US20050267702A1 - Method for developing a unified quality assessment and providing an automated fault diagnostic tool for turbine machine systems and the like - Google Patents
Method for developing a unified quality assessment and providing an automated fault diagnostic tool for turbine machine systems and the like Download PDFInfo
- Publication number
- US20050267702A1 US20050267702A1 US10/855,985 US85598504A US2005267702A1 US 20050267702 A1 US20050267702 A1 US 20050267702A1 US 85598504 A US85598504 A US 85598504A US 2005267702 A1 US2005267702 A1 US 2005267702A1
- Authority
- US
- United States
- Prior art keywords
- sensor data
- quality assessment
- data
- turbine
- event
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B23/00—Testing or monitoring of control systems or parts thereof
- G05B23/02—Electric testing or monitoring
- G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
- G05B23/0218—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults
- G05B23/0224—Process history based detection method, e.g. whereby history implies the availability of large amounts of data
- G05B23/0227—Qualitative history assessment, whereby the type of data acted upon, e.g. waveforms, images or patterns, is not relevant, e.g. rule based assessment; if-then decisions
- G05B23/0229—Qualitative history assessment, whereby the type of data acted upon, e.g. waveforms, images or patterns, is not relevant, e.g. rule based assessment; if-then decisions knowledge based, e.g. expert systems; genetic algorithms
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01D—MEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
- G01D3/00—Indicating or recording apparatus with provision for the special purposes referred to in the subgroups
- G01D3/08—Indicating or recording apparatus with provision for the special purposes referred to in the subgroups with provision for safeguarding the apparatus, e.g. against abnormal operation, against breakdown
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M15/00—Testing of engines
- G01M15/02—Details or accessories of testing apparatus
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B23/00—Testing or monitoring of control systems or parts thereof
- G05B23/02—Electric testing or monitoring
Definitions
- a new and improved approach toward developing a quality assessment for complex wind/steam/gas turbine systems, fluid compressor/pumping systems, generators, and the like is described.
- This approach combines the benefits of disparate statistical methods (such as, for example, the “matched filter” and the “multiple model hypothesis test”) to result in more accurate analysis and assessment of a particular machine/system operational event.
- the overall system quality as well as individual component quality is examined for deviations, which may correspond to or at least be indicative of specific faults. By comparing recent event signatures to selected archived signatures, system and component faults can be readily detected, identified and diagnosed.
- a computer implemented method for characterizing the relative degree of success or failure (i.e., providing a quality assessment) of a particular machine/system operational event by rating it over a continuous (contiguous) type assessment scale—as opposed to the more conventional “pass/fail” or “trip/no-trip” binary type assessment. It is contemplated that using a continuous type scale for characterizing a relative degree of “success” or “failure” of an operational event will better assist field technicians and operations personnel in assessing and communicating the quality of a particular operational event.
- Another aspect of this computer implemented assessment method is that it assesses and characterizes not only the quality of the system response to an operational event, but also the quality of individual component response to the event—thus enabling field engineers to identify and localize potential faults or failures within the machine system.
- the exemplary computer implemented quality assessment method described herein realizes the above improvements and benefits through a process of analyzing acquired system sensor and/or operational parameters data in conjunction with information concerning the existing ambient conditions and the fuel type/quality in a manner that eliminates or at least significantly reduces variability in the acquired data that is introduced by such known factors.
- a set of “corrected” parameters may be used to compensate for a known variability in operating conditions
- one aspect of the disclosed assessment method is to use such a set of corrected parameters to transform sensor and/or system operational parameter data collected during the operation of a particular machine/system into a “corrected parameter space” that effectively eliminates, or at least reduces, variability in the acquired data that is caused by known variations in ambient conditions and fuel type/quantity.
- Such transformed/corrected data corresponding to one or more operational variables of the system is then statistically analyzed and compared with a set of expected (“normal”) operational values and the results are used to diagnose and predict faults.
- available non-transformed (e.g., uncorrectable) operational event data may also evaluated in a manner which lessens the degree of confounding which may occur with the transformed/corrected data.
- separate quality assessments of the turbine operational event are developed (i.e., an assessment of the transformed data and an assessment of the non-transformed data). These assessments are then combined to provide a single overall “unified” comprehensive operational event assessment. This unified comprehensive operational event assessment is then tracked and updated over time and may be used to provide an early warning of machine/component degradation for a particular turbine system.
- event signatures corresponding to different anomalies produced by known faults may be saved or archived so that subsequent outlier event signatures can be diagnosed by being matched to an archived anomaly signature to identify a particular problem or component failure.
- quality assessments of operational events and/or particular system operational variables may be performed either in real-time while the monitored system is operational or implemented by recording system sensor data at predetermined times followed by a post-processing of the acquired data at a remote facility.
- a numerical quality assessment value for a particular operational event and/or a particular operational variable is computed and the event may be deemed as a “success” or “failure” based upon the degree to which the acquired transformed/corrected sensor data falls within certain predetermined numerical limits or “bounds” defining different quality categories.
- Numerical quality assessment values that are computed for different operational variables and/or events are saved and also used in developing an overall quality assessment for a particular gas turbine system.
- the quality assessment method disclosed and described herein may be used to provide a unified quality assessment of operational events, as well as provide component fault detection/identification, for a variety of different types of complex machine and machine systems such as power generator systems and turbine systems including wind/steam/gas turbines and/or fluid compressor/pump systems such as oil/gas pumping systems.
- complex machine and machine systems such as power generator systems and turbine systems including wind/steam/gas turbines and/or fluid compressor/pump systems such as oil/gas pumping systems.
- a gas turbine system is referenced and illustrated throughout the discussion of the invention herein, that particular example serves solely as one non-limiting example application.
- the computer implemented quality assessment and fault diagnostic method disclosed herein is not intended to be limited solely for use with gas turbine systems but is also intended as applicable for use in assessing and diagnosing most types of turbine machines/fleets/systems, compressors, pumps and other complex machine systems.
- FIG. 1 is a procedural diagram providing a basic overview of the operational event quality assessment/diagnostic process
- FIG. 2 is a process flow diagram illustrating example procedural blocks implemented on a computer/controller for developing and updating a turbine unit-specific signature of a turbine operational event;
- FIG. 3 is a process flow diagram illustrating example procedural blocks implemented on a computer/controller for developing and updating a turbine site-specific signature of a turbine operational event;
- FIG. 4 is a process flow diagram illustrating example procedural blocks implemented on a computer/controller for developing and updating a turbine machine fleet-specific signature of a turbine operational event;
- FIG. 5 is a process flow diagram illustrating example quality assessment procedural blocks implemented on a computer/controller for developing anomaly fault signatures associated with site and/or fleet-wide operational events;
- FIG. 6 is a process overview flow diagram illustrating example quality assessment procedural blocks implemented on a computer/controller for the evaluation of unified quality assessments of an operational event with respect to unit, site and fleet signatures;
- FIG. 7 is a flow diagram illustrating an example computer implemented process for providing automated fault detection/identification based on the operational event quality assessment process
- FIG. 8 is a flow diagram illustrating an example real-time local computer processing implementation of the operational event quality assessment process for a gas turbine system
- FIG. 9 is a flow diagram illustrating an example non-real-time computer processing implementation of the operational event quality assessment process for a gas turbine system that may be performed either locally or centrally;
- FIG. 10 is a pair of graphs illustrating examples of sub-signature signal data plots used in forming event signatures for a particular turbine machine
- FIG. 11A is diagram illustrating the computer/controller implemented processes of collection, transformation and fusion of signal data information to provide a single unified quality assessment
- FIG. 11B is a series of graphs illustrating transformation of an example collected data set via the computer/controller implemented quality assessment processes.
- FIG. 12 is an example computer output screen display for the computer implemented operational event quality assessment/diagnostic process for evaluating a turbine system operational event.
- Operational events which take place in large/complex turbine systems, fluid compressor/pumping systems and the like are often characterized by one or more operational variables that may be influenced by uncontrollable commonplace variations in ambient conditions and fuel type/quality.
- a computer implemented process is provided for developing a unified quality assessment of one or more of such turbine operational events despite such uncontrollable variations.
- a unique approach involves removing, or at least reducing, the effects of variations in ambient operating conditions and variations in fuel quality by initially performing a mathematical transform upon at least some of the acquired system/sensor data to effectively transform the data into a “corrected” parameter space, after which both transformed and non-transformed data corresponding to certain predetermined operational variables are statistically analyzed using, for example, matched filter, multiple model hypothesis tests, fault detection, etc. From that statistical analysis, a numerical quality assessment value for the event is developed and then compared with a pre-determined “expected” operational value or range. This expected value may be initially selected, for example, based on known variations particular to a specific commercial line, fleet type/model of machine or system (e.g., turbine fleet).
- this expected value is modified and continually updated so as to become more characteristic of that particular turbine system or a specific turbine unit at that site or a particular component of that turbine machine unit.
- the specific parameter corrections applied in any one incidence are not necessarily limited to the conventional corrections known in the industry, but may also include parameter corrections developed specifically for particular sensors such that the known/observed variation(s) due to ambient conditions/fuel quality is reduced in that particular sensor reading.
- Individual assessments of the same operational events made at different times and/or from different gas turbines of the same mechanical fleet may also be combined to develop a more complete and comprehensive assessment that effectively covers the entire duration of a particular turbine operational event in a contiguous manner.
- a computer processor or machine controller is programmed to perform one or more of the following operations:
- the method for developing a unified quality assessment described herein may be implemented via computer either at the turbine system site in real-time during the occurrence of a particular operational event or as part of a subsequent diagnostic process conducted remotely after storing and forwarding the acquired event data over a digital communications network to a central diagnostic site.
- FIG. 1 a procedural diagram is illustrated which provides an overview of the operational event quality assessment process.
- the process begins with the selection of a particular operational event of interest.
- a technical systems expert/analyst familiar with the particular system being evaluated, e.g., one who has an through understanding of the physics of the system and its various processes, will be instrumental in selecting or predefining at least the following initial parameters:
- Data acquisition may be performed manually, at least initially, but preferably would be automated through the implementation of an appropriate automatic data capturing process.
- Data acquisition is initiated and a data file created whenever a predetermined “trigger” condition is met.
- the “trigger” is used as an alignment point for analysis of the acquired signal/sensor data.
- An exemplary data acquisition process may also include:
- site and unit-specific signatures and corresponding thresholds are created for a given event of a particular configuration type.
- the corresponding corrected parameters are computed (e.g., the data is transformed to a corrected parameter space), as indicated in block 120 .
- the corresponding signal signatures may at least be checked to insure that the acquired data falls within pre-specified range or bounds (block 120 ).
- Developing corrections is done by using domain knowledge of the process, identifying the variables or signals that characterize the process, using the physics of the process to determine either dimensionless quantities that characterize the process or those that have minimal effect of the ambient conditions like temperature, pressure, humidity etc. when plotted against another virtual variable.
- a signature associated with an operational event for a particular equipment type/configuration is formed from a set of sub-signature plots of predetermined sensor signals or parameters (e.g., Fuel or turbine acceleration), plotted in an appropriately corrected domain for that signal.
- Each sub-signature plot is characterized by a nominal value having associated quality thresholds boundaries/ranges (e.g., red, yellow and green) defined about the nominal value.
- the event signature is initially created by overlaying the sub-signature plots created from multiple data sets of acquired sensor or monitored parameter data and determining the nominal or representative plot for each sub-signature. (See examples shown in FIG. 10 .)
- the validity of the corrections and corrected space applied to data for a given event is determined by utilizing data acquired from several machine units and developing probability distributions indicative of the corresponding range of variations across the fleet.
- event signatures are developed from parameter plots that correspond to the acquired sensor data, at least some of which is corrected by utilizing one or more corrected parameter coefficients which reduces or eliminates variabilities in the sensor data caused by ambient conditions and/or fuel type/quality.
- a numerical quality assessment value is then determined for the event based upon comparison with a pre-determined expected value/range and the event is then accordingly classified into an appropriate quality assessment category. All corrected (transformed) parameter data and non-corrected (non-transformed) data are combined using, for example, a weighted average or rule-based averaging. This combined overall assessment is then classified into a “red,” “yellow” or “green” quality category. As indicated at block 160 , signals that fall into “suspect” categories (e.g., “red” or “yellow”) are tagged for further analysis to identify potential operational problems. Ultimately, the combined overall numerical quality assessment, as well as the quality assessment of signals/parameters that fall into suspect categories, are tracked over time to provide an early warning and identification of component or system degradation, component and system modifications and potential failures, as indicated in block 160 .
- a process flow diagram illustrates example procedural blocks implemented by a computer or system controller for developing a “unit-specific” signature for an operational event occurring on a single turbine machine at a turbine system site (at any one particular turbine system site the turbine system may include more than one turbine machine unit).
- This process may be implemented by a computer located at the turbine system site, such as the local turbine controller, or it may be implemented by a remotely located processing system which receives data from the turbine controller.
- a fleet signature or other generic signature may initially be used for performing the quality assessment analysis until sufficient data is collected over time to develop an adequate site-specific event signature.
- a database containing historical operational event data for one or more turbine systems is maintained and updated with new data at regular intervals (e.g., block 221 ).
- a database may also contain corrected parameters that are predefined for various operational events corresponding to specific turbine units located at different sites. Assuming that sufficient previously recorded historical event data exists for a particular turbine unit of interest (block 210 ), database files containing historical operational event data for the unit and/or for the particular site where the unit is situated are accessed (block 220 ) to identify corrected parameters that are predefined for the particular operational event (block 230 ). A conventional data set optimization is then performed to determine corrected parameter coefficients that will minimize the variance in the observed data set (block 240 ).
- corrected parameter plots are developed and the mean signature and variation are determined using, one or more conventional statistical methods (e.g., matched filter and multiple model hypothesis test).
- quality assessment category “thresholds” or “boundaries” are computed for use in classifying signal data into one of a plurality of quality categories (e.g., red, yellow and green).
- these quality range boundaries/thresholds may be initially set or verified by a system operator or user, as indicated at block 260 .
- the signature and the threshold may be validated by comparison with archived data stored in a validation database (block 270 ).
- the resulting signature associated with that turbine unit and saved (presumably in a historical operational event database) so that it may also be accessed and used by field personnel (block 290 ). Otherwise, as indicated at block 281 , the thresholds and/or corrected parameter coefficients (and/or the detection algorithm) is adjusted and blocks 250 through 280 are repeated until the desired performance is obtained. This entire process may be repeated as additional or new data from the turbine unit is collected, as indicated in block 291 .
- FIG. 3 shows a process flow diagram illustrating example quality assessment procedural blocks implemented on a computer/controller for developing a turbine system site-specific signature of a turbine operational event.
- a historical database consisting of operational event files corresponding to turbine systems located at one or more sites exists and is maintained to serve as a source of information concerning the turbine units at a particular site.
- this database is accessed to obtain generalized operationally corrected parameters and information corresponding to the turbine units being analyzed at a particular site. Corrected parameters that are predefined for the particular operational event at that site being assessed are identified, as indicated in block 310 , and corrected parameter coefficients are computed based on this information, as indicated in block 320 .
- a mean signature and variation is determined from the corrected parameter plots and used to compute the three red, yellow and green category quality thresholds (bounds) for defining a quality assessment.
- the computed thresholds are verified with a system operator/user, as indicated at block 340 .
- the signatures and the thresholds are then validated using a validation database that contains a record of successful and unsuccessful operational events, as indicated in block 350 .
- the results are saved as an updated site-specific operational event signature (preferably in a historical operational event database) and made available for future use and/or access by field personnel, as indicated in block 370 .
- site-specific operational event signature preferably in a historical operational event database
- the monitored events may be partitioned into different sets, and a signature and the corresponding thresholds and boundaries may be determined individually for each set. In practice, this may correspond to a change over time in the signature for a specific unit or of different configurations across multiple units. Accordingly, the thresholds and/or the corrected parameter coefficients (and/or the particular detection process used) are adjusted and recomputed as indicated at blocks 361 and 330 .
- FIG. 4 shows a process flow diagram illustrating example procedural blocks implemented on a computer/controller to develop a turbine fleet-specific signature of a turbine operational event.
- creation of a fleet-specific signature may be performed before the creation of a unit-specific signature after collecting fleet-wide data sets.
- a historical database of operational event files associated with turbine systems and various sites exists and is maintained to provide turbine fleet information.
- this database is accessed to obtain generalized operationally corrected parameters and information corresponding to the particular type of gas turbine machine being analyzed. Corrected parameters that are predefined for the particular operational event being assessed are identified, as indicated in block 410 , and corrected parameter coefficients are computed based on this information, as indicated in block 420 .
- a mean signature and variation is determined from the corrected parameter plots and used to compute the three red, yellow and green category quality thresholds (bounds) for defining a quality assessment.
- the computed thresholds are verified with the user, as indicated at block 440 .
- the signatures in the thresholds are then validated using a validation database that contains a record of successful and unsuccessful operational events, as indicated in block 450 .
- the results are saved as an updated fleet operational event signature (preferably in a historical operational event database and made available for future use and/or access by field personnel, as indicated in block 470 .
- the monitored events may be partitioned into different sets, and a signature and the corresponding fleet thresholds and boundaries may be determined individually for each set. In practice, this may correspond to a change over time in the signature for a specific unit or of different configurations across multiple units. Accordingly, the thresholds and/or parameter coefficients are adjusted and recomputed as indicated at blocks 461 and 430 .
- the sensor data and/or parameter data corresponding to various “anomalous” or outlier operational events is also saved in the historical database (e.g., at procedural blocks 290 , 370 , 70 ) and flagged as data which comprise an anomaly event database.
- This anomaly event database may then be used for troubleshooting purposes by providing a means for identifying those operational events that evidence a “best fit” relationship with an anomalous event signature/data previously saved in the anomaly event database.
- An example process for creating an anomaly signature and/or an individual component fault signature is illustrated by the procedural flow diagram of FIG. 5 .
- the diagram shows example procedural blocks which may be implemented on a computer processor/controller as part of the operational event quality assessment process to develop both system and individual component anomaly fault signatures associated with a particular unit, site or fleet-wide operational events.
- specific storage space or files in a historical database in a computer memory are set up or allocated for use as an anomaly event database.
- the corrected parameter specific to the event and anomaly is identified and CPC corrections to variations associated with ambient conditions are applied, as indicated in procedural blocks 510 and 520 .
- the mean signature and the anomaly threshold/boundaries are determined, as indicated in block 530 , and the signature is validated then compared against existing anomaly signatures stored in the anomaly database, as indicated in block 540 .
- the validated anomaly signature is not currently in the anomaly database, it is saved along with appropriate identifying information or comments, as indicated in block 550 .
- a procedural flow diagram is shown which may serve as a general overview of an exemplary computer/controller implementation of the operational event quality assessment process.
- a particular operational event is initiated on the turbine and operational parameter data from various sensors is monitored during and throughout the operational event.
- an on-site real-time turbine unit controller (or a comparable remote monitoring system) is configured to recognize the particular type of operational event taking place (e.g., start-up, mode transfer, etc.) and verify that valid signals are being acquired from the various turbine sensors (block 611 ).
- Sets of appropriately verified sensor signals i.e., verified as appropriate for the particular operational event
- This real-time quality assessment analysis may either be implemented locally by the turbine system site controller itself or the acquired sensor data may be transmitted via an appropriate digital communications network to a remote real-time processing facility.
- a “post-processing” arrangement may also be implemented wherein one or more system events are monitored and all of the appropriate sensor data during each event is collected and saved in a historical event/trend file which may be stored on site or at a remote facility. Subsequently, as indicated at block 612 , a particular operational event may be selected for analysis. If it exists, all prerecorded pertinent data corresponding to that particular event is retrieved from a historical event/trend file (block 614 ) and the sensor signal data may then be examined and verified as valid (block 616 ) before being passed on for further processing.
- the corresponding corrected parameters are computed (e.g., the data is transformed to a corrected parameter space), as indicated in block 620 .
- the corresponding signal signatures may at least be checked to insure that the acquired data falls within pre-specified range or bounds (block 620 ). Accordingly, event signatures are developed from parameter plots that correspond to the acquired sensor data, at least some of which is corrected by utilizing one or more corrected parameter coefficients which reduces or eliminates variabilities in the sensor data caused by ambient conditions and/or fuel type/quality.
- a numerical quality assessment value is then determined for the event based upon comparison with a pre-determined expected value/range and the event is then accordingly classified into an appropriate quality assessment category.
- all corrected (transformed) parameter data and non-corrected (non-transformed) data are combined using, for example, a weighted average or rule-based averaging.
- This combined overall assessment is then classified into a “red,” “yellow” or “green” quality category according to.
- signals that fall into “suspect” categories e.g., “red” or “yellow” are tagged for further analysis to identify potential operational problems.
- the combined overall numerical quality assessment, as well as the quality assessment of signals/parameters that fall into suspect categories are tracked over time to provide an early warning and identification of component or system degradation, component and system modifications and potential failures, as indicated in block 660 .
- FIG. 7 illustrates an example computer implemented process blocks for providing automated fault detection and diagnosis/identification that operates as part of the basic quality assessment processing described above.
- a database of outlier anomaly event signatures is developed during the quality assessment processing.
- This database may also include historical/archival operational event data indicative of component degradation and fault signatures corresponding to both individual machine units and/or specific machine fleets.
- operational events that fall into categories indicative of poor or bad quality such as the “red” and “yellow” quality classifications, are treated as suspect events and are used as candidates for detecting and identifying system and component faults.
- a first candidate operational event is selected for examination and, if not already done, corrections for ambient conditions are applied to the event signals and a corrected parameter specific to that operational event is identified.
- a list of anomaly signatures associated with the selected event is formed (block 730 ) and then the identified event parameter is compared with each of the anomaly signatures in the list to determine if a close match exists (blocks 740 and 750 ). Event parameters and anomaly signatures exhibiting a high degree to matching are then used to identify the particular fault type and component or system malfunction.
- FIG. 8 a flow diagram is shown which illustrates an example real-time implementation of an operational event quality assessment process for a gas turbine.
- turbine sensor data from an operational event is provided to the local or a remote computer processor/turbine-controller 801 .
- processor/controller 801 uses site and fleet signal threshold information obtained from a historical event database to compute a combined overall numerical assessment quality value in real-time and determines which quality category (red, yellow, green) the operational event is classified (block 810 ). This information is then saved in a historical tracking database as indicated at block 820 .
- FIG. 9 shows a flow diagram illustrating an example of a non-real-time computer processing implementation of the operational event quality assessment process for a gas turbine system.
- gas turbine sensor data during an operational event is provided to processor/controller 901 which stores the acquired information locally or sends it to a central server for performing further analysis at a later time.
- An on-site monitor/user interface 902 may be provided to provide a means for an operator to locally access, control and display the acquired data and results from any quality assessment and fault diagnostic analysis that is performed. 1002 also suggests that this function could be performed a remote central site.
- Unit, site and fleet signatures and threshold information are obtained from a historical event database and used in computing the combined overall quality assessment value and determining the quality category of the event, as indicated at block 910 . This information may then be saved in a historical tracking database as indicated at block 920 .
- FIG. 10 shows two example sets of parameter/sensor time domain sub-signature signal data plots obtained during a turbine “startup” event that are used in forming event signatures for a particular turbine machine.
- the left example illustrates plots of acceleration vs. time and the right example illustrates plots of percent fuel vs. time.
- the first action is to time align the data.
- a signature is formed from a set of sub-signature data plots.
- Each sub-signature plot is characterized by a nominal value having associated quality thresholds boundaries/ranges (e.g., red, yellow and green) defined about the nominal value.
- An event signature is produced by obtaining multiple data sets and overlaying the corresponding sub-signature plots to determine the “nominal” or representative plot for each sub-signature.
- FIG. 11A shows an example diagram illustrating the computer/controller implemented processes of collection, transformation and fusion of signal data information to provide a single unified quality assessment.
- multiple time domain plots of, for example, pressure, temperature, speed, etc. are developed from the data collected, and aligned in the time domain.
- An arrow from block 1101 to block 1102 illustrates the transformation of the data of block 1101 into a corrected parameter space to remove the effects of ambient conditions, fuel quality and/or other known causes of variability in the data. Transformed data from this corrected parameter space is used to generate X-Y virtual parameter plots, as shown in block 1102 , that are effectively corrected for ambient conditions and systemic variations and will provide a statistically better indication of the underlying process.
- a single unified assessment of success is produced by performing a probabilistic averaging of the sub-signature assessments, as described above with respect to FIG. 6 .
- FIG. 11B provides a series of graphs illustrating an example transformation of an example collected data set using the above described computer/controller implemented quality assessment processes.
- data from a time domain plot of compressor discharge pressure (CPD), 1103 is transformed to the corrected parameters of compressor pressure ratio (CPR) vs. corrected speed, shown in graph 1104 .
- CPR compressor pressure ratio
- This sub-signature information is then used for producing a unified quality assessment as discussed above and illustrated in FIG. 11A .
- the appropriate correction (corrected parameter space) for the above example and for each data set for other operational variables, such as temperatures, fuel, etc. is developed using known conventional techniques familiar to those skilled in the art and typically involves using domain knowledge of the operational event, identifying the variables or signals that characterize the operational event and applying knowledge of the underlying physics of the operational event to determine either dimensionless quantities that characterize the event and minimize the effect of ambient conditions like temperature, pressure, humidity, etc. when plotted against another virtual variable.
- the quality assessment information developed by the present computer implemented operational event quality assessment/diagnostic process may be output to a display device, a laptop or a printer.
- FIG. 12 shows an example of an output screen display that may be produced by the computer implemented operational event quality assessment/diagnostic process upon evaluating a turbine or other machine system operational event.
- the machine site, ID, equipment configuration, operational event and date information are displayed in separate columns ( 1201 ) for each event assessed.
- an associated “Status” column ( 1202 ) the unified quality assessment value developed for each machine fleet, machine site and machine unit for each operational event evaluated is displayed along with a color indicator showing the corresponding quality range (e.g., red, yellow or green).
Abstract
Description
- For turbine electric power generation systems, large fluid compressor/pump systems and the like, a great number of sensor signals and operational data often needs to be acquired and analyzed to properly evaluate a particular significant operational “event” (e.g., operational “events” such as start-up operations, mode transfer events, FSNL-FSFL, etc.). Consequently, it is usually not possible to quickly perform and obtain an accurate assessment of such events. In addition, variations in ambient operating conditions and/or fuel quality result in inconsistent and inaccurate sensor readings. This makes comparisons of operational events from one operational “run” of a particular turbine/compressor system to the next (as well as comparisons between operational runs of different turbine/compressor systems) impracticable. Moreover, for the same reasons, these problems make it impracticable to attempt to compare an operational run of a particular turbine machine with any sort of standardized data indicative of a normal operation for that particular system. Consequently, operations personnel (e.g., field engineers, technicians, remote tuning and systems operations center personnel) often resort to relying solely on signals from an individual sensor or an individual system parameter to determine whether a particular turbine or compressor machine unit or component is operating below an appropriate safety limit or within a proper tolerance range—such limit/range often being based generally upon some known variability inherent to a particular machine's design fleet or upon some known consistent variability in the particular ambient operating conditions or fuel type/quality.
- Although fault detection mechanisms and statistical tests useful for analyzing and evaluating operational events of complex machine systems and equipment have been developed, the known conventional procedures for such have significant efficiency limitations and often produce inaccurate or erratic results. A much more accurate and efficient approach for developing quality assessments and providing fault diagnosis of operational events occurring in complex compressor/pump and turbine machine systems is needed and is highly desirable.
- A new and improved approach toward developing a quality assessment for complex wind/steam/gas turbine systems, fluid compressor/pumping systems, generators, and the like is described. This approach combines the benefits of disparate statistical methods (such as, for example, the “matched filter” and the “multiple model hypothesis test”) to result in more accurate analysis and assessment of a particular machine/system operational event. In addition to providing a unified quality assessment, the overall system quality as well as individual component quality is examined for deviations, which may correspond to or at least be indicative of specific faults. By comparing recent event signatures to selected archived signatures, system and component faults can be readily detected, identified and diagnosed.
- In one aspect, a computer implemented method is described herein for characterizing the relative degree of success or failure (i.e., providing a quality assessment) of a particular machine/system operational event by rating it over a continuous (contiguous) type assessment scale—as opposed to the more conventional “pass/fail” or “trip/no-trip” binary type assessment. It is contemplated that using a continuous type scale for characterizing a relative degree of “success” or “failure” of an operational event will better assist field technicians and operations personnel in assessing and communicating the quality of a particular operational event. Another aspect of this computer implemented assessment method is that it assesses and characterizes not only the quality of the system response to an operational event, but also the quality of individual component response to the event—thus enabling field engineers to identify and localize potential faults or failures within the machine system.
- Basically, the exemplary computer implemented quality assessment method described herein realizes the above improvements and benefits through a process of analyzing acquired system sensor and/or operational parameters data in conjunction with information concerning the existing ambient conditions and the fuel type/quality in a manner that eliminates or at least significantly reduces variability in the acquired data that is introduced by such known factors. Based on the premise that a set of “corrected” parameters may be used to compensate for a known variability in operating conditions, one aspect of the disclosed assessment method is to use such a set of corrected parameters to transform sensor and/or system operational parameter data collected during the operation of a particular machine/system into a “corrected parameter space” that effectively eliminates, or at least reduces, variability in the acquired data that is caused by known variations in ambient conditions and fuel type/quantity. Such transformed/corrected data corresponding to one or more operational variables of the system is then statistically analyzed and compared with a set of expected (“normal”) operational values and the results are used to diagnose and predict faults.
- In a further aspect of the exemplary computer implemented quality assessment method disclosed herein, available non-transformed (e.g., uncorrectable) operational event data may also evaluated in a manner which lessens the degree of confounding which may occur with the transformed/corrected data. In the example implementation, separate quality assessments of the turbine operational event are developed (i.e., an assessment of the transformed data and an assessment of the non-transformed data). These assessments are then combined to provide a single overall “unified” comprehensive operational event assessment. This unified comprehensive operational event assessment is then tracked and updated over time and may be used to provide an early warning of machine/component degradation for a particular turbine system. In yet a further aspect of the disclosed method, event signatures corresponding to different anomalies produced by known faults may be saved or archived so that subsequent outlier event signatures can be diagnosed by being matched to an archived anomaly signature to identify a particular problem or component failure. In still yet a further aspect of the disclosed method, quality assessments of operational events and/or particular system operational variables may be performed either in real-time while the monitored system is operational or implemented by recording system sensor data at predetermined times followed by a post-processing of the acquired data at a remote facility.
- In at least one non-limiting example implementation discussed and illustrated herein, a numerical quality assessment value for a particular operational event and/or a particular operational variable is computed and the event may be deemed as a “success” or “failure” based upon the degree to which the acquired transformed/corrected sensor data falls within certain predetermined numerical limits or “bounds” defining different quality categories. The operational event is then classified accordingly into one of three different categories (e.g., red, yellow or green) that are intended as being generally indicative of its relative operational “success” or “failure” (e.g., “red”=failure; “green”=success). Numerical quality assessment values that are computed for different operational variables and/or events are saved and also used in developing an overall quality assessment for a particular gas turbine system.
- The quality assessment method disclosed and described herein may be used to provide a unified quality assessment of operational events, as well as provide component fault detection/identification, for a variety of different types of complex machine and machine systems such as power generator systems and turbine systems including wind/steam/gas turbines and/or fluid compressor/pump systems such as oil/gas pumping systems. Although a gas turbine system is referenced and illustrated throughout the discussion of the invention herein, that particular example serves solely as one non-limiting example application. The computer implemented quality assessment and fault diagnostic method disclosed herein is not intended to be limited solely for use with gas turbine systems but is also intended as applicable for use in assessing and diagnosing most types of turbine machines/fleets/systems, compressors, pumps and other complex machine systems.
- Other advantages and objects of the present invention will be described in detail with reference to the accompanying drawings, in which:
-
FIG. 1 is a procedural diagram providing a basic overview of the operational event quality assessment/diagnostic process; -
FIG. 2 is a process flow diagram illustrating example procedural blocks implemented on a computer/controller for developing and updating a turbine unit-specific signature of a turbine operational event; -
FIG. 3 is a process flow diagram illustrating example procedural blocks implemented on a computer/controller for developing and updating a turbine site-specific signature of a turbine operational event; -
FIG. 4 is a process flow diagram illustrating example procedural blocks implemented on a computer/controller for developing and updating a turbine machine fleet-specific signature of a turbine operational event; -
FIG. 5 is a process flow diagram illustrating example quality assessment procedural blocks implemented on a computer/controller for developing anomaly fault signatures associated with site and/or fleet-wide operational events; -
FIG. 6 is a process overview flow diagram illustrating example quality assessment procedural blocks implemented on a computer/controller for the evaluation of unified quality assessments of an operational event with respect to unit, site and fleet signatures; -
FIG. 7 is a flow diagram illustrating an example computer implemented process for providing automated fault detection/identification based on the operational event quality assessment process; -
FIG. 8 is a flow diagram illustrating an example real-time local computer processing implementation of the operational event quality assessment process for a gas turbine system; -
FIG. 9 is a flow diagram illustrating an example non-real-time computer processing implementation of the operational event quality assessment process for a gas turbine system that may be performed either locally or centrally; -
FIG. 10 is a pair of graphs illustrating examples of sub-signature signal data plots used in forming event signatures for a particular turbine machine; -
FIG. 11A is diagram illustrating the computer/controller implemented processes of collection, transformation and fusion of signal data information to provide a single unified quality assessment; -
FIG. 11B is a series of graphs illustrating transformation of an example collected data set via the computer/controller implemented quality assessment processes; and -
FIG. 12 is an example computer output screen display for the computer implemented operational event quality assessment/diagnostic process for evaluating a turbine system operational event. - Operational events which take place in large/complex turbine systems, fluid compressor/pumping systems and the like are often characterized by one or more operational variables that may be influenced by uncontrollable commonplace variations in ambient conditions and fuel type/quality. A computer implemented process is provided for developing a unified quality assessment of one or more of such turbine operational events despite such uncontrollable variations. As briefly outlined above, a unique approach is described that involves removing, or at least reducing, the effects of variations in ambient operating conditions and variations in fuel quality by initially performing a mathematical transform upon at least some of the acquired system/sensor data to effectively transform the data into a “corrected” parameter space, after which both transformed and non-transformed data corresponding to certain predetermined operational variables are statistically analyzed using, for example, matched filter, multiple model hypothesis tests, fault detection, etc. From that statistical analysis, a numerical quality assessment value for the event is developed and then compared with a pre-determined “expected” operational value or range. This expected value may be initially selected, for example, based on known variations particular to a specific commercial line, fleet type/model of machine or system (e.g., turbine fleet). For example, in a turbine power generating system, as additional operational data from a particular turbine site is acquired over time, this expected value is modified and continually updated so as to become more characteristic of that particular turbine system or a specific turbine unit at that site or a particular component of that turbine machine unit. The specific parameter corrections applied in any one incidence are not necessarily limited to the conventional corrections known in the industry, but may also include parameter corrections developed specifically for particular sensors such that the known/observed variation(s) due to ambient conditions/fuel quality is reduced in that particular sensor reading. Individual assessments of the same operational events made at different times and/or from different gas turbines of the same mechanical fleet may also be combined to develop a more complete and comprehensive assessment that effectively covers the entire duration of a particular turbine operational event in a contiguous manner.
- In one non-limiting example implementation of the method for developing a unified quality assessment, as illustrated herein for a gas turbine system, a computer processor or machine controller is programmed to perform one or more of the following operations:
-
- acquiring and/or recording pertinent sensor data which characterizes the operation of the turbine during occurrence of an operational event, wherein the acquired sensor data includes information concerning the ambient operating conditions of the turbine and/or the fuel quality/type;
- using a predetermined mathematical transform or a set of correction parameters to transform/correct acquired sensor data (e.g., by transforming or converting the data into a corrected parameter space) to effectively remove or correct for variability in the data that results from variations in ambient operating conditions at the turbine and/or fuel type/quality;
- comparing both the transformed data and other non-transformed sensor data (e.g., unaffected acquired sensor data) relevant to the operational event with a predetermined expected or “normal” data value or range and determining a statistical degree to which that data matches the expected value/range—i.e., comparisons are made against an expected “normal” value using both the non-transformed data and the transformed data as acquired from operational events occurring on the same turbine machine and/or from operational events occurring across different turbine machines to determine the degree to which the data falls within or outside of certain predetermined bounds—the comparing process being basically statistical in nature in that it utilizes variability information in the transformed or non-transformed parameter space (depending on the operational variable(s) being considered); in this example, the expected “normal” value/range may be initially based upon historical (e.g., archived information for a particular turbine fleet (a “fleet” being a group of turbine machine production models having the same or similar configuration, size, etc.) and as additional data is subsequently collected for a particular operational event occurring on a particular turbine machine or component, the corresponding expected “normal” value is updated/adjusted to more accurately reflect the turbine's actual “normal” operation during that particular type of operational event;
- classifying the data into a plurality of quality categories (e.g., “red”, “yellow” and “green”) according to the statistical degree to which the data matches the expected value/range;
- combining the statistical evaluations of both transformed data and non-transformed data into a single comprehensive quality assessment value that is indicative of the quality of a particular turbine operational event—i.e., an information “fusion” process is implemented which combines all individual comparison statistics to produce an overall comprehensive quality metric of the operational event (e.g., a comprehensive numeric quality assessment value) that is contiguous and continuously evolving (as opposed to being a static, non-evolving, binary-type indication of event quality, e.g., “good/no-good” or “pass/fail”); and
- continuously tracking and updating the developed comprehensive event assessment value over time and identifying when a deviation in the assessment value violates a pre-determined threshold/range so that the cause of the deviation may be identified and appropriate corrective action initiated before a severe problem develops; in this regard, the assessment tracking procedure is preferably made somewhat tolerant of noise present in the event assessment data so as to reduce the occurrence of false positives.
- As will become evident from the non-limiting exemplary application discussed below, the method for developing a unified quality assessment described herein may be implemented via computer either at the turbine system site in real-time during the occurrence of a particular operational event or as part of a subsequent diagnostic process conducted remotely after storing and forwarding the acquired event data over a digital communications network to a central diagnostic site.
- Referring first to
FIG. 1 , a procedural diagram is illustrated which provides an overview of the operational event quality assessment process. As indicated atprocedural block 100, the process begins with the selection of a particular operational event of interest. Ideally, a technical systems expert/analyst familiar with the particular system being evaluated, e.g., one who has an through understanding of the physics of the system and its various processes, will be instrumental in selecting or predefining at least the following initial parameters: -
- a “trigger” to start data acquisition
- a list of signals/sensors to be sampled
- a rate at which to sample the signals/sensors
- Data acquisition may be performed manually, at least initially, but preferably would be automated through the implementation of an appropriate automatic data capturing process. Data acquisition is initiated and a data file created whenever a predetermined “trigger” condition is met. The “trigger” is used as an alignment point for analysis of the acquired signal/sensor data. There may be multiple points or signals in the data that may be used as a trigger from which re-alignment of the acquired data may also be beneficial. This approach to data acquisition ensures that any signature creation has a consistent beginning point from which all subsequent data can be analyzed. An exemplary data acquisition process may also include:
-
- Normalizing any sensor or signal as separate value while retaining the original information;
- Filtering any signal or signals as needed; and
- Zero shifting any sensor drift if possible.
The described procedures for data acquisition correspond to a single operational event “run” which may be iteratively performed for collecting data for a plurality of data runs from multiple sites as indicated inprocedural block 110.
- As indicated at
procedural block - Many signals from the system sensors will be appropriate candidates for transformation to a corrected parameter space to reduce/remove the effects of ambient, fuel and machine state variability. However, for certain signals, a correction/transformation to remedy such effects will not be available or appropriate. For the signal types which can be corrected, the corresponding corrected parameters are computed (e.g., the data is transformed to a corrected parameter space), as indicated in
block 120. For those signals that do not have appropriate correction parameters but which may be critical for event assessment, the corresponding signal signatures may at least be checked to insure that the acquired data falls within pre-specified range or bounds (block 120). - Developing corrections is done by using domain knowledge of the process, identifying the variables or signals that characterize the process, using the physics of the process to determine either dimensionless quantities that characterize the process or those that have minimal effect of the ambient conditions like temperature, pressure, humidity etc. when plotted against another virtual variable.
- A signature associated with an operational event for a particular equipment type/configuration is formed from a set of sub-signature plots of predetermined sensor signals or parameters (e.g., Fuel or turbine acceleration), plotted in an appropriately corrected domain for that signal. Each sub-signature plot is characterized by a nominal value having associated quality thresholds boundaries/ranges (e.g., red, yellow and green) defined about the nominal value. The event signature is initially created by overlaying the sub-signature plots created from multiple data sets of acquired sensor or monitored parameter data and determining the nominal or representative plot for each sub-signature. (See examples shown in
FIG. 10 .) The validity of the corrections and corrected space applied to data for a given event is determined by utilizing data acquired from several machine units and developing probability distributions indicative of the corresponding range of variations across the fleet. - Every time, a new event data set is obtained, it is transformed into the corrected space and then matched versus each of the sub-signature plots. A quantitative measure of the fit versus each of the sub-signature plots is obtained, and a single assessment of fit versus the signature is computed as a probabilistically weighted average. Accordingly, event signatures are developed from parameter plots that correspond to the acquired sensor data, at least some of which is corrected by utilizing one or more corrected parameter coefficients which reduces or eliminates variabilities in the sensor data caused by ambient conditions and/or fuel type/quality.
- As indicated in
block 150, a numerical quality assessment value is then determined for the event based upon comparison with a pre-determined expected value/range and the event is then accordingly classified into an appropriate quality assessment category. All corrected (transformed) parameter data and non-corrected (non-transformed) data are combined using, for example, a weighted average or rule-based averaging. This combined overall assessment is then classified into a “red,” “yellow” or “green” quality category. As indicated atblock 160, signals that fall into “suspect” categories (e.g., “red” or “yellow”) are tagged for further analysis to identify potential operational problems. Ultimately, the combined overall numerical quality assessment, as well as the quality assessment of signals/parameters that fall into suspect categories, are tracked over time to provide an early warning and identification of component or system degradation, component and system modifications and potential failures, as indicated inblock 160. - In
FIG. 2 , a process flow diagram illustrates example procedural blocks implemented by a computer or system controller for developing a “unit-specific” signature for an operational event occurring on a single turbine machine at a turbine system site (at any one particular turbine system site the turbine system may include more than one turbine machine unit). This process may be implemented by a computer located at the turbine system site, such as the local turbine controller, or it may be implemented by a remotely located processing system which receives data from the turbine controller. Initially, as indicated atprocedural block 200, it is determined whether any particular quality category range or “boundary” information exists for sensor data originating from the specific turbine system site being assessed. As indicated atblock 201, if no site-specific quality category ranges or boundary information exists, a fleet signature or other generic signature may initially be used for performing the quality assessment analysis until sufficient data is collected over time to develop an adequate site-specific event signature. - Preferably, a database containing historical operational event data for one or more turbine systems (or other fluid compressor systems or the like) is maintained and updated with new data at regular intervals (e.g., block 221). Such a database may also contain corrected parameters that are predefined for various operational events corresponding to specific turbine units located at different sites. Assuming that sufficient previously recorded historical event data exists for a particular turbine unit of interest (block 210), database files containing historical operational event data for the unit and/or for the particular site where the unit is situated are accessed (block 220) to identify corrected parameters that are predefined for the particular operational event (block 230). A conventional data set optimization is then performed to determine corrected parameter coefficients that will minimize the variance in the observed data set (block 240). Next, as indicated at
block 250, corrected parameter plots are developed and the mean signature and variation are determined using, one or more conventional statistical methods (e.g., matched filter and multiple model hypothesis test). Using this information, quality assessment category “thresholds” or “boundaries” are computed for use in classifying signal data into one of a plurality of quality categories (e.g., red, yellow and green). Before being applied, these quality range boundaries/thresholds may be initially set or verified by a system operator or user, as indicated atblock 260. Next, the signature and the threshold may be validated by comparison with archived data stored in a validation database (block 270). If a valid signature or the desired performance has been achieved, the resulting signature associated with that turbine unit and saved (presumably in a historical operational event database) so that it may also be accessed and used by field personnel (block 290). Otherwise, as indicated atblock 281, the thresholds and/or corrected parameter coefficients (and/or the detection algorithm) is adjusted and blocks 250 through 280 are repeated until the desired performance is obtained. This entire process may be repeated as additional or new data from the turbine unit is collected, as indicated inblock 291. -
FIG. 3 shows a process flow diagram illustrating example quality assessment procedural blocks implemented on a computer/controller for developing a turbine system site-specific signature of a turbine operational event. Preferably, a historical database consisting of operational event files corresponding to turbine systems located at one or more sites exists and is maintained to serve as a source of information concerning the turbine units at a particular site. Initially, as indicated inblock 300, this database is accessed to obtain generalized operationally corrected parameters and information corresponding to the turbine units being analyzed at a particular site. Corrected parameters that are predefined for the particular operational event at that site being assessed are identified, as indicated inblock 310, and corrected parameter coefficients are computed based on this information, as indicated inblock 320. Next, as indicated inblock 330, a mean signature and variation is determined from the corrected parameter plots and used to compute the three red, yellow and green category quality thresholds (bounds) for defining a quality assessment. Next, the computed thresholds are verified with a system operator/user, as indicated atblock 340. The signatures and the thresholds are then validated using a validation database that contains a record of successful and unsuccessful operational events, as indicated inblock 350. - As indicated in
block 360, if a valid signature or desired performance was achieved, then the results are saved as an updated site-specific operational event signature (preferably in a historical operational event database) and made available for future use and/or access by field personnel, as indicated inblock 370. If a valid signature is not obtained, the monitored events may be partitioned into different sets, and a signature and the corresponding thresholds and boundaries may be determined individually for each set. In practice, this may correspond to a change over time in the signature for a specific unit or of different configurations across multiple units. Accordingly, the thresholds and/or the corrected parameter coefficients (and/or the particular detection process used) are adjusted and recomputed as indicated atblocks -
FIG. 4 shows a process flow diagram illustrating example procedural blocks implemented on a computer/controller to develop a turbine fleet-specific signature of a turbine operational event. In practice, creation of a fleet-specific signature may be performed before the creation of a unit-specific signature after collecting fleet-wide data sets. - Preferably, a historical database of operational event files associated with turbine systems and various sites exists and is maintained to provide turbine fleet information. Initially, as indicated in
block 400, this database is accessed to obtain generalized operationally corrected parameters and information corresponding to the particular type of gas turbine machine being analyzed. Corrected parameters that are predefined for the particular operational event being assessed are identified, as indicated inblock 410, and corrected parameter coefficients are computed based on this information, as indicated inblock 420. Next, as indicated inblock 430, a mean signature and variation is determined from the corrected parameter plots and used to compute the three red, yellow and green category quality thresholds (bounds) for defining a quality assessment. Next, the computed thresholds are verified with the user, as indicated atblock 440. The signatures in the thresholds are then validated using a validation database that contains a record of successful and unsuccessful operational events, as indicated inblock 450. - As indicated in
block 460, if a valid signature or desired performance was achieved, then the results are saved as an updated fleet operational event signature (preferably in a historical operational event database and made available for future use and/or access by field personnel, as indicated inblock 470. If a valid signature is not obtained, the monitored events may be partitioned into different sets, and a signature and the corresponding fleet thresholds and boundaries may be determined individually for each set. In practice, this may correspond to a change over time in the signature for a specific unit or of different configurations across multiple units. Accordingly, the thresholds and/or parameter coefficients are adjusted and recomputed as indicated atblocks - Although not mentioned above in the detailed discussion of
FIG. 2, 3 or 4, the sensor data and/or parameter data corresponding to various “anomalous” or outlier operational events is also saved in the historical database (e.g., atprocedural blocks FIG. 5 . The diagram shows example procedural blocks which may be implemented on a computer processor/controller as part of the operational event quality assessment process to develop both system and individual component anomaly fault signatures associated with a particular unit, site or fleet-wide operational events. - As indicated at
block 500, specific storage space or files in a historical database in a computer memory are set up or allocated for use as an anomaly event database. The corrected parameter specific to the event and anomaly is identified and CPC corrections to variations associated with ambient conditions are applied, as indicated inprocedural blocks block 530, and the signature is validated then compared against existing anomaly signatures stored in the anomaly database, as indicated inblock 540. Next, assuming the validated anomaly signature is not currently in the anomaly database, it is saved along with appropriate identifying information or comments, as indicated inblock 550. - Referring to
FIG. 6 , a procedural flow diagram is shown which may serve as a general overview of an exemplary computer/controller implementation of the operational event quality assessment process. As indicated in procedural block 600, a particular operational event is initiated on the turbine and operational parameter data from various sensors is monitored during and throughout the operational event. If real-time processing is employed (block 610), an on-site real-time turbine unit controller (or a comparable remote monitoring system) is configured to recognize the particular type of operational event taking place (e.g., start-up, mode transfer, etc.) and verify that valid signals are being acquired from the various turbine sensors (block 611). Sets of appropriately verified sensor signals (i.e., verified as appropriate for the particular operational event) are immediately processed to provide real-time analysis of the event. This real-time quality assessment analysis may either be implemented locally by the turbine system site controller itself or the acquired sensor data may be transmitted via an appropriate digital communications network to a remote real-time processing facility. - As indicated at
procedural block 610, a “post-processing” arrangement may also be implemented wherein one or more system events are monitored and all of the appropriate sensor data during each event is collected and saved in a historical event/trend file which may be stored on site or at a remote facility. Subsequently, as indicated atblock 612, a particular operational event may be selected for analysis. If it exists, all prerecorded pertinent data corresponding to that particular event is retrieved from a historical event/trend file (block 614) and the sensor signal data may then be examined and verified as valid (block 616) before being passed on for further processing. - Many signals from the system sensors will be appropriate candidates for transformation to a corrected parameter space to reduce/remove the effects of ambient, fuel and machine state variability. However, for certain signals, a correction/transformation to remedy such effects will not be available or appropriate. For the signal types which can be corrected, the corresponding corrected parameters are computed (e.g., the data is transformed to a corrected parameter space), as indicated in
block 620. For those signals that do not have appropriate correction parameters but which may be critical for event assessment, the corresponding signal signatures may at least be checked to insure that the acquired data falls within pre-specified range or bounds (block 620). Accordingly, event signatures are developed from parameter plots that correspond to the acquired sensor data, at least some of which is corrected by utilizing one or more corrected parameter coefficients which reduces or eliminates variabilities in the sensor data caused by ambient conditions and/or fuel type/quality. - As indicated in
block 630, a numerical quality assessment value is then determined for the event based upon comparison with a pre-determined expected value/range and the event is then accordingly classified into an appropriate quality assessment category. Next, as indicated atblock 640, all corrected (transformed) parameter data and non-corrected (non-transformed) data are combined using, for example, a weighted average or rule-based averaging. This combined overall assessment is then classified into a “red,” “yellow” or “green” quality category according to. As indicated atblock 650, signals that fall into “suspect” categories (e.g., “red” or “yellow”) are tagged for further analysis to identify potential operational problems. Ultimately, the combined overall numerical quality assessment, as well as the quality assessment of signals/parameters that fall into suspect categories, are tracked over time to provide an early warning and identification of component or system degradation, component and system modifications and potential failures, as indicated inblock 660. -
FIG. 7 illustrates an example computer implemented process blocks for providing automated fault detection and diagnosis/identification that operates as part of the basic quality assessment processing described above. As previously discussed with respect toFIGS. 2 through 6 , a database of outlier anomaly event signatures is developed during the quality assessment processing. This database may also include historical/archival operational event data indicative of component degradation and fault signatures corresponding to both individual machine units and/or specific machine fleets. For this aspect of the invention, operational events that fall into categories indicative of poor or bad quality, such as the “red” and “yellow” quality classifications, are treated as suspect events and are used as candidates for detecting and identifying system and component faults. - As indicated at
blocks 700 through 720, a first candidate operational event is selected for examination and, if not already done, corrections for ambient conditions are applied to the event signals and a corrected parameter specific to that operational event is identified. A list of anomaly signatures associated with the selected event is formed (block 730) and then the identified event parameter is compared with each of the anomaly signatures in the list to determine if a close match exists (blocks 740 and 750). Event parameters and anomaly signatures exhibiting a high degree to matching are then used to identify the particular fault type and component or system malfunction. - Referring now to
FIG. 8 , a flow diagram is shown which illustrates an example real-time implementation of an operational event quality assessment process for a gas turbine. In this example, turbine sensor data from an operational event is provided to the local or a remote computer processor/turbine-controller 801. Using site and fleet signal threshold information obtained from a historical event database, processor/controller 801 computes a combined overall numerical assessment quality value in real-time and determines which quality category (red, yellow, green) the operational event is classified (block 810). This information is then saved in a historical tracking database as indicated atblock 820. -
FIG. 9 shows a flow diagram illustrating an example of a non-real-time computer processing implementation of the operational event quality assessment process for a gas turbine system. In this example, gas turbine sensor data during an operational event is provided to processor/controller 901 which stores the acquired information locally or sends it to a central server for performing further analysis at a later time. An on-site monitor/user interface 902 may be provided to provide a means for an operator to locally access, control and display the acquired data and results from any quality assessment and fault diagnostic analysis that is performed. 1002 also suggests that this function could be performed a remote central site. Unit, site and fleet signatures and threshold information are obtained from a historical event database and used in computing the combined overall quality assessment value and determining the quality category of the event, as indicated atblock 910. This information may then be saved in a historical tracking database as indicated atblock 920. -
FIG. 10 shows two example sets of parameter/sensor time domain sub-signature signal data plots obtained during a turbine “startup” event that are used in forming event signatures for a particular turbine machine. The left example illustrates plots of acceleration vs. time and the right example illustrates plots of percent fuel vs. time. The first action is to time align the data. As explained above, a signature is formed from a set of sub-signature data plots. Each sub-signature plot is characterized by a nominal value having associated quality thresholds boundaries/ranges (e.g., red, yellow and green) defined about the nominal value. An event signature is produced by obtaining multiple data sets and overlaying the corresponding sub-signature plots to determine the “nominal” or representative plot for each sub-signature. -
FIG. 11A shows an example diagram illustrating the computer/controller implemented processes of collection, transformation and fusion of signal data information to provide a single unified quality assessment. As shown inblock 1101, multiple time domain plots of, for example, pressure, temperature, speed, etc. are developed from the data collected, and aligned in the time domain. An arrow fromblock 1101 to block 1102 illustrates the transformation of the data ofblock 1101 into a corrected parameter space to remove the effects of ambient conditions, fuel quality and/or other known causes of variability in the data. Transformed data from this corrected parameter space is used to generate X-Y virtual parameter plots, as shown inblock 1102, that are effectively corrected for ambient conditions and systemic variations and will provide a statistically better indication of the underlying process. As also illustrated inblock 1102, a single unified assessment of success is produced by performing a probabilistic averaging of the sub-signature assessments, as described above with respect toFIG. 6 . -
FIG. 11B provides a series of graphs illustrating an example transformation of an example collected data set using the above described computer/controller implemented quality assessment processes. In this example, data from a time domain plot of compressor discharge pressure (CPD), 1103, is transformed to the corrected parameters of compressor pressure ratio (CPR) vs. corrected speed, shown ingraph 1104. This is followed by the creation of the sub-signature illustrated ingraph 1105. This sub-signature information is then used for producing a unified quality assessment as discussed above and illustrated inFIG. 11A . The appropriate correction (corrected parameter space) for the above example and for each data set for other operational variables, such as temperatures, fuel, etc., is developed using known conventional techniques familiar to those skilled in the art and typically involves using domain knowledge of the operational event, identifying the variables or signals that characterize the operational event and applying knowledge of the underlying physics of the operational event to determine either dimensionless quantities that characterize the event and minimize the effect of ambient conditions like temperature, pressure, humidity, etc. when plotted against another virtual variable. - The quality assessment information developed by the present computer implemented operational event quality assessment/diagnostic process may be output to a display device, a laptop or a printer.
FIG. 12 shows an example of an output screen display that may be produced by the computer implemented operational event quality assessment/diagnostic process upon evaluating a turbine or other machine system operational event. In this example, the machine site, ID, equipment configuration, operational event and date information are displayed in separate columns (1201) for each event assessed. In an associated “Status” column (1202), the unified quality assessment value developed for each machine fleet, machine site and machine unit for each operational event evaluated is displayed along with a color indicator showing the corresponding quality range (e.g., red, yellow or green). - While the invention has been described in connection with what is presently considered to be the most practical and preferred embodiments, it is to be understood that the invention is not to be limited to the disclosed embodiments, but on the contrary, is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims.
Claims (25)
Priority Applications (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US10/855,985 US6973396B1 (en) | 2004-05-28 | 2004-05-28 | Method for developing a unified quality assessment and providing an automated fault diagnostic tool for turbine machine systems and the like |
GB0510159A GB2414560B (en) | 2004-05-28 | 2005-05-18 | A method for developing a unified quality assessment and providing an automated fault diagnostic tool for turbine machine systems and the like |
CA2508008A CA2508008C (en) | 2004-05-28 | 2005-05-19 | A method for developing a unified quality assessment and providing an automated fault diagnostic tool for turbine machine systems and the like |
JP2005154800A JP4856396B2 (en) | 2004-05-28 | 2005-05-27 | Method for creating a unified quality assessment for turbine mechanical systems and the like and providing an automatic fault diagnosis tool |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US10/855,985 US6973396B1 (en) | 2004-05-28 | 2004-05-28 | Method for developing a unified quality assessment and providing an automated fault diagnostic tool for turbine machine systems and the like |
Publications (2)
Publication Number | Publication Date |
---|---|
US20050267702A1 true US20050267702A1 (en) | 2005-12-01 |
US6973396B1 US6973396B1 (en) | 2005-12-06 |
Family
ID=34711950
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US10/855,985 Expired - Fee Related US6973396B1 (en) | 2004-05-28 | 2004-05-28 | Method for developing a unified quality assessment and providing an automated fault diagnostic tool for turbine machine systems and the like |
Country Status (4)
Country | Link |
---|---|
US (1) | US6973396B1 (en) |
JP (1) | JP4856396B2 (en) |
CA (1) | CA2508008C (en) |
GB (1) | GB2414560B (en) |
Cited By (36)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20060058898A1 (en) * | 2004-09-10 | 2006-03-16 | Emigholz Kenneth F | System and method for abnormal event detection in the operation of continuous industrial processes |
US20060073013A1 (en) * | 2004-09-10 | 2006-04-06 | Emigholz Kenneth F | Application of abnormal event detection technology to fluidized catalytic cracking unit |
US20070061626A1 (en) * | 2005-09-14 | 2007-03-15 | Microsoft Corporation | Statistical analysis of sampled profile data in the identification of significant software test performance regressions |
US20070233428A1 (en) * | 2004-09-10 | 2007-10-04 | Emigholz Kenneth F | Application of abnormal event detection technology to hydrocracking units |
US20070250292A1 (en) * | 2006-04-21 | 2007-10-25 | Perry Alagappan | Application of abnormal event detection technology to delayed coking unit |
US20080071656A1 (en) * | 2006-09-15 | 2008-03-20 | General Electric Company | Method for constructing searchable data patterns of interest |
US20080097637A1 (en) * | 2006-03-21 | 2008-04-24 | Nguyen Anh T | Application of abnormal event detection (AED) technology to polymers process |
US7424395B2 (en) | 2004-09-10 | 2008-09-09 | Exxonmobil Research And Engineering Company | Application of abnormal event detection technology to olefins recovery trains |
US20100091022A1 (en) * | 2005-08-01 | 2010-04-15 | Noboru Shinohara | Material creation support device, material creation support system, and program |
US20100121781A1 (en) * | 2008-11-05 | 2010-05-13 | Clifford James Berg | Mechanisms for illustrating the choices in an optimal solution to a set of business choices |
CN101807164A (en) * | 2009-02-13 | 2010-08-18 | 富士施乐株式会社 | Monitoring apparatus, information processing system and monitoring method |
US20100257410A1 (en) * | 2007-06-15 | 2010-10-07 | Michael Edward Cottrell | Framework and method for monitoring equipment |
US20100262858A1 (en) * | 2009-04-09 | 2010-10-14 | Nec Laboratories America, Inc. | Invariants-Based Learning Method and System for Failure Diagnosis in Large Scale Computing Systems |
US20100268395A1 (en) * | 2007-12-11 | 2010-10-21 | Vestas Wind Systems A/S | System and method for detecting performance |
US20100274400A1 (en) * | 2009-04-22 | 2010-10-28 | Vestas Wind Systems A/S | Wind turbine configuration system |
US20110072130A1 (en) * | 2009-09-18 | 2011-03-24 | Nec Laboratories America, Inc. | Extracting Overlay Invariants Network for Capacity Planning and Resource Optimization |
US20120105199A1 (en) * | 2010-10-29 | 2012-05-03 | Cisco Technology, Inc. | Validating Sensor Data at a Community Sensor-Coordinating Entity |
FR2966878A1 (en) * | 2010-11-02 | 2012-05-04 | Snecma | System for monitoring validity of temperature measurement chain of exhaust gases of jet engine of aircraft, has computer analyzing results of statistical test on abnormality detection indicator to assess validity of chain |
US8219356B2 (en) * | 2010-12-23 | 2012-07-10 | General Electric Company | System and method for detecting anomalies in wind turbines |
US20120232806A1 (en) * | 2011-03-10 | 2012-09-13 | General Electric Company | System and method for analyzing and retrieving data obtained from turbine operations |
US20120283885A1 (en) * | 2011-05-04 | 2012-11-08 | General Electric Company | Automated system and method for implementing statistical comparison of power plant operations |
US8327189B1 (en) * | 2009-12-22 | 2012-12-04 | Emc Corporation | Diagnosing an incident on a computer system using a diagnostics analyzer database |
US20130013138A1 (en) * | 2011-07-06 | 2013-01-10 | Yinghui Lu | System and method for predicting mechanical failure of a motor |
US20130274899A1 (en) * | 2010-12-28 | 2013-10-17 | International Business Machines Corporation | Method, computer program, and computer for determining system situation |
CN103969067A (en) * | 2014-04-01 | 2014-08-06 | 中北大学 | Generalized-morphology-based automatic filling system fault diagnosis method |
US8862250B2 (en) | 2010-05-07 | 2014-10-14 | Exxonmobil Research And Engineering Company | Integrated expert system for identifying abnormal events in an industrial plant |
US20140358398A1 (en) * | 2013-03-15 | 2014-12-04 | United Technologies Corporation | Use of SS Data Trends in Fault Resolution Process |
WO2015011395A1 (en) * | 2013-07-23 | 2015-01-29 | Snecma | Method of estimation on a curve of a relevant point for the detection of an anomaly of a motor and data processing system for the implementation thereof |
EP2290233B1 (en) | 2009-08-28 | 2015-03-25 | General Electric Company | System and method for managing wind turbines and enhanced diagnostics |
US8996334B2 (en) | 2011-03-02 | 2015-03-31 | General Electric Company | Method and system for analysis of turbomachinery |
JP2015084176A (en) * | 2013-10-25 | 2015-04-30 | 日立建機株式会社 | Fault diagnostic apparatus and active machine |
WO2016040085A1 (en) * | 2014-09-10 | 2016-03-17 | Siemens Aktiengesellschaft | Gas turbine failure prediction utilizing supervised learning methodologies |
US20160195082A1 (en) * | 2015-01-02 | 2016-07-07 | General Electric Company | System and method for health management of pumping system |
US20180103049A1 (en) * | 2016-10-11 | 2018-04-12 | General Electric Company | Systems and Methods for Protecting a Physical Asset Against a Threat |
US11563814B2 (en) * | 2018-02-13 | 2023-01-24 | Omron Corporation | Session control apparatus, session control method, and program |
US20230168641A1 (en) * | 2021-11-30 | 2023-06-01 | Caterpillar Inc. | On-board machine component failure detection |
Families Citing this family (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7228461B2 (en) * | 2003-01-09 | 2007-06-05 | Siemens Energy & Automation, Inc. | System, method, and user interface for acceptance testing |
US7726169B2 (en) * | 2006-02-06 | 2010-06-01 | Westinghouse Electric Co. Llc | Method of assessing the performance of a steam generator |
EP1837506B1 (en) | 2006-03-24 | 2013-08-28 | Rolls-Royce plc | Method for monitoring thrust in gas turbine engines |
GB0606022D0 (en) * | 2006-03-24 | 2006-05-03 | Rolls Royce Plc | Monitoring gas turbine engines |
JP5127163B2 (en) * | 2006-06-12 | 2013-01-23 | 中国電力株式会社 | Coal type property evaluation feedback system |
US8261599B2 (en) | 2008-04-24 | 2012-09-11 | Rbt, Lp | Method and system for determining an imbalance of a wind turbine rotor |
US8230266B2 (en) * | 2008-06-03 | 2012-07-24 | General Electric Company | System and method for trip event data acquisition and wind turbine incorporating same |
US8185260B2 (en) * | 2009-02-12 | 2012-05-22 | Honeywell International Inc. | Prognostic and health management accuracy maintenance system and method |
US8041540B2 (en) * | 2009-12-09 | 2011-10-18 | General Electric Company | System, device, and method for acoustic and visual monitoring of a wind turbine |
US20110135473A1 (en) * | 2009-12-22 | 2011-06-09 | Singamsetti V N S Raju | System, device, and method for monitoring a wind turbine using data quality indicators |
US10137542B2 (en) | 2010-01-14 | 2018-11-27 | Senvion Gmbh | Wind turbine rotor blade components and machine for making same |
ES2794015T3 (en) | 2010-01-14 | 2020-11-17 | Siemens Gamesa Renewable Energy Service Gmbh | Wind turbine rotor blade components and methods for making them |
WO2012050474A1 (en) * | 2010-10-11 | 2012-04-19 | General Electric Company | Systems, methods, and apparatus for detecting shifts in redundant sensor signals |
US9752960B2 (en) * | 2011-11-22 | 2017-09-05 | Electric Power Research Institute, Inc. | System and method for anomaly detection |
US20140298097A1 (en) * | 2013-03-28 | 2014-10-02 | General Electric Company | System and method for correcting operational data |
JP6411769B2 (en) * | 2014-04-14 | 2018-10-24 | 株式会社日立製作所 | Condition monitoring device |
US10809156B2 (en) * | 2016-02-15 | 2020-10-20 | General Electric Company | Automated system and method for generating engine test cell analytics and diagnostics |
CN111140356B (en) * | 2019-12-25 | 2021-07-27 | 中国人民解放军陆军航空兵学院陆军航空兵研究所 | Fault assessment method and device for helicopter engine fuel system |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5225724A (en) * | 1990-10-22 | 1993-07-06 | Sgs-Thomson Microelectronics S.R.L. | Operational analysis device of the scan path type having a single scanning clock and a single output phase for an integrated circuit |
US20030034995A1 (en) * | 2001-07-03 | 2003-02-20 | Osborn Brock Estel | Interactive graphics-based analysis tool for visualizing reliability of a system and performing reliability analysis thereon |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5748500A (en) * | 1995-11-14 | 1998-05-05 | Electric Power Research Institute, Inc. | System to assess the starting performance of a turbine |
US6584434B1 (en) * | 2000-04-24 | 2003-06-24 | General Electric Company | Method for data filtering and anomoly detection |
US6343251B1 (en) * | 2000-10-20 | 2002-01-29 | General Electric Company | Method and system for monitoring the operation of and predicting part life consumption for turbomachinery |
US6643600B2 (en) * | 2001-04-26 | 2003-11-04 | General Electric Company | Method and system for assessing adjustment factors in testing or monitoring process |
US7089154B2 (en) * | 2001-08-09 | 2006-08-08 | Rovsing Dynamics A/S | Automatic machinery fault diagnostic method and apparatus |
-
2004
- 2004-05-28 US US10/855,985 patent/US6973396B1/en not_active Expired - Fee Related
-
2005
- 2005-05-18 GB GB0510159A patent/GB2414560B/en not_active Expired - Fee Related
- 2005-05-19 CA CA2508008A patent/CA2508008C/en not_active Expired - Fee Related
- 2005-05-27 JP JP2005154800A patent/JP4856396B2/en not_active Expired - Fee Related
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5225724A (en) * | 1990-10-22 | 1993-07-06 | Sgs-Thomson Microelectronics S.R.L. | Operational analysis device of the scan path type having a single scanning clock and a single output phase for an integrated circuit |
US20030034995A1 (en) * | 2001-07-03 | 2003-02-20 | Osborn Brock Estel | Interactive graphics-based analysis tool for visualizing reliability of a system and performing reliability analysis thereon |
Cited By (65)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20060073013A1 (en) * | 2004-09-10 | 2006-04-06 | Emigholz Kenneth F | Application of abnormal event detection technology to fluidized catalytic cracking unit |
US20070233428A1 (en) * | 2004-09-10 | 2007-10-04 | Emigholz Kenneth F | Application of abnormal event detection technology to hydrocracking units |
US20060058898A1 (en) * | 2004-09-10 | 2006-03-16 | Emigholz Kenneth F | System and method for abnormal event detection in the operation of continuous industrial processes |
US7349746B2 (en) | 2004-09-10 | 2008-03-25 | Exxonmobil Research And Engineering Company | System and method for abnormal event detection in the operation of continuous industrial processes |
US8005645B2 (en) | 2004-09-10 | 2011-08-23 | Exxonmobil Research And Engineering Company | Application of abnormal event detection technology to hydrocracking units |
US7424395B2 (en) | 2004-09-10 | 2008-09-09 | Exxonmobil Research And Engineering Company | Application of abnormal event detection technology to olefins recovery trains |
US7567887B2 (en) * | 2004-09-10 | 2009-07-28 | Exxonmobil Research And Engineering Company | Application of abnormal event detection technology to fluidized catalytic cracking unit |
US20100091022A1 (en) * | 2005-08-01 | 2010-04-15 | Noboru Shinohara | Material creation support device, material creation support system, and program |
US8044958B2 (en) * | 2005-08-01 | 2011-10-25 | I-N Information Systems, Ltd. | Material creation support device, material creation support system, and program |
US20070061626A1 (en) * | 2005-09-14 | 2007-03-15 | Microsoft Corporation | Statistical analysis of sampled profile data in the identification of significant software test performance regressions |
US7577875B2 (en) * | 2005-09-14 | 2009-08-18 | Microsoft Corporation | Statistical analysis of sampled profile data in the identification of significant software test performance regressions |
US7761172B2 (en) | 2006-03-21 | 2010-07-20 | Exxonmobil Research And Engineering Company | Application of abnormal event detection (AED) technology to polymers |
US20080097637A1 (en) * | 2006-03-21 | 2008-04-24 | Nguyen Anh T | Application of abnormal event detection (AED) technology to polymers process |
US7720641B2 (en) | 2006-04-21 | 2010-05-18 | Exxonmobil Research And Engineering Company | Application of abnormal event detection technology to delayed coking unit |
US20070250292A1 (en) * | 2006-04-21 | 2007-10-25 | Perry Alagappan | Application of abnormal event detection technology to delayed coking unit |
US20080071656A1 (en) * | 2006-09-15 | 2008-03-20 | General Electric Company | Method for constructing searchable data patterns of interest |
GB2462047B (en) * | 2007-06-15 | 2012-08-15 | Shell Int Research | Method and system for monitoring oilfield operations |
US20100257410A1 (en) * | 2007-06-15 | 2010-10-07 | Michael Edward Cottrell | Framework and method for monitoring equipment |
US8612029B2 (en) * | 2007-06-15 | 2013-12-17 | Shell Oil Company | Framework and method for monitoring equipment |
US9035750B2 (en) | 2007-06-15 | 2015-05-19 | Shell Oil Company | Method and system for state encoding |
US20100268395A1 (en) * | 2007-12-11 | 2010-10-21 | Vestas Wind Systems A/S | System and method for detecting performance |
EP2223048A4 (en) * | 2007-12-11 | 2014-12-03 | Vestas Wind Sys As | System and method for detecting performance |
US20100121781A1 (en) * | 2008-11-05 | 2010-05-13 | Clifford James Berg | Mechanisms for illustrating the choices in an optimal solution to a set of business choices |
US20100211814A1 (en) * | 2009-02-13 | 2010-08-19 | Fuji Xerox Co., Ltd. | Monitoring apparatus, information processing system, monitoring method and computer readable medium |
US8145955B2 (en) * | 2009-02-13 | 2012-03-27 | Fuji Xerox Co., Ltd. | Monitoring apparatus, information processing system, monitoring method and computer readable medium |
CN101807164A (en) * | 2009-02-13 | 2010-08-18 | 富士施乐株式会社 | Monitoring apparatus, information processing system and monitoring method |
US20100262858A1 (en) * | 2009-04-09 | 2010-10-14 | Nec Laboratories America, Inc. | Invariants-Based Learning Method and System for Failure Diagnosis in Large Scale Computing Systems |
US8185781B2 (en) * | 2009-04-09 | 2012-05-22 | Nec Laboratories America, Inc. | Invariants-based learning method and system for failure diagnosis in large scale computing systems |
US20100274400A1 (en) * | 2009-04-22 | 2010-10-28 | Vestas Wind Systems A/S | Wind turbine configuration system |
EP2290233B1 (en) | 2009-08-28 | 2015-03-25 | General Electric Company | System and method for managing wind turbines and enhanced diagnostics |
US20110072130A1 (en) * | 2009-09-18 | 2011-03-24 | Nec Laboratories America, Inc. | Extracting Overlay Invariants Network for Capacity Planning and Resource Optimization |
US9098342B2 (en) * | 2009-09-18 | 2015-08-04 | Nec Laboratories America, Inc. | Extracting overlay invariants network for capacity planning and resource optimization |
US8327189B1 (en) * | 2009-12-22 | 2012-12-04 | Emc Corporation | Diagnosing an incident on a computer system using a diagnostics analyzer database |
US8862250B2 (en) | 2010-05-07 | 2014-10-14 | Exxonmobil Research And Engineering Company | Integrated expert system for identifying abnormal events in an industrial plant |
US20120105199A1 (en) * | 2010-10-29 | 2012-05-03 | Cisco Technology, Inc. | Validating Sensor Data at a Community Sensor-Coordinating Entity |
FR2966878A1 (en) * | 2010-11-02 | 2012-05-04 | Snecma | System for monitoring validity of temperature measurement chain of exhaust gases of jet engine of aircraft, has computer analyzing results of statistical test on abnormality detection indicator to assess validity of chain |
US8219356B2 (en) * | 2010-12-23 | 2012-07-10 | General Electric Company | System and method for detecting anomalies in wind turbines |
US20130274899A1 (en) * | 2010-12-28 | 2013-10-17 | International Business Machines Corporation | Method, computer program, and computer for determining system situation |
EP2660675A1 (en) * | 2010-12-28 | 2013-11-06 | International Business Machines Corporation | Method, computer program, and computer for determining status of system |
EP2660675A4 (en) * | 2010-12-28 | 2017-04-05 | International Business Machines Corporation | Method, computer program, and computer for determining status of system |
US9857775B2 (en) * | 2010-12-28 | 2018-01-02 | International Business Machines Corporation | Method, computer program, and computer for determining system situation |
US8996334B2 (en) | 2011-03-02 | 2015-03-31 | General Electric Company | Method and system for analysis of turbomachinery |
US20120232806A1 (en) * | 2011-03-10 | 2012-09-13 | General Electric Company | System and method for analyzing and retrieving data obtained from turbine operations |
US20120283885A1 (en) * | 2011-05-04 | 2012-11-08 | General Electric Company | Automated system and method for implementing statistical comparison of power plant operations |
US9845012B2 (en) * | 2011-07-06 | 2017-12-19 | General Electric Company | System and method for predicting mechanical failure of a motor |
US20130013138A1 (en) * | 2011-07-06 | 2013-01-10 | Yinghui Lu | System and method for predicting mechanical failure of a motor |
US9896961B2 (en) | 2013-03-15 | 2018-02-20 | Untied Technologies Corporation | Use of ss data trends in fault resolution process |
US20140358398A1 (en) * | 2013-03-15 | 2014-12-04 | United Technologies Corporation | Use of SS Data Trends in Fault Resolution Process |
US9494492B2 (en) * | 2013-03-15 | 2016-11-15 | United Technologies Corporation | Use of SS data trends in fault resolution process |
US20160163132A1 (en) * | 2013-07-23 | 2016-06-09 | Snecma | Method of estimation on a curve of a relevant point for the detection of an anomaly of a motor and data processing system for the implementation thereof |
FR3009021A1 (en) * | 2013-07-23 | 2015-01-30 | Snecma | METHOD OF ESTIMATING A CURVE OF A RELEVANT POINT FOR ANOMALY DETECTION OF AN ENGINE AND A DATA PROCESSING SYSTEM FOR ITS IMPLEMENTATION |
RU2667794C2 (en) * | 2013-07-23 | 2018-09-24 | Сафран Эркрафт Энджинз | Method of estimation of a relevant point on a curve for detecting an anomaly of an engine and data processing system for its implementation |
CN105408828A (en) * | 2013-07-23 | 2016-03-16 | 斯奈克玛 | Method of estimation on a curve of a relevant point for the detection of an anomaly of a motor and data processing system for the implementation thereof |
WO2015011395A1 (en) * | 2013-07-23 | 2015-01-29 | Snecma | Method of estimation on a curve of a relevant point for the detection of an anomaly of a motor and data processing system for the implementation thereof |
US9792741B2 (en) * | 2013-07-23 | 2017-10-17 | Snecma | Method of estimation on a curve of a relevant point for the detection of an anomaly of a motor and data processing system for the implementation thereof |
JP2015084176A (en) * | 2013-10-25 | 2015-04-30 | 日立建機株式会社 | Fault diagnostic apparatus and active machine |
CN103969067A (en) * | 2014-04-01 | 2014-08-06 | 中北大学 | Generalized-morphology-based automatic filling system fault diagnosis method |
WO2016040085A1 (en) * | 2014-09-10 | 2016-03-17 | Siemens Aktiengesellschaft | Gas turbine failure prediction utilizing supervised learning methodologies |
CN107111309A (en) * | 2014-09-10 | 2017-08-29 | 西门子能源有限公司 | Utilize the combustion gas turbine failure predication of supervised learning method |
US9777723B2 (en) * | 2015-01-02 | 2017-10-03 | General Electric Company | System and method for health management of pumping system |
US20160195082A1 (en) * | 2015-01-02 | 2016-07-07 | General Electric Company | System and method for health management of pumping system |
US20180103049A1 (en) * | 2016-10-11 | 2018-04-12 | General Electric Company | Systems and Methods for Protecting a Physical Asset Against a Threat |
US10819719B2 (en) * | 2016-10-11 | 2020-10-27 | General Electric Company | Systems and methods for protecting a physical asset against a threat |
US11563814B2 (en) * | 2018-02-13 | 2023-01-24 | Omron Corporation | Session control apparatus, session control method, and program |
US20230168641A1 (en) * | 2021-11-30 | 2023-06-01 | Caterpillar Inc. | On-board machine component failure detection |
Also Published As
Publication number | Publication date |
---|---|
CA2508008A1 (en) | 2005-11-28 |
JP4856396B2 (en) | 2012-01-18 |
CA2508008C (en) | 2013-07-09 |
JP2005339558A (en) | 2005-12-08 |
GB2414560B (en) | 2007-07-18 |
GB0510159D0 (en) | 2005-06-22 |
GB2414560A (en) | 2005-11-30 |
US6973396B1 (en) | 2005-12-06 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US6973396B1 (en) | Method for developing a unified quality assessment and providing an automated fault diagnostic tool for turbine machine systems and the like | |
US7840332B2 (en) | Systems and methods for steam turbine remote monitoring, diagnosis and benchmarking | |
US7734443B2 (en) | System and method for prediction of gas turbine trips due to thermocouple failures | |
EP2015186B1 (en) | Diagnostic systems and methods for predictive condition monitoring | |
EP3105644B1 (en) | Method of identifying anomalies | |
US6947797B2 (en) | Method and system for diagnosing machine malfunctions | |
US9740993B2 (en) | Detecting anomalies in field failure data | |
US20170284896A1 (en) | System and method for unsupervised anomaly detection on industrial time-series data | |
US20180231394A1 (en) | Gas turbine sensor failure detection utilizing a sparse coding methodology | |
US20090037121A1 (en) | System and method for detection of rotor eccentricity baseline shift | |
US8756977B2 (en) | Developments in or relating to a condensate recovery system | |
JP6856443B2 (en) | Equipment abnormality diagnosis system | |
AU2002246994A1 (en) | Diagnostic systems and methods for predictive condition monitoring | |
US10032322B2 (en) | Validation tool for an aircraft engine monitoring system | |
JP6523815B2 (en) | Plant diagnostic device and plant diagnostic method | |
EP3663919A1 (en) | System and method of automated fault correction in a network environment | |
US6502018B1 (en) | Method for diagnosis of equipment | |
CN113255783A (en) | Sensor fault detection method and device based on unsupervised learning | |
US11339763B2 (en) | Method for windmill farm monitoring | |
Smith et al. | Automated Test Equipment Data Analytics in a PBL Environment | |
Schoeller et al. | Sensor validation and recovery for combustion turbines in electrical power generation | |
CN114526865A (en) | Valve internal leakage monitoring and diagnosing method | |
CN117808456A (en) | Equipment fault early warning method and device based on intelligent operation management | |
CN117687382A (en) | Vehicle fault checking method, system and computer medium | |
CN117640636A (en) | Cloud computing-based dynamic ring monitoring method and system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AS | Assignment |
Owner name: GENERAL ELECTRIC COMPANY, NEW YORK Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:SHAH, MINESH;RAO, KOTESH K.;NORMAN, BRUCE;AND OTHERS;REEL/FRAME:015975/0700;SIGNING DATES FROM 20040527 TO 20041103 |
|
FPAY | Fee payment |
Year of fee payment: 4 |
|
FPAY | Fee payment |
Year of fee payment: 8 |
|
REMI | Maintenance fee reminder mailed | ||
LAPS | Lapse for failure to pay maintenance fees |
Free format text: PATENT EXPIRED FOR FAILURE TO PAY MAINTENANCE FEES (ORIGINAL EVENT CODE: EXP.) |
|
STCH | Information on status: patent discontinuation |
Free format text: PATENT EXPIRED DUE TO NONPAYMENT OF MAINTENANCE FEES UNDER 37 CFR 1.362 |
|
FP | Lapsed due to failure to pay maintenance fee |
Effective date: 20171206 |