Summary of the invention:
Technical matters to be solved by this invention is the deficiency that overcomes the existing structure health monitor method, a kind of structure health monitoring method based on distributed strain dynamic test is provided, adopt strain transducer structure distributed sensing network and reference sensor is set, carry out model analysis and obtain with distributed strain modal vector (Distributed Modal Strain Vector, be abbreviated as DMSV) be the structural system characteristic of core, reach the purpose of accurate recognition structure damage, dynamic monitoring structural health.Described structure health monitoring method based on distributed strain dynamic test has that error is little, precision is high, diagnostic procedure is simple and advantage such as monitoring display is directly perceived, can be uncertain at the load value, under the condition that structure mathematics is unknown, get rid of the interference of noise and environment, the generation and the position thereof of capturing structure damage effectively and accurately, real-time online ground carries out structural damage detection and monitoring structural health conditions.
It is as follows that the present invention solves the technical scheme that its technical matters takes:
A kind of structure health monitoring method based on distributed strain dynamic test, it realizes that by sensing network, sensor signal demodulation acquisition system and data processing monitoring system this method comprises the following steps:
(1) makes up the distributed sensing network: in the zone of monitored structure, arrange some strain transducers, reach the scope that whole strain transducer tests and to cover the Zone Full of the required monitoring of this structure, to constitute the distributed sensing network;
(2) determine reference sensor: in some strain transducers of above-mentioned layout, as the reference sensor, and judge its validity according at least one strain transducer of following Standard Selection, other strain transducer is as common sensor:
(a) the structure not damaged of this reference sensor region;
(b) response of the structural strain of this reference sensor region should not be too little;
(3) data acquisition and rough handling: load application on monitored structure, under this load excitation, produce the raw data of strain-responsive through the structure that the collection of sensor signal demodulation acquisition system obtains on by above-mentioned distributed sensing network, with the described data processing monitoring system of it input, and tentatively judge the validity of these data;
(4) model analysis: above-mentioned data are carried out model analysis, obtain distributed strain modal vector DMSV, each component δ of this distributed strain modal vector DMSV corresponds respectively to each strain transducer, and it is considered the mutual ratios relation between them and ignores the size of absolute value; The ratio of the component of each common sensor constitutes the target feature vector α of structure to the component that replaces the table reference sensor again as a group component with other respectively;
(5) component of realtime graphic display-object proper vector α: the coordinate diagram of on the visualization interface of data processing monitoring system, setting up each component images of expression target feature vector α, this coordinate diagram is the x axle with the component of reference sensor among the distributed strain modal vector DMSV, component with common sensor among the distributed strain modal vector DMSV is the y axle respectively, marks the picture point of the component of the target feature vector α that characterizes corresponding each common sensor in each coordinate diagram respectively in real time;
(6) damage identification and location: afterwards under the condition of load application, continuously image data, carry out model analysis and the figure of display-object proper vector α component in real time on visualization interface; If in certain period, institute's image data departs from former target feature vector α, and has obviously stable rule and trend, show that then new damage takes place monitored structure, damage location determines in the layout area corresponding to the strain transducer of the target feature vector α component that abnormal graph occurs, otherwise then shows monitored structure that new damage does not take place and be in the original state of health maintenance.
(7) data compression and storage: in time keep with the relevant damage of storage before, in the damage process and damage take place and stablize after the data of target feature vector α.
Structure health monitoring method based on distributed strain dynamic test of the present invention, it is a distributed long gauge length optical fibre Bragg optical grating strain sensor in order to the strain transducer that constitutes the distributed sensing network, the not damaged zone at described reference sensor place be on the monitored structure suffered internal force not quite, away from weak part and the good section of Service Environment; In the described method step, be the time dependent dynamic load of numerical value to the load that monitored structure applied, the minimum sampling frequency of data is greater than the twice of monitored structure fundamental frequency, and timely time-histories data of deleting the monitored structure of being gathered; The data processing monitoring system that this method adopts is a computer.
Compare with traditional monitoring method to building structure health, the structure health monitoring method based on distributed strain dynamic test of the present invention has reached following beneficial effect:
At first, the present invention has adopted distributed long gauge length optical fibre Bragg optical grating (FBG) strain transducer to make up sensing network, overcome traditional strain transducer and can only do a distribution, carry out the defective that " point " measured, utilize distributed long gauge length optical fibre Bragg optical grating strain sensor in its long gauge length zone, to record mean strain, realize the characteristics of enhanced sensitivity and temperature compensation simultaneously, constitute the distributed sensing network, the scope that makes this sensing network test can cover whole monitored structure as wide as possible and carry out " distributed measurement ", thereby make measurement data comprise the information of monitored structure as far as possible all sidedly, reach more effective, catch the generation of unpredictable structural damage and the purpose of position thereof exactly.
Secondly, it is the structural system characteristic of core that the present invention has adopted with distributed strain modal vector DMSV, the method that utilization originally simultaneously is provided with reference sensor is introduced the target feature vector α of structure, and the variation of this target feature vector α is taken place as the monitored structural damage of identification and the criterion of position.Therefore this method has been brought following advantage to the present invention: (1) is because distributed strain modal vector DMSV is the build-in attribute of monitored structure, irrelevant with external loads, therefore test needn't be in advance to adding load make restriction, also needn't stop the operate as normal of monitored structure, can under the condition of dynamic load, carry out online monitoring, so structure health monitoring method of the present invention is loose to the conditional request that detects, process simple, it is convenient to implement, and helps structure is carried out the online real-time health monitoring of long-term dynamics.(2) described distributed strain modal vector DMSV is based on that the structural strain data of being gathered directly constitute, and do not need the mathematical model of monitored structure, thereby the present invention can avoid structural model distortion and systematic analysis to calculate the error of bringing.(3) the present invention to the variation tendency of target feature vector α in a time period repeatedly repeatedly image data just obtain, so the variation of this target feature vector α has statistical significance, thereby can eliminate the accidental error in the measurement, reduce the adverse effect that measurement noise and environmental perturbation bring.(4) the present invention is with the variation of the target feature vector α criterion as the recognition structure damage, and each strain transducer has directly one-to-one relationship clearly in the component of this target feature vector α and the distributed sensing network, so just inevitable variation that only influences the component of this position upper sensor response accordingly of the damage of somewhere positional structure, and other component is not exerted an influence, so by the generation and the position thereof that can capture structural damage immediately, thereby improved the validity and the accuracy of damage identification, simplified diagnostic procedure.
Again it, therefore the present invention adopts the coordinate diagram picture of each component images of the direct display-object proper vector of the visualization interface of data processing monitoring system α, can observe monitoring result intuitively, helps real-time online ground and carries out monitoring structural health conditions work.
At last, because structure health monitoring method of the present invention only keeps before and after the relevant damage with storage in observation process and the data that produce in the process, the redundant data that deletion in time is irrelevant therewith so saved the equipment storage resources, helps improving information storage and efficiency of managing.
Embodiment:
The invention will be further described below in conjunction with specific embodiments and the drawings.
At first see also Fig. 1 hardware system structure synoptic diagram of the present invention, described structure health monitoring method based on distributed strain dynamic test is to be used for engineering structure or product structure are carried out structural damage or degeneration under the condition of load real time on-line monitoring, and its employed hardware system includes sensing network 1, sensor signal demodulation acquisition system 2 and data processing monitoring system 3.Described sensing network 1 is in order to the response of induction structure, described sensor signal demodulation acquisition system 2 is gathered the raw data of monitored structural response in order to separate mediation, it can be the sensor (FBG) demodulator, described data processing monitoring system 3 generally adopts computer in order to data processing, storage, management and monitoring result output.
Now with shown in Figure 2 be the content that embodiment describes the structure health monitoring method based on distributed strain dynamic test of the present invention in detail by geodesic structure.Monitored structure shown in Figure 2 is that a cross section is the free beam of 50 * 3mm, and the spacing of two ends constraint is 1m.Now the observation process with real-time online is described as follows.
The first step makes up distributed sensing network 1.
On monitored free beam, arrange some strain transducers, promptly as shown in Figure 3, arrange 4 strain transducer F1~F4 equably in the bottom of free beam.This strain transducer F1~F4 is distributed long gauge length optical fibre Bragg optical grating (FBG) strain transducer, its gauge length is free beam span length's 1/4, the length that is its outer sleeve is 0.25m, this distributed long gauge length FBG strain transducer F1~F4 can record mean strain in its long gauge length zone, realize enhanced sensitivity and temperature compensation simultaneously.They are end to end, in series be distributed in free beam vertically on, the scope that therefore whole distributed long gauge length FBG strain transducers are tested can cover the Zone Full of the required monitoring of this free beam just, thereby constitute distributed sensing network 1, this distributed sensing network 1 is connected with sensor signal demodulation acquisition system 2 via the optical cable of band FC/APC joint.The scope that such distributed sensing network 1 is tested can cover whole monitored structure as wide as possible, and it can make any damage on the monitored simple beam structure all can fall into the scope that a certain strain transducer detects.Certainly, the difficult all standing layout that realizes as embodiment in actual engineering, but adopt distributed long gauge length FBG strain transducer to make up sensing network and realize distributed measurement, always the strain transducer that distributes than " point " formula of employing carries out the generation that point measurement more can be caught unpredictable structural damage effectively.
Second step was determined reference sensor.
In some strain transducer F1~F4 of above-mentioned layout, select at least one strain transducer as the reference sensor, other strain transducer is as common sensor, in the present embodiment we to select strain transducer F1 be reference sensor, other strain transducer F2, F3, F4 then are common sensor.The principle of selected reference sensor should guarantee that reference sensor F1 is installed in the undamaged zone of structure, although practical structures damage be at random, can't predict in advance, but judge through theoretical analysis and experience, still can be chosen in the position of damage probability of happening minimum as far as possible.Its selection is generally carried out according to following standard:
(a) the structure not damaged of this reference sensor region, in general such zone be on the monitored structure suffered internal force little, away from weak part and the good section of Service Environment;
(b) response of the structural strain of this reference sensor region should not be too little, in order to avoid the measured noise of real structural response is fallen into oblivion.
Judge the validity of reference sensor F1 then, promptly check this reference sensor F1 itself whether to lose efficacy on the one hand, check this reference sensor F1 whether to be installed in really on the position of non-structure damage on the other hand.This checking can be undertaken by number of ways, for example: can check directly targetedly whether the position of this reference sensor F1 is damaged with the method for Non-Destructive Testing; Also can repeat to arrange two sensors in this position, relatively whether both measurement results are consistent; The measurement data of all right Direct observation this sensor F1 in certain period if these a series of measurement data instabilities or the overwhelming majority have all departed from regime values regularly, then can be queried the validity of this reference sensor F1.
In the present embodiment, we have only selected a strain transducer F1 as the reference sensor, and in order to prevent that this reference sensor F1 from causing the not normal rough sledding that " reference " can't be provided of measurement data because of damage takes place in the position, usually the strain transducer conduct simultaneously that can select an above quantity is with reference to sensor, thereby can contrast the results of structural analysis that adopts different reference sensor gained, avoid the monitoring result that may cause invalid because only adopt a reference sensor.
The 3rd step data collection and rough handling.
After finishing sensing network layout and reference sensor selection, load application on monitored free beam at first, this load can be a dead load, it also can be dynamic load, this just means, structure health monitoring method of the present invention can be under the normal running conditions of dynamic load in monitored structure, carries out the monitoring of real-time online.So-called herein " dead load " is meant big or small time-independent load, i.e. the constant load of constant magnitude; " dynamic load " is meant big or small time dependent load, and promptly numerical value is passed and the load that changes in time.In the present embodiment, adding load is the dynamic load that single-point impacts.
Gather then by above-mentioned distributed sensing network 1, promptly the raw data of the strain-responsive that produced under the excitation of this single-point impact load of the free beam that obtains on strain transducer F1~F4 as shown in Figure 4, obtains 4 strain time-histories records.In order to realize dynamic test, the minimum sampling frequency of data is decided on monitoring target, general twice greater than monitored structure fundamental frequency.
The raw data of strain-responsive is imported described data processing monitoring system 3 thereupon, and tentatively judged the validity of these data according to the rule of Measurement and Data Processing.Whether judgment data validity can have unusually and carry out by observing these data, if find to have unusual, then analyzes the reason of its generation and determines the choice of these data at this reason.
The 4th step model analysis.
At first in computer, above-mentioned data are carried out model analysis, in conjunction with to the determining of dynamic exciting form, ask for the method for proper vector, obtain distributed strain modal vector DMSV according to test modal analysis.Ask for the method for proper vector here in any test modal analysis, all be applicable to and make up distributed strain modal vector DMSV, its method has a lot, for example amplitude method, admittance circule method, least square complex exponential method, ERA method or the like, these methods are known technology in the industry, give unnecessary details no longer one by one at this, what use in the present embodiment is amplitude method (see figure 4).
Described distributed strain modal vector DMSV is the build-in attribute of structure, and is irrelevant with external loads.Its mathematic(al) representation is:
DMSV={δ
1r,δ
2r,…,δ
Nr}
T
In the formula, δ is each component of DMSV,
N is the quantity of strain transducer in the sensing network,
R is the exponent number of strain-responsive function.
By following formula as can be known, each monitored structure can be considered r DMSV, and its quantity can be chosen according to the needs of analytical calculation, and therefore low order measuring accuracy height adopts the first rank measurement result proper generally speaking; Each DMSV has N component δ, and promptly the number of components of DMSV is identical with the strain transducer quantity of being arranged.Each component δ of this distributed strain modal vector DMSV corresponds respectively to each strain transducer, and it is considered the mutual ratios relation between them and ignores the size of absolute value.
In the present embodiment, N=4, r=1, then the DMSV={ δ of free beam
11, δ
21, δ
31, δ
41}
T, δ wherein
11Be reference sensor component, δ
21, δ
31, δ
41Be common sensor component, they be the amplitude of this free beam of obtaining of corresponding FBG sensor measurement frequency response function of strain under the excitation of single-point impact load, and only considered proportionate relationship each other and ignored the size of absolute figure.
For the different health status (as before damaging and damage back) of monitored structure being discerned and being judged, the distributed strain modal vector DMSV that the relatively different periods are obtained, just need carry out the numerical value processing and make it standardization its each component δ, therefore the present invention introduces and makes up one special in to discern the important structure parameter of monitored structural damage: the component that target feature vector α, this target feature vector α replace the table reference sensor respectively with other ratio of the component of each common sensor as the group component of itself.That is:
Target feature vector
In the formula, α is each component of target feature vector α,
δ is each component of DMSV,
N is the quantity of strain transducer in the sensing network,
N is the quantity of reference sensor,
R is the exponent number of strain-responsive function,
B is the label (b=b of reference sensor
1, b
2... b
n).
Following formula shows that if selected n strain transducer as the reference sensor, then target feature vector α just is made of (N-n) individual component.
In the present embodiment, target feature vector α has 3 components:
DMSV is the same with the distributed strain modal vector, and target feature vector α is the build-in attribute of structure, and is irrelevant with external loads, that is to say, analysis and judged result and external loads size are irrelevant.Therefore use the method for the invention structure monitored, to adding load needn't make restriction in advance, also needn't stop the operate as normal of monitored structure, so that this method is implemented is convenient, help structure is carried out the online real-time health monitoring of long-term dynamics.In addition, each strain transducer has directly one-to-one relationship clearly in each component of this target feature vector α and the distributed sensing network, for example in the present embodiment, and α
11Corresponding F2, α
21Corresponding F3, α
31Corresponding F4.So just inevitable variation that only influences the component of this position upper sensor response accordingly of the damage of position, somewhere, and other component is not exerted an influence, this is the occurrence positions that seizure rapidly damages, and improves accuracy, the simplification diagnostic procedure of damage identification and has created condition.
The component of the 5th step realtime graphic display-object proper vector α.
At first on the visualization interface of data processing monitoring system 3, set up the coordinate diagram of each component α figure of expression target feature vector α, this coordinate diagram is the x axle with the component of certain reference sensor among the distributed strain modal vector DMSV, component with each common sensor among the distributed strain modal vector DMSV is the y axle respectively, because target feature vector α is made of (N-n) individual component, so can set up altogether ((N-n) * n) individual x-y coordinate diagram (if consider the exponent number r of strain-responsive function, the ading up to of coordinate diagram ((N-n) * n * r) individual) then; In each coordinate diagram, mark the coordinate points of the component α of the target feature vector α that characterizes corresponding each common sensor then respectively in real time: for example at component α
1rCoordinate diagram on to mark the x-y coordinate be (δ
Br, δ
1r) representative component α
1rPoint (b ≠ 1), other is by that analogy.Can show simultaneously on the visualization interface of this data processing monitoring system 3 that ((N-n) * n) individual coordinate diagram picture also can show each coordinate diagram item by item through switching for all.
See also Fig. 4, present embodiment has been set up on the display of computer and has been represented each component α of free beam target feature vector α respectively
11, α
21, α
31Three x-y coordinate diagram.The x axle is represented δ in the 1st coordinate diagram
11Component, y axle are represented δ
21Component marks then that (x, y) coordinate figure is (δ
11, δ
21) expression α
11The picture point of component; In like manner, mark respectively in the 2nd, the 3rd coordinate diagram that (x, y) coordinate figure is (δ
11, δ
31), (δ
11, δ
41) correspondence represent α
21, α
31The picture point of component.Structure health monitoring method of the present invention utilizes the coordinate diagram picture of each component images of the direct display-object proper vector of the visualization interface of data processing monitoring system 3 α, therefore can observe monitoring result intuitively, help real-time online ground and carry out monitoring structural health conditions work.
Damage identification of the 6th step and location.
In implementing the process of health monitoring to the continuous load application of structure, and constantly the repeated acquisition data, carry out model analysis and on visualization interface the real-time series of steps such as figure of display-object proper vector α component, the maximum sample frequency of data is greater than the twice of monitored structure fundamental frequency.Through a time period repeat repeatedly measure after, on each coordinate diagram of data processing monitoring system 3 visualization interfaces, a plurality of picture point of expression different measuring each component of target feature vector α constantly are recorded demonstration, to form a curvilinear figure after on each coordinate diagram these match, in fact the curve tangent line on the each point, promptly slope of a curve is exactly the value corresponding to the target feature vector α component of this this strain transducer of moment.
Damage identification and location thereof according to following criterion subsequently: if in certain period, the data of being gathered depart from former target feature vector α, be that great changes will take place for shown slope of a curve, and has obviously stable rule and trend, show that then new damage takes place monitored structure, damage location determines in the layout area corresponding to the strain transducer of the target feature vector α component that abnormal graph occurs, otherwise then shows monitored structure that new damage does not take place and be in the original state of health maintenance.This is because in theory, if structure at this moment in the section not damaged take place, these picture point should linear dependence, be that all curves after the match are straight line, in other words, the target feature vector α no change of this monitored structure of period, promptly each component images slope of a curve remains unchanged; Otherwise if damage in the zone of structure interior certain strain transducer test of section at this moment, then the target feature vector α component slope of a curve corresponding to this strain transducer will change.
Among the present invention, described distributed strain modal vector DMSV and target feature vector α directly utilize the structural strain data of being gathered to constitute, the mathematical model that does not need monitored structure, thereby avoided structural model distortion and systematic analysis to calculate the error of bringing, in addition, the present invention to the variation tendency of target feature vector α in a time period repeatedly repeatedly image data just obtain, so the delta data of this target feature vector α has statistical significance, thereby can eliminate the accidental error in the measurement, reduce the adverse effect that measurement noise and environmental perturbation bring, improved the precision of system performance analytical structure monitoring result.
In the present embodiment, in a period of time, free beam is constantly applied impact load (position and amplitude are not limit), and image data and carry out model analysis constantly, on computer monitor, just demonstrate coordinate graphic as shown in Figure 4.As can be seen from the figure, in the time 1 of monitoring, correspond respectively to the target feature vector α component α of common strain transducer F2, F3, F4
11, α
21, α
31Three matched curves significant change is not all arranged, its slope is constant substantially, this illustrates that free beam does not damage in this period, is in the state of keeping of former unsoundness.In the time 2 of monitoring, corresponding to the α of common strain transducer F3, F4
21, α
31Curve does not have significant change yet, and its slope and the rate of curve of time 1 are basic identical, and corresponding to the component α of common strain transducer F2
11The matched curve slope clear regularity has taken place to change and shown, the data of this explanation strain transducer F2 have departed from former target feature vector α, can judge this moment, the position A of this period inherent layout strain transducer F2 goes up tested free beam damage (seeing the synoptic diagram of time 2 among Fig. 3) has taken place, and can in time send the damage early warning simultaneously.
Compression of the 7th step data and storage.
In the data acquisition and processing (DAP) process of structure monitoring, if the data that obtain do not depart from former target feature vector α all the time, just in time delete the redundant time-histories data of the sign pilot process of measuring process accumulation, these time-histories data after observation process in without any value, and only keep and store data about target feature vector α.That is to say, when monitored structure keeps original health status and during all-the-time stable, in time this period of deletion gather with the irrelevant time-histories data of target feature vector α, only keep and the data of storing relevant target feature vector α itself, so that the capacity of the storage resources of the equipment of saving, compressed database, raising information storage and efficiency of managing.If in certain period, the data that obtain have departed from former target feature vector α and have shown clear regularity, illustrate that then monitored structure in the section damage has taken place at this moment, at this moment just not only to keep and store the data of the fresh target proper vector α that tends towards stability after preceding former target feature vector α of damage and the damage, and the data that will keep and store change procedure between the curve of two sections Different Slope, to understand the moment that damage takes place and the overall process of development.
By above-mentioned to Monitoring Data selectively house get, last historical data under preserving in data processing monitoring system 3 includes only three contents: the target signature data before the damage, the data of damage process and damage take place and stable after the target signature data.Meanwhile also realized the optimization of information storage and management.
In the present embodiment, in the monitoring time 1 that target feature vector α data remain unchanged, computer is left out the middle time-histories data of basically identical at any time, only keeps and the data of storing relevant target feature vector α itself, so that soar the more data storage area; In monitoring time 2, because the data of strain transducer F2 have departed from former target feature vector α, point out the A place of tested free beam that damage has taken place, therefore computer with keep with the storage damage before in (in the time 1), the damage process and damage the back takes place and tend towards stability after whole detection data of relevant target feature vector α, information accumulation work is carried out in post analysis research for it.
For the analysis of the target feature vector α of n>1 and r>1 and basic identical in order to the monitoring method of identification of damage and process and the foregoing description.
In sum, structure health monitoring method based on distributed strain dynamic test of the present invention, it adopts the distributed strain sensor to make up the distributed sensing network and selects to be provided with reference sensor, carry out the structural system model analysis and obtain distributed strain modal vector DMSV, originally whether target feature vector α is changed criterion, obtained rapidly, the effect of capturing structure damage accurately and efficiently and location thereof as Damage Assessment Method.The present invention has that error is little, precision is high, diagnostic procedure is simple and advantage such as monitoring display is directly perceived, can be uncertain at the load value, under the condition that structure mathematics is unknown, get rid of the interference of noise and environment, real-time online ground carries out structural damage detection, be applicable to various engineering structures the especially long-term health of civil structure monitoring.