DE10257793A1 - Model based service life monitoring system, especially for forecasting the remaining service life of motor vehicle components, whereby existing instrumentation is used to provide data for a model for calculating wear - Google Patents

Abstract

Model-based service life monitor comprises the following modules: module for determining the system loading from available instrumentation, especially motor vehicle instrumentation, and processing of sensor signals to determine loading behavior; module for determining component loading and loading behavior and; a calculation module for determining the remaining service life of components from the measured accumulated component damage and operating strength analyses.

Description

The invention relates to a model-based Lifetime observer for the calculation of the remaining life of a selected component. The model-based Lifetime observer linked the measurement of operational loads by means of available sensors with a model-based preparation of the measured loads to local Component load profiles and their rating with a damage accumulation calculation.

From the DE 199 37 203 A1 is a monitoring of the life of a warehouse known. By means of sensors in the warehouse, the loads at the warehouse are measured on site. From the bearing geometry, the structural deformation of the bearing under the measured loads is calculated by means of a finite element program. In a further step, the local roller load is determined from the deformation of the bearing and summed up over the service life of the bearing. From the total applied bearing load, the residual life of the bearing rollers and thus of the bearing is then determined by means of a damage progression calculation. All monitoring is done after calibration of the sensors in real time.

Lifetime monitoring of the aforementioned Art have the disadvantage that the Sensors for determining the component loads as close as possible to the component, whose life is monitored should be, must be appropriate. In component monitoring of multi-body systems, their individual assemblies form a complex model, therefore, must at least for the individual assemblies are always subjected to load measurements. Will you in this way a lifetime observation of all components realizing the overall model in real time increases the effort Sensor technology enormously.

The invention is therefore to Task the life of individual components or groups of components within an overall model with as possible few sensors to realize.

The task is solved with a Model-based life observers with the features of the claim 1. Further advantageous embodiments are in the subclaims in the embodiments contain.

Proposed is a model-based Lifetime observer, with the following modules:

- Determination of system loadings:

First of all, the different ones have to the existing system sensors, in particular a vehicle sensor available Analyzed sensor signals in a step of signal preprocessing, be evaluated and integrated. The goal is the time courses of the original Sensor signals corresponding time courses for a complete and to generate consistent set of system loads.

This initially includes the standard methods a signal processing, such as the elimination of unwanted Measuring and interference signals, the evaluation of signal accuracy, the selection of relevant time windows or a frequency analysis with filtering and smoothing. In addition, in usually signals from different sensors correlated with each other and, if appropriate, the time course of a corresponding system load can be integrated.

- Model-based determination of component loads:

The determined system loads have to the model-based determination of local component-specific stresses enable. This is achieved by the system sensors, in particular from the vehicle sensor system, the extracted set of system loads the Stimulation of a component-specific model consistent in time and completely describes and thus allows a simulation of the system behavior. By temporal integration of the model behavior under influence of the complete Set of time-dependent System loads then become time courses of local component stresses determined. These component stresses can, for example, in the form of a Time course of local reaction forces, stresses or strains at the selected Component present.

- calculation of component damage:

The evaluation of the component damage accumulated as a result of the experienced stresses takes place in the environment of an analysis of the fatigue life by evaluation of the stress time courses in one classical damage accumulation calculation. The time courses of the stresses are first compressed under targeted loss of information in the collective of demands. In principle, freely selectable evaluation intervals can be integrated with the help of a suitable description of the tolerable stress (eg a Wöhler line of the considered component) and a damage hypothesis to a incurred in the evaluation interval component damage or a prognosis for the remaining life remaining in the meantime accumulated.

This results mainly in the following advantages:
By including the system model, it is no longer necessary to equip each component whose service life is to be monitored with its own sensor system. As a result, the sensor effort can be reduced considerably.

Especially in the case of application in a vehicle, the vehicle models are already consistently as today CAD models in front. Also in motor vehicles of today's design are numerous Sensors for the control of various mechatronic components available. For vehicles with active chassis, sensors are for determining the time courses from Radnabenkräfte available. When the vehicle is usually a major part of Forces over the Wheel hubs introduced into the overall vehicle model. Another source of power is the drive of the vehicle. But here is one today electronic power control of the internal combustion engine available, the above has built-in sensors. Here are addressed crankshaft sensor, torque sensors. This Already existing sensors result in a complete detection of the in one Vehicle model each selectively introduced forces. By the inclusion of the vehicle model in the lifetime monitoring can from this existing sensor kit on the life of almost all Components or component groups are closed in the vehicle. Also on the life of such component groups, by the introduction of force are relatively far away. It is therefore necessary in the case of a modern vehicle no additional Sensors to one as possible full Life monitoring perform.

By accumulation of the stresses in load collectives and evaluation of the load collective can a residual life monitoring introduced and from the remaining life a maintenance profile for the observed System or the observed vehicle.

Embodiments of the invention will be without limitation the general public explained on the basis of drawings.

Showing:

1 The functional diagram of a model-based lifetime observer;

2 The functional diagram of a model-based observation of the available remaining lifespan;

3 A comparison of two possible versions of a model-based lifetime observer:

4 The functional diagram of a model-based stress analysis; 5 The functional diagram of an integration of measurement data and models;

6 A model-based load determination; 7 A complete vehicle model;

8th A modular overview of the complete vehicle model; 9 A rear axle model;

10 A possible structural representation of the rear axle model for the rear axle 8th ;

11 The flowchart of a damage calculation;

12 Strength areas based on a Wöhler line;

13 Determination of the tolerable level of stress;

14 Tool chain and interfaces for modeling and damage calculation;

15 FEM-MBS interface in SIMPACK;

16 Input information and functionality of the damage calculation;

17 Four different possible implementations of a model-based lifetime observer;

18 Schematic representation of a possible procedure for determining the remaining service life;

19 A scheme for identifying standard usage cases;

Description of the embodiments:

For A calculation of the remaining life of a selected component is information about local Demands, ie time courses local forces, Tensions or strains required. Vehicles that produced in larger quantities are for cost reasons only with the minimal for equipped with the required set of sensors, so that sensors, the information about the mechanical stress individual components in daily Deliver operation, usually not available.

The following is based on the 1 The concept of a model-based life observer for selected, mechanically stressed components of a vehicle can be represented. It combines a measurement of the operational loads by means of available vehicle sensors with a model-based preparation of the measured loads for local component load profiles and their evaluation with the aid of a damage accumulation calculation.

From the signals of existing vehicle sensors can directly or indirectly the time functions of the operational charges be determined. Using suitable models, e.g. by Modeling of a elastic multibody system can, under certain restrictions, from the time functions of the sensor signals local operational component stresses extracted and in the form of suitable classification methods be represented by stress collectives. A damage accumulation calculation should using geometry and cyclic material properties of the component (Wöhler line of the Component) the total damage of the component accumulate under the action of the respective stress collective and predict remaining life. A data store allows the recording of intermediate results and separates one time-consuming and memory-critical high-frequency data channel of Signal recording from a low-frequency (offline) data channel the evaluation.

The objective of the model-based lifetime observation for a selected component is the determination of the damage accumulated over the entire useful life of the component under consideration or the prognosis for its currently still expected remaining service life. This requires a determination and evaluation of the local stresses of the component occurring in operation in the context of a damage calculation. (Please refer 2 )

The prediction of the component remaining service life it should be based on information that already exists Vehicle sensors during the operational use, as well as, using a suitable model, by a simulation of the associated component stresses respectively. Finally The determined component stresses are based on evaluated by material properties and a damage hypothesis.

A lifecycle observer, who accumulates the component-specific damage during vehicle use with the help of a model, is therefore functionally divided into three components:
Determination of system loadings: First, the various sensor signals available from the existing vehicle sensor system must be analyzed, evaluated and integrated in a signal pre-processing step. The aim here is to generate corresponding time profiles from the time sequences of the original sensor signals for a complete and consistent set of system loads.

This initially includes the standard methods a signal processing, such as the elimination of unwanted noise, the evaluation of the signal accuracy, the selection of relevant time windows or a frequency analysis with filtering and smoothing. In addition, in usually signals from different sensors correlated with each other and, if appropriate, the time course of a corresponding system load can be integrated.

Model-based determination of component loads: The system loads determined must be the model-based determination allow for local component-specific stresses. This only works if the set of system loads extracted from the vehicle sensor system the stimulation of a component-specific model is temporally consistent and completely describes and thus allows a simulation of the system behavior. By temporal integration of the model behavior under influence of the complete sentence the time-dependent System loads then become time courses of local component stresses determined. These component stresses can, for example, in the form of a Time course of local reaction forces, stresses or strains at the selected Component present.

The scope and system limits required for determining the component-specific stresses Hang model strong on the validity of the information from the available vehicle sensors for the considered component. In any case, the model must be the determination of stresses, the following an assessment of the injury allow the component in the context of an operating strength calculation. This must be the Component in the system model used with the help of a suitable behavioral description and using physical and geometric parameters be imaged.

Calculation of component damage: The evaluation of accumulated as a result of the experienced stresses component damage takes place, in the context of a fatigue analysis, by evaluation the stress time courses in a classical damage accumulation calculation. The time courses of the Stresses are first under targeted loss of information in collective workloads compressed. In principle freely selectable evaluation intervals can the accumulated in the meantime stress collective with Help of a suitable description of the sustainable use (e.g., a Wöhler line of the considered component) and a damage hypothesis to a in the evaluation interval incurred component damage or a forecast for the remaining life remaining to be integrated.

The direct observer principle is to stimulate the observer, who allows to simulate system behavior model-based, with the same stimulation as the system to be observed. By comparison In accordance with the system and observer output variables, the behavior of the observer model can be corrected in a control process on the basis of the real system behavior.

Due to the restrictions, the ones available in the vehicle Sensors for measuring and tracking the condition of the entire vehicle imposed, can be this principle for the model-based lifetime observation is usually not immediate realize. The suggestion of the considered system would have to be the same Completed be detected by the existing vehicle sensors, such as the to control the observer relevant system reactions. This requires a component-specific acquisition of state variables which, at least in a production vehicle, not necessarily available.

It is possible to supplement the limited information about the state of the entire vehicle from the vehicle sensor system for the application of the model-based lifetime observation, in which an 'environment model' is integrated into the model used. If the excitation of the target system under consideration can not be measured directly, the scope of the system model used and its external interfaces can be adapted to the available information from the vehicle sensor system. The model used then contains the actual observer model as a subsystem, supplemented by the environment model, which generates the necessary excitation for the observer model from the available information from the vehicle sensor system. The environment model serves as an interface between the signals of the vehicle sensor system and the actual model for lifetime observation of a selected component. (please refer 3 )

In the strict sense it could as part of the signal preprocessing, but should here as it is part of the complete Simulation model, than for model-based stress analysis belonging to be viewed as.

Scope of the required system model

The scope of the model-based Lifetime observation needed Environment model depends heavily from the validity of the available Vehicle sensors for the considered component from.

Are those for the model-based lifetime observation needed a component Information immediately before, is given suitable preprocessing and integration of the sensor signals no actual environmental model needed. The more indirect but the vehicle sensor information about the State and behavior of the selected component provides, all the more this must be more extensive System model, whose simulation of the model-based lifetime observation this component is used. The outer interface of the needed Model is determined by the available vehicle sensors.

In the model-based lifetime observation are given from a given model using a physical model Set of signal time courses the vehicle sensor time courses generated local loads of a selected component. In This physical model is made up of subsystems which the model is composed of defines interactions, the i.a. from the states depend on the connected subsystems. In a mechanical system can this state-dependent Forces or the dynamics of the subsystems are constraining constraints. The actual state of the subsystems determines the interaction and thus the further temporal development of the linked subsystems. This physical structure of the interactions can on the one hand the necessary scope of the system model for determining the loads restrict a component, On the other hand, this results in minimum requirements for the required input signals.

Completeness condition: outgoing from the considered component all interaction chains traceable to model excitation be defined as the system loads from vehicle sensors is. The system loads that can be determined from the vehicle sensor system define model scope as the scope of the required model. A minimal required set of system loads results from the condition in the model, the interaction chains to the component with corresponding Model suggestions complete to be able to cover. This is to measure the different arrows in the illustrated model for different sets Illustrate system load. Areas of the system that on the Component non-reactive are not needed to be included in the model.

Causality condition: In principle, it would be conceivable by evaluation of an additional Sensors in an already fully determined by appropriate system loads We chain back to an impact chain that is not measured by an associated one System load is set. An example of this would be the inference from the sensor signals measured in the vehicle to the parameters an otherwise unknown route. However, this requires one Reversal of the cause-and-effect relationship within the model. Such an inverse model application containing the required interactions determined, which lead to a certain final state of the system, tracked an interaction chain is directed backwards in time and therefore leads usually on causality injuries. Additional measurement signals in one already complete certain chain of effects Therefore, do not be used to the lack of measurement signals to compensate in another chain of effects.

Model-based stress analysis according to 4

MBS tools and virtual prototyping

The model-based stress analysis as a building block of a lifetime observer takes place with a system model based on the multibody dynamics. Complex multi-body systems (MBS), which are also electrical, hydraulic or pneumatic elements can contain are modeled, simulated with the help of appropriate tools and analyzed.

This is done in three sequential steps:
Preprocessing: In an MBS model, the mechanical structure of the system to be modeled is mapped into the body and its connections, which contribute significantly to the dynamic behavior of the system. Forced conditions at the connection points and forces between the bodies define the mobility of the individual elements. The description of the mechanical system is supplemented by parameters describing the properties of the bodies, the coupling elements and the spatial position of the coupling points, by depositing three-dimensional geometries for the graphical representation and by defining all input functions used.

Calculation: The used MBS tool supports not only the model generation, but also generates a System behavior descriptive set of differential equations, the under selection of different analysis options by means of optimized solved numerical method can be. These include Method for time integration of the system, for finding a static equilibrium, for frequency analysis, for investigation of parameter dependencies and codes generated for export.

Postprocessing: In the analysis step can the numerical results are displayed (2D plots, 3D animation ...), analyzed (data filter, extreme values, FFT ...) or from the tool be exported.

The use of tools for simulation mechanical systems is significantly expanded by integration this technique into a consistent, model-based Development process in which the various areas of system development are integrated.

Application-specific toolboxes (for example from the application areas of automotive, rail vehicles and robotics) can be re-used by predefined modules and submodels and drastically accelerate the construction of a model. By supporting interfaces between development tools from different subfields, developers can, by building and applying consistent and consistent models as virtual prototypes, perform the various development steps in an optimal environment. Depending on the requirements, the interfaces range from a simple or bidirectional data transfer based on files to an interactive coupling of two independent simulation tools (co-simulation):
By means of appropriate interfaces, MBS tools can exchange data with a CAD system bi-directionally or even be integrated directly into an existing CAD development environment, in order to transfer design and construction features (kinematic models, component geometries, mass parameters) of a component and directly based on it to evaluate a multi-body dynamic simulation.

Of particular importance for the model-based Lifetime observer is the possibility the model of a component from a finite element analysis by means of an interface in the modeling of mechanical behavior to integrate the entire system. This information about the structural elasticity of components from a FE analysis transferred into the MBS model, so that the nonlinear dynamics an arbitrarily shaped body can be simulated in a system even if for large deflections structural elasticities are of importance. Accordingly, determined reaction forces as Loads and boundary conditions used in the finite element analysis become.

In the sense of a continuous virtual Prototyping can mechanical and control engineering (CACE) systems in a common Environment can be integrated and simulated. This can be different Wise ways, so by replacing linear system matrices (A, B, C, D), by exporting the equations of motion or symbolic Code of the mechanical system model in the tool environment of the Control technology or by a co-simulation of the mechanical and control subsystems with interactive exchange of data.

By generating code and by support Special input and output formats can be the results of a virtual Prototyping or a mechanical simulation in another Environment, such as to support hardware and software test environments or aspects of durability already in the design of a product account to be able to.

4 shows an overview of the tools required for the module of the online life cycle observer model-based stress assessment. The core of the model-based stress analysis is a mechanical multi-body dynamics model of the considered system. The geometric representations and mass parameters of individual bodies, as well as the marker points defined on the bodies or a given 3D kinematics can be extracted directly from CAD databases. The component for which A lifetime observation is to be performed must be present as an elastic body in the system model. The definition of an elastic body is extracted by means of an interface tool from a finite element model of the respective component. In turn, the reaction forces determined in a simulation can serve as load conditions for a detailed structural analysis in the FE application, from which the time profiles of local stresses or strains then result as component-specific stresses.

Integration of measurement data and simulation models based on 5 :

The overall model for the model-based determination of the component stresses must be stimulated at its interfaces to the environment by appropriate model excitations. In 5 is compared with the purely model-based approach of a metrological approach across the various hierarchical levels of the entire vehicle. 5 is intended to illustrate the various ways in which a model-based simulation can be related to the metrological information obtained at various system levels.

• – dem Gesamtfahrzeug, das im realen Umfeld angeregt wird (Straßen- und Trassierungsverhältnisse, Fahrerinteraktion) und, entsprechend der verfügbaren Fahrzeugsensorik, entweder die Anregung des Gesamtfahrzeugs selbst oder Systembelastungen in Form von fahrzeug-internen Kräften und Momenten detektiert,
• – einem Fahrzeug im Prüfstand, an dem entweder Systembelastungen in Form von fahrzeug-internen Kräften und Momenten oder auch, durch zusätzliche Sensorik, Bauteilbeanspruchungen direkt sensiert werden können,
• – dem betrachteten Bauteil in einer Prüfumgebung, für das die entsprechenden Bauteilbeanspruchungen direkt sensiert und zudem die ertagbaren Beanspruchen (in Form von Wöhler- oder Lebensdauerlinien) bestimmt werden können,
• – den Materialeigenschaften des betrachteten Bauteils (z.B. in Form von synthetischen Wöhlerlinien eines Probestabs).

The system levels shown correspond to:

• The whole vehicle, which is excited in the real environment (road and track conditions, driver interaction) and, according to the available vehicle sensors, either the excitation of the entire vehicle itself or system loads detected in the form of vehicle-internal forces and moments,
• - A vehicle in the test bench, where either system loads in the form of internal forces and moments of the vehicle or else, by additional sensors, component stresses can be sensed directly,
• - the component under consideration in a test environment for which the corresponding component stresses are directly sensed and, in addition, the tolerable loads (in the form of Wöhler or life-time curves) can be determined,
• - the material properties of the component under consideration (eg in the form of synthetic Wöhler lines of a test piece).

The one for online lifecycle observation provided information flow is marked with arrows. The system model to determine local stresses must be determined by system loads, measured during operation on the real vehicle, stimulated. The evaluation of the determined stresses in a damage calculation requires additional Component-specific information about the tolerable for the component under consideration Strains, usually in the form of a parameterized SN curve available. The other illustrated possibilities of an information flow are natural for the Prototype implementation and verification of an online life observer also of great Meaning.

Model - based load determination and possible submodels for a lifetime observation based on the 6 :

At the time of the draft of a Lifetime observation module for a selected component can be detailed information about available in the vehicle Sensors and the quality their signals are not available his. In particular, as a rule, no measured data will be available, be it from production vehicles in the daily Operation or even of measuring vehicles on special test tracks.

On the other hand, however, must also be considered be drawn that vehicles in the future with a variety intelligent mechatronic components, in which technologies of mechanics, electrics / electronics, control and information technology are integrated to the reliability, Availability, Improve maintainability and safety of the vehicle during operation. This includes a drastic extension of the sensors for monitoring the vehicle conditions.

In order to be able to flexibly consider the range of state information available in a vehicle, a model-based load determination is therefore proposed by way of example for a module for lifetime observation, as described in US Pat 6 outlined.

The direct approach would be one Determination of loads required to operate the lifetime observation model are necessary, from sensor data measured on the component. It is because of that only possible, if at the time of the design of the lifespan observer for the considered Vehicle availability the corresponding sensor is ensured and measured data available.

A load determination on the Base of a model for the entire vehicle, it allows with the help of a simulation calculation for the Vehicle a variety of different sensor signals and thus variable Sets of Generate state and load information. This allows for the selected component different complex models for model-based stress analysis to be examined.

• – Verwendung typischer und repräsentativer Anregungen für das Gesamtfahrzeugmodell, die den erforderlichen Amplituden- und Frequenzbereich abdecken und die realen betrieblichen Verhältnisse widerspiegeln.

If using the generated from a total vehicle model loads below an Ermitt The following conditions must be met in order to be able to perform stress tests for the lifetime observation of a component:

• - Use of typical and representative suggestions for the entire vehicle model, covering the required amplitude and frequency range and reflecting the real operating conditions.

• – Verwendung eines verifizierten Gesamtfahrzeugmodells. Dazu muss gezeigt werden, dass die mit Hilfe der Modellsimulation ermittelten Belastungen, über die Bandbreite der verwendeten Anregungen hinweg, gemessene Sensorsignale reproduzieren können.

If, as suggestions for the vehicle model, typical route designs and roadway descriptions are used, this requirement can be implemented, but if, for example, the course of typical wheel hub forces is required as vehicle excitation in order to avoid detailed modeling of the tire-road contact and the tire dynamics, one is dependent on type-specific measurement data because these forces also involve the reactions of the vehicle to the stimulus.

• - Use of a verified complete vehicle model. It must be shown that the loads determined with the help of the model simulation can reproduce measured sensor signals over the bandwidth of the applied excitations.

As a rule, available complete vehicle models for specific Applications designed and only for these special applications Verified. The use of such models in the context of lifetime observation then requires extensive verification under the typical ones chosen for this purpose Stimulus patterns that cover the full breadth of a business use cover.

If these conditions are met, leave the effects of different sets of available sensor information for the Flexible inspection of the lifecycle observer module. In extreme cases, for example. the excitation of the entire vehicle model of the sensed load information for the directly correspond to the component to be observed. In the other extreme case The complete vehicle model directly delivers the demands of the to be observed component.

Be with the help of a complete vehicle model both the loads and the loads are model-based generated, the lifetime observer model of a component is more integral Part of the overall system description model in the sense of elastic multi-body dynamics. The loads to be determined for the lifetime observation of a component then provide model-internal interfaces between an 'environment model' and the actual one System model dar.

Simulation model for a complete vehicle according to 7 :

boundary conditions

As part of a model-based online Lifetime observer should not and can not a specific vehicle model be developed and built. Rather, the CAD models available Models from the design and development phases of an existing one Vehicle are used.

The available CAD model of the vehicle is complete as a complete vehicle model and contains both Body, the e lastish body were extracted from NASTRAN models (FE modeling), as well as bodies whose geometric representations taken from CAD databases were. Its analysis and application therefore requires particular Tool-specific knowledge about the tool SIMPACK for MBS modeling a mechanical system and its interfaces to FE and CAD software packages.

The CAD model, however, contains neither an explicit model for the vehicle tires, still are specific in the database available Vehicle suggestions defined or present, whether in the form of Definition of a route with appropriate routing and vehicle connection or in the form of other time-dependent ones Excitation functions.

The following is the overall model the vehicle in its rough structure are presented to the possibilities an application of this model in the development of a model-based Lifetime observer for show the Hinterachsfahrschemel the vehicle.

7 shows a representation of a vehicle model, obtained as a simplified screenshot from the corresponding SIMPACK model database. A simpler overview of the model of the vehicle gives 8th , in which the coarse structure of the individual components, that is, the bodies and substructures used, and their mutual couplings are shown. In 8th The designation of substructures, ie modular submodels, which in turn consist of several bodies and their couplings, has been added to a star.

The model presented in these figures was originally developed and used to investigate the vertical dynamics of the vehicle in detail based on simulation calculations. The central body of the model is the vehicle body, which is directly connected to the inertial system (referred to in SIMPACK with Isys) and, moreover, to the front and rear axles of the vehicle, the substructure tures motor and integral carrier and the easily mapped (6 possible) Fahrzeuginsasssen is linked. Both the front and the rear axle consist of two substructures (VA_LI, VA_RE and HA_LI, HA_RE), which each include a wheel carrier, a wheel, a handlebar and a damper, and two further structural elements (stabilizer and handlebar on the front axle , Stabilizer and Hinterachsfahrschemel on the rear axle) are interconnected. The wheels of the vehicle are due to the action of a spring-damper force (with appropriate preload) on four cylinders, which in turn are fixed in the inertial system (s. 7 and 8th ).

The model contains a total of 3 elastic bodies, the Hinterachsfahrschemel (body HAFS_el in the substructure HRFS) and the rear axle link (body steering wheel_el in the substructures HA_LI and HA_RE). The part models for this body were using appropriate FE modeling (in NASTRAN) with the help The FEMBS interface is extracted and accessed via file interfaces in imported the vehicle model. Accordingly, 5 geometric Part representations for the Vehicle body (Primitive FZG_Haut for the body FZG), the Hinterachsfahrschemel (Primitive HAFS_el_Cubid for the body HAFS_el), the integral carrier (Primitive Integraltraeger_geo for the body Integral carrier) and the handlebars of the rear axle (Primitive handlebar_el_CAD the body handlebar_el, each for the right and left rear axle) via an interface from appropriate CAD data generated.

In the 8th shown compounds (together with the connections in the substructures not shown) limit the number of degrees of freedom of the individual bodies, so that the entire model comprises a total of 79 degrees of freedom: 62 dynamic degrees of freedom for the different rigid bodies, 9 elastic degrees of freedom for the elastic bodies and 8 Degrees of freedom that describe internal states for the state-dependent computation of dynamic force elements. The MBS tool SIMPACK generates a set of coupled equations of motion for these degrees of freedom. In a numerical integration calculation, the time behavior of the associated state vector can be simulated under a given model excitation. In postprocessing, the time behavior of the state vector is transformed into physically interpretable system variables and model-defined sensor variables, which can be analyzed numerically, graphically and in a 3D animation of the model. In addition, with the help of the LOADS interface, time curves for generalized reaction forces and moments on the markers taken from the corresponding FE model can be exported to a formatted output file for each of the defined elastic bodies in order to reuse them directly in the FE modeling.

Suggestion of the Model over arbitrary time courses of wheel forces

Since there were no suggestions in the original complete vehicle model, 4 new forces that stimulated the model were defined (F_ANREG_HL, F_ANREG_HR, F_ANREG_VL, F_ANREG_VR). They attack from the inertial system at the centers of the individual wheels, so act as a force excitation on the hubs. They are defined as type 93 of the SIMPACK force elements (Force / Torque by u (t)) by excitation functions u _i (t) with the assignment from Table 1:

Table 1 Assignment of the functions u _i (t) to the wheel hub forces

Via time excitation functions (T_Anreg_HL_inx, ..., T_Anreg_HL_iny, ..., T_Anreg_HL_inz, ...) of the SIMPACK type 02 (Input-Fct), the excitation functions u _i (t) are again assigned function values that result from reading in a sequence of value pairs (tk, u _i (tk)). In this case, function values for points in time that are not explicitly defined in the read-in value pairs can be interpolated in different ways from the temporally adjacent values. In addition, additional scaling and time delays tion and a temporal adaptation range for the read excitation functions u _i (t) are defined.

The input function with the sequence of value pairs (t _k , u _i (t _k )) itself must be provided in a suitable form at a specific point in the model database. In simple cases, this input function can be generated manually (eg a staircase function). In addition, a simple MATLAB routine has been created that can generate more complex input functions in the required form (eg by superimposing sine functions of given frequencies and amplitudes with random phases at selected interpolation points of time).

This procedure makes it possible practically without restrictions, by copying corresponding input functions simulation runs under perform different excitation patterns, without each in the model even fundamental changes to have to make.

Various exemplary simulation runs for the model under the influence of time-dependent hub forces have been performed, their results should not be presented in detail here.

modeling the rear axle

9 shows enlarged and in different perspectives a 3D representation of the rear axle of the SIMPACK model.

The rear axle stool (HAFS) is as an elastic body modeled using the FEMBS interface directly from a FE modeling was extracted in NASTRAN. There are 5 eigenmodes taken with frequencies between 100 Hz and 300 Hz, so that respectively 5 initial values for the preload forces, the elastic coordinates and associated speeds for the reduced Partial model of the HAFS in the MBS system to be defined. In addition, 22 relevant, distributed over the entire geometry Node of the FE model as marker for the corresponding HAFS body in the MBS system taken over.

Accordingly, the submodel is the Handlebar for each of the right and left rear axles is modeled as an elastic body, extracted with FEMBS from a FE modeling in NRSTRAN. There were 2 eigenmodes with frequencies between 150 Hz and 350 Hz taken over. In addition, each 20 relevant, distributed over the entire geometry Node of the FE model as a marker for the corresponding handlebar body in the MBS system taken over.

Clearly visible in 9 nor the stabilizer, which is similar to the HAFS, the two rear axle substructures linked, as well as the two spring-damper systems between the HAFS and the handlebars.

10 shows one of the 8th Comparable structural representation in which only the rear axle of the model is shown, but now the participating substructures are resolved into their individual bodies, compounds are again listed as black lines, forces as red arrows.

The interface of the rear axle their environment is the result of the connection of the HAFS on the vehicle body, on the other hand, the wheels via a spring-damper force to the cylinder punch and over the stimulating wheel hub forces bound to the inertial system.

part models for one Life observation

In order to be able to flexibly take into account the range of state information available in the vehicle, a model-based load determination is necessary for the prototypical implementation of the HAFS lifetime monitoring module, as described in 6 was sketched. With the aid of the load-time profiles generated from a complete vehicle simulation, time profiles of the stresses for the lifetime observation of the rear-axle traveling platform HAFS are to be determined, also model-based. Thus, the effects of different sets of available sensor information for the lifetime observer module can be flexibly examined.

Since both the loads and the stresses are generated model-based, the HAFS lifetime observation model is an integral part of the overall vehicle model of elastic multi-body dynamics used. The loads to be determined for the lifetime observation of the HAFS then represent model-internal interfaces between an 'environment model' and the actual system model. Looking at the vehicle, various realistic possibilities for the available load information are conceivable. This corresponds to various interfaces between the overall vehicle model for generating the load information (sensor signals) and the model for lifetime observation of the HAFS contained therein. In 10 are marked as dash-dot areas two different sub-models, each corresponding to a different set sensed loads.

damage calculation

The evaluation of the model-based determined component-specific time profiles of the stresses for a statement about the accumulated component damage or a prediction of the remaining life of the component is done in the classical manner with the aid of a damage calculation on the basis of the fatigue analysis ( 11 ).

The objectives, fundamentals and Procedures of a fatigue analysis were used in the scientific Literature presented several times and shall only be summarized here to be viewed as.

A calculation of the component damage is based to be performed sequentially on 3 Partial steps:

Accumulation of the stresses in stress collective

An operating life or service life calculation of a component subject to vibration must be based on a material behavior meaningful Description of all vibration stresses occurring during operation based. Usually receives these description forms by analysis and preparation of the determined time courses the stresses. You will be classified in single swing games and then compressed in the form of stress collectives. there targeted information is eliminated for a life cycle calculation not needed anymore become.

From the passage of time of a stress will be first extracting the sequence of upper and lower reversal points, where on information about the shape of the vibration amplitude and the frequency the vibrations is dispensed with. With an amplitude filter as data filter are also undesirable small or big Amplitudes eliminated (Omission, Truncation). By waiving information about the chronological order (sequence) of the reversal points is through statistical classifying and counting procedures achieved a drastic reduction of the data volume. One for the life cycle calculation appropriate description of a relevant Contains load sequence only the frequency of a given amplitude of the amplitude and mean So a two-parametric stress collective. Every turning point of the stress process is accumulated in a stress collective and increased the frequency of one Oscillation game from a predefined grid of possible amplitudes and mean values. The resolution the grid determines the effort required to store the collective, which therefore does not depend more on the accumulation time. there are different statistical counting methods common (One-parametric such as the class passage and the span pair method, Two-Parametric like the From Bis Count or the Rainflow Count.

The accumulation of stress time courses in Collective delivers via a defined evaluation algorithm a compressed description the stress process in the form of multi-parametric frequency distributions for the Amplitudes and mean values of the oscillatory cycles occurring in the stresses.

parameterization of the sustainable demands

A Wöhler test determines the up to the oscillation break or the Schwinganriß sustainable number of oscillatory cycles N for an oscillation stress of any amplitude shape, but given amplitude s _a and fixed average s _m . The Wöhler line results from a statistical evaluation of all vibration frequencies N measured at oscillation fracture (oscillation crack) as a function of the parameters s _a and s _m (s. 12 ) and separates the different strength ranges (short-time, time and fatigue strength) from each other.

For a given probability of failure P _Ü and amplitude mean value s _m , the range of the time stability is limited by a curve N (s _A ) whose shape is approximated by N = N A (s a / s A ) -k given is. This corresponds in a double-logarithmic representation of a straight line with the gradient -k. The point (N _A , s _A ) marks the lower limit of the range of validity of the straight approximation (s. 12 ). In the area of fatigue strength, the life of a component, which is limited by vibration fracture, is therefore determined by the parameters N _A , S _A and k of the Wöhler line. In addition to the dependence on the mean stress s _m, they depend largely on the type of stress (axial, bending, torsional stress), the tensile strength and surface treatment of the material, the ambient conditions (temperature and corrosion). The dependence on the component size and geometry is particularly complex and requires special consideration.

rating the stress collective

One of the methods for calculating the life of a component under a variable amplitude vibration stress is referred to as a Palmgren and Miner linear damage accumulation hypothesis, or Miner's Rule for short. It links the description of the acting stress (in the form of the stress collective) with the sustainable stress (in the form of the component Wöhler line) (see 12 ) to a statement about the total accumulated damage or the remaining life remaining. The basic idea of the linear damage accumulation is that each vibration of the stress in the material of the component causes a partial damage, which is independent of the damage already caused by previous oscillatory games.

In the time-stability range of the Wöhler line, the result for each load cycle of the amplitude s _{a, i is} a partial damage of D _i = 1 / N _i , where N _i = N (s _{a, i} ) the sustainable number of oscillations from the Wöhler line for the amplitude s _{a, i} is. The summation over an n-pitched stress collective with h _i cycles in the raster i yields:

The damage D is the sum of the partial damage of each individual applied load cycle from the collective, while the life N indicates how many load cycles from the above collective can be tolerated until the oscillation breakage (or crack). As a rule, large amplitudes s _{a, i are} so rare (h _i is so small) that, despite the associated large partial damage (1 / N _i ), they only provide a small contribution h _i / N _i to the total damage. Small amplitudes are much more common, but can only contribute to the damage sum to a limited extent, since the sustainable number of cycles N _i according to the Wöhler line becomes extremely large (in general, the sustainable number of oscillations falls with the 6th to 8th power of the amplitude s _{a, i} from). In the area of fatigue strength, the very small amplitudes no longer contribute to the damage at all.

For components of simple geometry, such as test bars and notched bars, there are a variety of measurements of Wöhler lines (N _A , s _A , k) for different parameter sets (type of stress, P _ü , s _m ...). In addition, empirical methods exist to calculate Wöhler lines from material properties, to transform available Wöhler lines to a given parameter set or to adapt them to a given geometry by means of shape and notch effect numbers.

Specific tools for durability and damage calculation support Therefore, not only the determination of the amount of damage with a damage hypothesis, but by accessing material databases and algorithms for consideration the component shape and size the calculation of sustainable burdens.

For the computational lifetime estimation of a component of complex geometry and structure, the Wöhler line valid for this component and its specific load (parameter type of stress, P _ü , s _m ...) is required. As a rule, this component selector line must be derived from Wöhler lines for small samples, taking into account size and surface influences of geometric, statistical, technological and surface-technical nature.

This should 13 illustrate. The vertical axis shows there the increasing complexity of the influences, ranging from single-stage load sequences (Wöhler tests) via standardized load sequences to the real operational load sequences. On the basis of a hypothesis for damage accumulation it can be concluded mathematically from the behavior with simple load sequences on the behavior with complex load sequences (vertical lines in 13 ).

The horizontal axis represents the complexity of the test object (smooth sample rod, notch bar or molded element up to the real component). By taking into account size and surface influences, it is possible to conclude empirically on the service life of a complex component, with an identical load sequence (horizontal lines in 13 ). The aim is to determine the tolerable stress for the real component under an operational load sequence. The availability of measured data determines the number of empirical or computational steps that are necessary to reach this goal and thus also the significance of the calculated accumulated component damage or the predicted remaining service life.

toolchain

The process for model-based lifetime observation presented in the previous chapters ei A selected component is preferably to be implemented on the basis of a suitable tool environment. 14 summarizes the tools needed for modeling, model simulation and damage calculation and their interfaces with each other.

Basic tool is a tool for Generation and simulation of mechanical multi-body models. With your help can Created, simulated, tested and validated system models as 'virtual prototypes' be that realistic to reproduce the behavior of a physical system. The dynamics, internal forces and the vibrational behavior of the virtual prototype can be detailed be analyzed and optimized. The most common tools (SIMPACK and ADAMS) have interfaces to other CAD / CAM / CAE tools, in the virtual prototype partial models, model geometries, component elasticities, control engineering To incorporate elements and metrological stimuli in other, developed tool environments adapted to the particular problem were.

For the design of vehicle components become FE models of a component used for (linear or nonlinear) structural analysis. through FE program systems, such as ABAQUS, ANSYS or NASTRAN, become equivalent Models generated, for example, in static strength analyzes the Deformation and mechanical stress, in modal analyzes the natural frequencies and forms and in dynamic analyzes the vibration behavior to examine and evaluate the component.

Programs for fatigue analysis like FALANCS, FEMFAT or MSC / FATIGUE can be used to communicate with measured or model-based generated descriptions of local Stresses and cyclical material parameters local individual damage of a considered component to determine and a remaining life remaining to forecast.

The integration of behavioral descriptions a mechatronic system by analyzing different aspects in a tool environment adapted to the respective problem were created, to a virtual prototype allows it, already at an early age At the time of system development, the fulfillment of the requirements to verify, analyze and demonstrate system behavior to document. economical and in a short time you can examined on the basis of virtual prototype design variants and be optimized. However, this is only possible if it succeeds various tools used by suitable interfaces to integrate into a closed tool environment.

In the project 'online lifecycle observers' will not be project-specific Models are developed, rather, on already available models resorted so that within the project no free choice of which to use Modeling tools exists more. As in Chap. 0 described, the overall vehicle model used is available as a SIMPACK model, in that both elastic body, whose description was extracted from NASTRAN models as well geometric representation from CAD databases are integrated.

FEMBS / LOADS interface from SIMPACK to FE tools

FEMBS is an interface program between program systems for finite element analysis and for multibody dynamics simulation, the input data flexible body for the MBS system selected and generated. This link is not just done in the form of a pure data transfer, but requires the calculation additional physical quantities for the nonlinear MBS model. While FE programs the movement of complex flexible structures through a size The number of nodal coordinates is described in MBS programs Dynamics of elastic bodies by a rigid body movement modeled, superimposed by small elastic deformations becomes. These elastic deformations are considered a superposition of a few, usually FE modes pre-calculated, eigenmodes described.

The used eigenmodes are taken from an FE modal analysis of the free structure. As homogeneous solutions only describe the vibrational behavior of the elastic body then correctly, if the acting loads no local deformations effect at the force introduction points. The inclusion of static Loads at the force introduction points into the eigenmodes (, static modes') in the superposition of elastic deformations very many partly high-frequency eigenmodes. In FEMBS, however, are based on the chosen one free natural oscillations so-called frequency response modes used. With these eigenmodes, the harmonic forces at the force introduction points consider the elastic structure, can the elastic body very few eigenmodes very effectively connected to the surrounding MBS structure become.

The coefficients of the equations of motion for the elastic body arise from time-independent Volume integrals, which are calculated once in the interface and in the form of development coefficients (in deformation) up to 2nd order stored as Standard Input Data (SID) in a file become.

This results in the transfer of information for a reduced model of an elastic body from an FE modeling of the body to the MBS model as seen on the left side of FIG 15 shown. In this case, the deductive derivation of the measured loads on construction in the form of an interactive loop Parts of the multi-body system with the inductively calculated loads from the multi-body system on the components and resulting loads by selecting the relevant components and by selecting the relevant eigenmodes of the finite element model brought to coincide.

FEMBS generates in interaction with the user in 5 sequential steps (see Table 2) a reduced Model of the selected elastic body in the form of a SID file whose format is also described in. When building the model, it is interpeted by the SIMPACK preprocessor and defines an elastic body in the MBS model.

Tabelle 2 : Schrittweises Vorgehen in der FEMBS-Anwendung

Table 2: Step-by-step procedure in the FEMBS application

The LOADS (or FEMBS ^-1 ) interface is part of the SIMPACK postprocessing and allows for any MBS model elastic body imported with FEMBS after performing a simulation of the overall model the dynamic reaction forces (loads) acting on the elastic bodies to calculate. Determined for the body-fixed coordinate system, the time courses of the reaction forces are stored in a formatted (SOD) file, which is directly interpretable by the FE software used. After interactively selecting an elastic body from the MBS model and a selection of simulation time steps, LOADS then calculates and stores reaction forces for those nodes of the FE model that are selected when importing a reduced model with the FEMBS interface and stored in MBS. Model of the body were described by a marker (see right side of 15 ).

Tools for damage calculation

• – Aufbereitung und Komprimierung lokaler Beanspruchungen, denen das Bauteil an seinen kritischen Querschnitten ausgesetzt ist;
• – Bereitstellung zyklischer Materialparameter unter Berücksichtigung der Bauteilgestalt;
• – Schadensakkumulation unter Verwendung eines entsprechenden physikalischen oder empirischen Ansatzes.

The damage calculation for a selected component takes place in 3 steps:

• - preparation and compression of local stresses to which the component is subjected at its critical cross sections;
• - Provision of cyclic material parameters taking into account the component shape;
• Accumulation of damage using a corresponding physical or empirical approach.

16 gives an overview of the input information and functionality required for a tool-based damage calculation. The various commercial and practical injury calculation tools differ considerably in their objectives, performance and scope.

• – Unterstützung und Verarbeitung verschiedener Darstellungsformen für die eingehende Beanspruchungsinformation (Zeitverläufe, Kollektive, Rainflow-Matrizen);
• – Verfügbarkeit verschiedener Varianten der Minerregel zur Schadensakkumulation nach dem Nennspannungskonzept (spannungs-basierte Schädigungsberechnung);
• – Verfügbarkeit verschiedener Teilmodule zur Schadensakkumulation nach dem örtlichen Konzept (dehnungs-basierte Schädigungsberechnung, Rissfortschrittsberechnung);
• – Verfügbarkeit von Datenbanken für die Bauteilwöhlerlinie und zyklische Materialparameter;
• – Berücksichtigung mehrachsiger Beanspruchungen und verschiedener Umgebungsbedingungen;
• – Anbindung an FE- und MBS-Modellierungswerkzeuge und an Software zur Prüfstandssteuerung;
• – Möglichkeiten zur grafischen oder dateibasierten Darstellung und Dokumentation der Ergebnisse.

The essential distinguishing features are:

• - support and processing of various forms of presentation for the incoming claim information (time courses, collectives, Rainflow matrices);
• - Availability of different variants of the mineral rule for damage accumulation according to the nominal stress concept (stress-based damage calculation);
• - Availability of different sub-modules for damage accumulation according to the local concept (strain-based damage calculation, crack propagation calculation);
• - Availability of databases for the component selector line and cyclic material parameters;
• - consideration of multi-axis loads and different environmental conditions;
• - Connection to FE and MBS modeling tools and software for test bench control;
• - Possibilities for graphical or file-based presentation and documentation of the results.

FEMFAT

FEMFAT is a software package for damage analysis, based on the results of a FE analysis. interfaces to MBS software packages and to the test bench control (RPC format) are also available. The damage calculation in FEMFAT based on the determination of S / N curves (Wöhler lines) at the individual nodes of the FE model, that is the sustainable number of oscillatory games as a function of the applied voltage amplitudes, which are determined by the Stresses and the respective local behavior of the component depend. A variety of expansion modules allows the use of FEMFAT for specific use cases (multiaxial stresses, welding points ...).

• – Import von Informationen über das FE-Modell des Bauteils; Geometriedaten des Bauteils (Knoten und Elemente) Übernahme im FE-Modell definierter Gruppen
• – Import von knoten-bezogener Spannungsinformation aus dem FE-Modell;
• – Generieren der benötigten Materialdaten; Einlesen aus FEMFAT-Datenbank oder Generierung aus Grundparametern
• – Import oder Generieren einer Beanspruchungsbeschreibung; Rainflow-Spektren
• – Schädigungsberechnung an den Knoten der aktuellen Gruppe; Berücksichtigung relevanter Einflüsse (Oberflächenrauheit, -Behandlung) Auswahl der Analyse und Einflußfaktoren (Spannungsgradienten, Oberfläche, Spots, Welding...) Selektion der Ausgabedaten (grafische Darstellung, Datei-Ausgabe für FE-Postprocessing)

A lifetime analysis with the FEMFAT tool involves the following steps:

• - import of information about the FE model of the component; Geometry data of the component (nodes and elements) Acceptance in the FE model of defined groups
• - import of node-related stress information from the FE model;
• - generating the required material data; Import from FEMFAT database or generation from basic parameters
• - import or generate a claim description; Rainflow spectra
• - Damage calculation to the nodes of the current group; Consideration of relevant influences (surface roughness, treatment) Selection of the analysis and influencing factors (stress gradients, surface, spots, welding ...) Selection of the output data (graphical representation, file output for FE postprocessing)

WAFO toolbox for MATLAB

WAFO (Wave Analysis for Fatigue and Oceanography) is a toolbox for Matlab with individual routines for statistical analysis and simulation of random waves and burdens that are significant at the institute for mathematical Statistics of Lund University in Sweden was developed. The Individual routines (m-files) of the toolbox, depending on the application in different Subdirectories include, inter alia, the analysis measurement data by determining parameters in statistical distributions, a variety of oceanographic applications, but also routines for one statistical damage analysis, general tools for statistical evaluations and graphic representations.

Some of the modules (SPEC, TRGAUSS, WAVEMODELS and MULTIDIM) were used for specific oceanographic Applications developed, other modules such as CYCLES, MARKOV and DAMAGE contain the essential applications for a damage calculation. These include u.a. tested and efficient algorithms for inverse point extraction and rainflow filtering, for the determination of excess frequencies, Minimum-maximum cycles and rainflow cycles, for the conversion of cycles in matrices and histograms, for extrapolation and smoothing of Rainflow matrices, for generating rainflow matrices from Markov models, for damage calculation from cycles and from matrices, as well as for the simulation of random loads. Table 3 shows a summary overview of the application areas of the for the damage calculation relevant submodules from the WAFO toolbox.

Tabelle 3 Für die Schädigungsrechnung relevante WAFO-Teilmodule

Table 3 WAFO submodules relevant for the injury calculation

Implementation concepts for Online life observation

Separation of recording and evaluation channel

Regardless of whether the investigation the operational demands of a selected component directly measured can be or model-based, can be a vehicle module for online lifetime observation of the component in two separate Separate subsystems, each with fundamentally different requirements subject.

A subsystem must be the interface represent to the vehicle sensors in which in high-frequency repetition rates current state information about the vehicle is present. Physically, this interface is in usually as a connection of the life observer to a vehicle data bus be realized. Under real-time requirements, the original sensor information must be processed and evaluated, both in the form of a with 'intelligent sensor' integrated units of sensor element, signal processing and A-D conversion or by transmission digital sensor data via the bus with subsequent Data preparation and integration in a separate module can. In any case, this unit is subject to the recording channel of the total lifetime observation module, the real-time requirements the vehicle sensor.

The second subsystem is used for processing the output data of the recording channel for a lifetime prediction of the selected Component. The damage calculation is based on the evaluation of component stresses that accumulate over extended periods of time are therefore not subject to the real-time requirements of the Recording channel. The calculation of a lifetime forecast can thereby cyclically (hourly, Every day ...) in predefined 'maintenance intervals' or event-driven (after completion of a Drive, when a predefined or critical operating condition occurs) respectively. In any case, the calculation cycles are in this value channel significantly larger than the cycles with which provided new data in the vehicle sensors become.

• – notwendige Rechenleistung im Aufzeichnungskanal (unter Realzeitanforderungen);
• – notwendiger Speicherbedarf in der Schnittstelle Aufzeichnungs- und Auswertekanal;
• – notwendiger (Kommunikations-)Aufwand zur Datenübertragung über den gesamten Prozess;
• – zulässiger Informationsverlust in der Datenübertragung vom Aufzeichnungs- zum Auswertekanal.

A data memory separates the recording channel, in which high-frequency data are processed, and the evaluation channel, in which input data accumulated over longer periods are used. In the realization of a module for online lifetime observation, there are various possibilities to assign the basic functions of a model-based lifetime observation to the recording or evaluation channel. However, every possible variant involves different requirements for the physical realization of the module for online lifetime observation. Overall, this can be a field of tension from the following influencing factors:

• - necessary computing power in the recording channel (under real-time requirements);
• - necessary storage space in the interface recording and evaluation channel;
• - necessary (communication) effort for data transmission over the entire process;
• - permissible loss of information in the data transmission from the recording channel to the evaluation channel.

17 is a graphic representation of this field of tension, in which various distribution variants for a module for model-based lifetime observation are entered. Due to the different requirements for the physical realization, the illustrated variants correspond to different implementation concepts for the online lifetime observation.

The following procedures are proposed for the implementation:
The various approaches are summarized in Table 4 and are grossly evaluated for the influential factors. The essential feature of a procedure is whether the model-based determination of the claim takes place in the recording channel (stress-based procedures) or in the evaluation channel (load-based procedures). The compression of the information before the transfer from the recording to the evaluation channel can be carried out directly (without information loss), by Fourier transformation in power density spectra (loss of phase information) or by formation of collectives (loss of sequence and phase information). Here are two extreme cases to briefly summarize the main approaches are treated: Table 4: Procedures for online lifecycle monitoring and their evaluation

Direct procedure (stress-based with collective formation):

The signal preprocessing and the Model-based determination of the stresses are used in this procedure executed in the recording channel under real-time requirements. This requires there, especially if to determine the stresses Models with many degrees of freedom are required, very much Computing power. From the different variants for compression and accumulation of the determined stress time courses here the one used, the stress collective, as minimal for the damage calculation necessary information, about transfers the data memory to the evaluation channel. There you can accumulated collective load after expiry of a maintenance interval be evaluated.

The available information then suffices however, only out to a summary damage assessment for the maintenance interval perform. A detailed (eg model-based) analysis of a damage-relevant Single event is not with this information in the evaluation possible.

load compression (load-based with direct compression):

In this implementation concept, as complete Description of the system load, the timings of the preprocessed and integrated sensor signals by means of a suitable, effective Compression algorithm to be cached directly to so that the information channel of the record of that of Separate evaluation. This relieves the data logging of the workload for the model execution under real-time requirements, but requires, depending on the effectiveness of the data compression, sufficient space. In the evaluation channel is then, but after one Data decompression, the complete information about the time courses the individual system loads available. In the model-based Determining the stresses that are here without real-time requirements can be done in the evaluation channel, therefore Even complex and complex models are used.

On the basis of the full description the system loads can in the evaluation, both the system models used modified and adapted to the requirements, as well as various approaches to damage calculation can be used and compared.

method using standard usage cases

In all the procedures mentioned in the previous section will the injury of the selected component with the help of available information about calculated the current state of the vehicle. Prior knowledge of the Component damage, e.g. under standardized load sequences or under use the vehicle at repetitive operating conditions, will not used.

The goal of the procedure under Use of standard usage cases is it, damage relevant To use foreknowledge for a part of the operating time a simplified damage calculation perform, so the limited resources of computing power and storage space for those Operating hours available stand for the a prior knowledge not available is.

The damage-relevant prior knowledge may, for example, from information about the operational use and the associated one damage behavior of the selected Component come from comparable or identical vehicles. It can also be made from bench tests a corresponding vehicle prototype or during the Development phase of the vehicle with the help of simulation calculations for a suitable vehicle model has been generated.

Standard use case (use case) and pre-determined injury rate

The damage-relevant prior knowledge combines a defined vehicle usage with an associated damage rate for the considered component. In advance, the load / stress curves or collectives who use a vehicle in the standard use case occur in a model-based stress assessment and / or a component-specific damage calculation be integrated into a damage rate associated with this use case.

This results in the procedure 18 with two alternative evaluation methods.

As long as the signals from the vehicle sensor or the time courses the system loads for the vehicle will display a standard usage case (USE-CASE) the associated damage relevant Previous knowledge in the form of a for In this case of use, the damage rate used to generate a perform greatly simplified damage accumulation. It therefore requires as little a model-based stress analysis and component-specific Damage calculation, like a complicated storage of information for the separation of Recording and evaluation channel.

The resources of computing power and storage space is then almost complete for those operating hours to disposal, in which the system load for the vehicle does not display a standard use case. For this purpose, depending according to design, the various described in the previous section Implementation concepts (from the direct procedure to the compression loads) for exceptional (non-standard) use and especially damage-relevant To capture events in detail.

Identification of standard use cases (use cases)

A standard use case (use case) is as a specific sequence of external stimuli of a system defined, which occur at predetermined times. This can e.g. as a consequence of predefined road sections and routing conditions and described as a sequential sequence of actions of the driver his. The system responds to these suggestions, in which the observable sensor signals from an initial value in a defined Change the time.

Taking into account the dependence of the temporal evolution under the use case conditions on the initial conditions or initial values of the sensor signals and uncertainties in the prior knowledge, a standard use case describes the temporal evolution of a limited volume in a space spanned by the sensor signals external suggestions that define the use case (see 19 ).

In a standard use case, therefore, the chronological sequence of the individual values of the sensor signals as predefined behavior and their inaccuracy are predefined, so that the component-specific damage accumulated in this use case can be predetermined with the aid of the damage-relevant prior knowledge (s. 19 ).

As long as no external stimulus for the system occurs, which deviates from the procedure defined in the use case, runs the temporal development of the system along the defined in the use case Nominal behavior. Are the external suggestions not completely detectable, must checked every time whether each sensor signal is still within the defined in the use case Value range is located. If this is no longer the case, the system behavior becomes of the nominal behavior of the use case deviate and a predetermination of the injury rate impossible.

The approach by means of standard usage cases becomes then particularly effective when long-term operational periods of use occur in which very few and only limited to the system dynamics impacting external stimuli occur. The system will then transition to a quasi-steady state and in this state to a highly localized subarea limit in the space of the sensor signals.

Two exemplary examples of standard usage cases are intended mentioned here become

In standard use case, "standstill while the engine is running "remains the vehicle at a standstill, without external intervention, left to itself. The vehicle sensors will be the transition in a quasi-stationary state Show that practically independent from the prehistory becomes. This operating case makes a not insignificant Share of the total useful life of a vehicle, at the same time can, for example, in test bench tests, in the apron of the operational use a lot of knowledge about the damage of the considered component are generated.

In a standard use case, highway travel 'is based on the available Vehicle sensor system recognized a highway ride (for example, the measured Driving speeds over a longer one Period at a large mean speed and a small fluctuation range demonstrate). Again, in the run-up to operational use can be very a lot of knowledge about the damage of the considered component are generated. So can on the basis of for a highway ride typical route profiles and driver activities an averaged injury rate for the Component to be determined.

Adaptation of the Standard use cases

A significant advantage that the Approach under consideration predefined standard use cases with brings, is the possibility the definition of standard use cases to the existing basis damage-relevant To adapt to prior knowledge. The current injury rate in a standard usage case results from an averaging over the damage rates the values of the different types permitted in the case of use at that time Sensor signals.

If you have little previous knowledge, you will to specify only a few 'gross' use cases can, the relatively large one allowed Include areas in the space of the sensor signals. The over this Ranges averaged damage rate provides accordingly only a rough estimate for the expected total damage. Corresponding is for a big Part of the total usage time a detailed damage analysis in the event channel be performed on the basis of current values of the sensor signals have to.

If you gain an improved base from the operational use of comparable vehicles For damage-relevant prior knowledge for the component under consideration, refined standard usage cases can be specified. Since in each standard use case only a defined course of the damage rate has to be stored for the damage analysis, the module for lifetime observation of a component can easily be adapted to a more refined specification of the standard use cases.

The integration of the damage-relevant Prior knowledge in which by analyzing the sensor information time periods standardized Use with predefined damage behavior for the considered component, allowed over the total usage time a flexible mixture of a simplified injury determination (with corresponding uncertainty) and a detailed model-based damage calculation (with corresponding need for storage space and computing power).

So is with appropriate prior knowledge For example, conceivable, the use case, standstill with the engine running 'with the idle speed to parametrize and store speed-dependent damage rates. Corresponding could the standard use, highway driving ', if appropriate prior knowledge is available, dependent be refined from one parameter into multiple use cases (ranges the driving speed ...).

Since the standard use cases are from a defined sequence of external suggestions they are not component-specific. So the specification of use cases can be considered when considering another Component taken over it will have to be but a component-specific damage information deposited become.

Claims (18)

Model based life observer with following modules: - Detection of system loads from the existing system sensors, in particular a vehicle sensor system, and processing of the sensor signals from the System sensors to load time courses; - Determination of local component loads from the system model and the load time courses; - Calculation of the remaining life the components contained in the system model from the accumulated component damage through a fatigue analysis. Model-based lifetime observer according to claim 1, in which the system model is supplemented by an environmental model, where the environmental model missing sensors in the system sensor through Simulation completed by experience. Model-based lifetime observer according to claim 1 or 2, where the system model is a complex multi-body system (MBS) is. Model based life observer after one the claims 1 to 3, in which the multi-body system Data exchange with a CAD system and the individual components of the multibody system are modeled with a finite element tool. Model based life observer after one the claims 1 to 4, in which the load time courses from the loads of System bearings, in particular from the wheel hub forces are obtained. Model based life observer after one the claims 1 to 4, in which the calculation of component damage to three sequential to be performed Sub-steps is divided: - accumulation of stresses in stress collective; - Parameterization of the sustainable stress; - and Evaluation of the load collective. Model based life observer after one the claims 1 to 6, in which the tool chain of the following software modules consists: - one Many-body system, the above an interface CAD data with a finite element tool exchanges; - the Finite element tool over another interface exchanges data with a damage calculation module; - and the Multibody system over one additional interface exchanges data with the damage calculation module. Model based life observer after one the claims 1 to 7, in which interactively a reduced model for the individual Components of the entire system is generated. Model based life observer after one the claims 1 to 8, where the damage calculation for one selected Component in three steps: - Preparation and compression local stresses that the component at its critical Cross sections is exposed; - Provision of material parameters considering the component shape; - damage accumulation using a corresponding physical or empirical Approach. Model based life observer after one the claims 1 to 9, with two subsystems separated by a data memory, which are referred to as a recording channel and as an evaluation channel, in which the recording channel is the interface to the vehicle sensor system and working under real-time conditions and data in the data store records, and the evaluation channel recorded in the data memory Data, cyclic, event-driven or at predefined maintenance intervals evaluates. Model-based lifetime observer according to claim 10, in the signal processing, model-based load calculation and load collective formation in the recording channel, and only the damage calculation takes place in the evaluation channel. Model-based lifetime observer according to claim 10, in which the signal processing and data compression in the recording channel takes place and the data decompression, the model-based load calculation, the load collective formation and the damage calculation in the evaluation channel he follows. Model-based lifetime observer according to claim 10, in the signal processing and load collective formation in the recording channel and model-based load calculation and damage calculation in the evaluation channel respectively. Model-based lifetime observer according to claim 10, in which the load time profiles are divided by evaluation into standard use cases and if a load time lapse into a standard usage case is divided, for this standard use case, a specific damage rate for the component-specific Remaining life. Model-based lifetime observer according to claim 14, where the specific injury rate is empirical gained damage knowledge is. Model-based lifetime observer according to claim 14, in which damage relevant Prior knowledge about the components are included in the specific damage rate. Model-based lifetime observer according to claim 14, in which a standard use case in the motor vehicle idle of the engine detected when the vehicle is stationary. Model-based lifetime observer according to claim 14, in which a standard use case in the motor vehicle, the highway a motor vehicle detected.