EP1599842A2

EP1599842A2 - Risk minimisation and maintenance optimisation by determining damage proportions from operational data

Info

Publication number: EP1599842A2
Application number: EP04712485A
Authority: EP
Inventors: Herbert Depping; Matthias Maisch
Original assignee: Voith Turbo GmbH and Co KG
Current assignee: Voith Turbo GmbH and Co KG
Priority date: 2003-03-06
Filing date: 2004-02-19
Publication date: 2005-11-30
Also published as: WO2004078543A2; DE10310116A1; WO2004078543A3

Abstract

The invention relates to a method for determining maintenance intervals for motor vehicle modules which comprise at least one maintenance unit. Each maintenance unit is associated with an amount of damage, whose the value increases during the operation of the motor vehicle. The load acting upon a maintenance unit is determined or estimated from operational data of the vehicle or from one part of said data in a selected time interval and divided into different categories of a load pattern. The increase in total damage for said time interval is calculated by subsequently comparing the load pattern of a time interval with stored data of the breakdown behaviour of the maintenance unit. The end of the maintenance interval for a maintenance unit is then achieved when the associated total damage exceeds a specific critical threshold value.

Description

Risk minimization and maintenance optimization by determining damage components from operating data

The invention relates to a method for setting maintenance intervals of motor vehicle assemblies or their subunits, which are to be replaced during the maintenance of the vehicle with a certain degree of wear or fatigue. These subunits are referred to below as maintenance units, since they are typically replaced as a whole in maintenance tasks.

For the maintenance of motor vehicle assemblies, individual maintenance units are usually replaced after a certain fixed service life or alternatively after a certain mileage. However, such a rigid determination of maintenance intervals has the disadvantage that maintenance units with very different wear and tear or

Degrees of fatigue are serviced or replaced entirely. It does not take into account the form in which these maintenance units were charged in the previous operation. Clear differences can be identified here with regard to wear and fatigue, due to a different period of use, the type of payload, the characteristics of the distance traveled

Routes, different climatic conditions and differences in individual driving behavior. A general assumption of maintenance intervals is based on a certain averaging of these loads. However, maintenance units are often replaced too early or vehicle failures occur because unusually heavily loaded components fail within the usual maintenance interval.

An individual decision about the replacement of maintenance units requires regular inspection and usually an exposure or expansion of the respective maintenance unit. This contrasts with fixing rigid

Maintenance intervals represent a considerable additional effort. In addition, a decision to replace maintenance units is usually based only on one

BESTATIGUNGSKOPIE external appearance in connection with the subjective overall impression of the motor vehicle. It therefore depends on the inspector how accurately the risk of failure of a particular maintenance unit is assessed. Nonetheless, even in the case of extensive experience, precise information about the actual exposure history is missing in most cases

Maintenance unit. For complex assemblies, such as Vehicle-driven, the components are usually not accessible, which means that there is no possibility for a visual assessment.

The invention has for its object a method for determining

To specify maintenance intervals for motor vehicle assemblies, which overcomes the disadvantages described above of general maintenance intervals or subjective individual decisions. The method should enable maintenance to be adapted as well as possible to the actual degree of wear and the resulting failure probability. At the same time

The method can be implemented as simply as possible with existing electronic control components of motor vehicles.

The inventors have recognized that the intervals after which maintenance units have to be inspected or replaced can be determined in a suitable manner by a multi-stage method.

In a first stage of the method, the relevant loads on a maintenance unit are analyzed, and the failure behavior associated with these loads is carried out by tests and calculations, in particular

Correlation calculations. Different criteria for the failure of a maintenance unit can be defined, such as for example inadmissible wear, breakage, deformation or an intolerable change in function.

The result of this first stage of the process is a map in which the failure behavior depending on the different load levels and Load frequency is shown. Depending on the selection of the relevant loads, the failure behavior is represented by a map with one or more parameters.

In the second stage of the process, those that occur while driving are

Loads on the maintenance unit determined. In the simplest case, this involves recording the frequency of the occurrence of a certain load. In many cases, however, a refined procedure is necessary in order to be able to record the occurring loads with sufficient accuracy, especially if there are several relevant load factors or if dynamic effects have to be taken into account.

For this purpose, the load can be divided into classes and the damage-related events counted for each of these classes. This counting can be done in different ways, for example by determining the length of stay in the respective class, the frequency of exceeding the class limit or the reversal of the load in the respective class. As a result of the second stage of the process, a load spectrum is created which is representative of the stress on the maintenance unit.

In the third stage of the method, the load spectrum determined during driving operation is appropriately offset against the failure behavior of the maintenance unit known from the first stage. The result is an abstract quantity called the sum of damage. This sum of damage, which increases during operation and can be stored in a simple manner, is the parameter in the method according to the invention which leads to an inspection of a maintenance unit, on the condition that the sum of damage exceeds a certain critical threshold. The critical threshold can be determined by knowing the failure behavior from the first

Stage of the procedure. However, the uncertainties that result from both the correlation calculations and the Map definition as well as from the simplified tests to determine the failure behavior result. The determination of a failure behavior representative of practice requires a lot of experimentation and cannot always be carried out satisfactorily.

This is followed by a fourth stage of the method according to the invention, according to which the failure behavior of the maintenance units can be determined in practice using a series of field data. For this purpose, the damage amounts achieved in the field are determined and compared with the test results from the first stage of the process. Adjustments to the real failure behavior can either be made by changing the determination of the damage sum in the electronic control device or by moving the critical threshold of the damage sum.

With knowledge of the damage amount, each maintenance unit will be one with a low one

Size allocated to effort to be stored and processed, which takes into account the actual load on the respective component and determines the degree of damage in a simplified, yet sufficiently precise manner. This characteristic of the damage allows the inspection to be carried out within a significantly better adjusted maintenance interval. This results in a significantly improved definition of maintenance intervals with a ^" significantly smaller bandwidth, within which, for cost reasons, early inspections can be dispensed with and at the same time the probability of failure can be reduced to a minimum.

The invention is explained in more detail with reference to the drawings. The following is shown in detail:

1 shows a flow diagram of the method according to the invention for optimizing maintenance in a control device of a vehicle. 2 schematically shows the subdivision of a motor vehicle assembly into

Maintenance units using the example of an automatic transmission.

Fig. 3 shows schematically the relationship between the amount and frequency of loading on a maintenance unit and the

Failure behavior.

Fig. 4 shows an example of the division of a burden on a

Maintenance unit in classes graded according to the load level in connection with the frequency of the occurrence of a corresponding load. i

Fig. 5 shows an example of the linkage of the load spectrum and

Failure behavior to determine the amount of damage.

Fig. 6: shows an example of the counting process for creating a

Belastungskolletivs.

FIG. 7 shows an example of the probability of failure of two maintenance units as a function of the damage sum, as well as the

Definition of a critical threshold for one of the maintenance units.

FIG. 2 shows the maintenance units (W1-W7) of an automatic transmission, which is used in the following as an example to illustrate the invention. The division of the maintenance units shown in FIG. 2 is also only an example. In practice, it will result from the design of the motor vehicle assemblies, which also defines the procedure for maintenance work. For design reasons, only certain units as a whole can be inspected or replaced, with the installation position and interaction with other maintenance units playing an important role here. Another division into Maintenance units will result from the type of load during operation and the resulting signs of wear and tear or the fatigue of this component.

Using the inventive method for determining

Maintenance intervals can also be used to classify motor vehicle assemblies into maintenance units so that the assignment of loads to damage amounts shown below can be carried out in a suitable manner. This means that the type of load on a maintenance unit should be classifiable as clearly as possible.

The operating data of the vehicle are the starting point for determining the load on a maintenance unit. This includes sensory data, for example on engine speeds, torques, temperatures of different vehicle components, but also vehicle dynamics data that characterize driving conditions. Furthermore, the operating data also include encoder signals which correspond to the driver's request, for example after acceleration. In today's vehicles, sensory data and encoder signals are available via communication networks, such as the CAN bus, and are used for vehicle control, such as optimizing the switching behavior in automatic transmissions. The term operating data in the following book is understood to mean all information about the vehicle, its payload and the vehicle components, which is typically stored in control units and is accessible via the communication links mentioned. For example, these are characteristic curves of the motor, the gearbox, the axle and gear ratios, and moments of inertia, etc.

Based on the operating data, the loads on the respective maintenance units can be estimated. For this purpose, the loads associated with the operating data or a part of these data must be stored in the control unit of the maintenance units or the motor vehicle modules. The load effects result from the effect of Torque, speed, pressure, temperature on the maintenance unit and the resulting effects such as friction, tension or deformation. Such an assignment between operating data and load is to be specified for each maintenance unit and is determined by the manufacturer of the maintenance unit as a load model. These assigned load data can also be stored in a central control unit for all maintenance units.

For the first step of the method according to the invention, it is necessary to establish a quantitative relationship between stress and the probability of failure using stress tests. In many cases it is

Stress tests with repetitions at a certain stress level up to failure in combination with correlation calculations. In the simplest case, the number of load events, for example the number of gear changes in a transmission, is linked to the failure behavior. Figure 3 shows in a schematically simplified manner the scatter band of the

Failure behavior for a selected load in relation to the load level and the frequency with which it occurs. The hatched area in FIG. 2 extends from a lower threshold, at which the probability that a given pair of stress level and stress frequency leads to a failure is approximately 1%, to an upper threshold, which is accordingly characterized in that the -The probability of failure is 99%. These curves can of course only be given approximately and, depending on the type of determination and the number and accuracy of the tests carried out for this purpose, have a certain spread. Thus, setting a threshold that is representative of the

Failure behavior is at the discretion of the person skilled in the art and, depending on the type or meaning of the assigned vehicle assembly, is on the safe side towards lower failure probabilities or, if this is tolerable, shifted towards a higher failure probability. In FIG. 3, this is referred to as critical failure threshold 1 in the following for illustration

Definition, drawn in with a probability of default of 50% and is shown as a dashed line. Furthermore, it can be advantageous that this is critical Failure threshold is not assigned to the same failure probability across the entire map. If, for example, it is known from load tests that the actual failure behavior fluctuates significantly in statistical terms, the critical failure threshold can be corrected in areas of great variation by an average failure probability so that the critical failure threshold is shifted to a safe side. Essentially, the result of the first stage of the method according to the invention will therefore be to determine relevant loads that lead to failure in the accumulation and to assign them a critical failure threshold in a characteristic diagram.

Non-destructive measurement methods can be used to characterize the signs of fatigue during load tests. An example of this is the X-ray refraction method, which can be used to detect microstructural defects such as microcracks resulting from the stress. Simulations of repeated component loads, damage accumulation calculations and plausibility assumptions are also suitable for establishing a quantitative relationship between load factors and damage.

Crucial for the subsequent second step of the invention

The method is to determine the loads relevant for a maintenance unit within a time interval from the operating data and to classify them into a load pattern graded according to the load level. Based on this classification, the damage-relevant events can now be counted. This is what a counting means in the broadest sense

Understand the frequency with which a class corresponding load occurs. An example of this is a division of a thermal load into several temperature overshoot levels, so that an example of an event-controlled counting in a single class is a temperature overshoot of 90 ° and the frequency with which this is recorded

Load occurs, is counted. In a generalized case, this frequency recording of a load in a class can also be a determination of the The length of time over which a load corresponding to the class occurs. Another way of recording the frequency of the occurrence of a load belonging to a class can be to determine the frequency of crossing the class boundary in a defined or any direction or to reverse the load in the respective class.

As a result of this second step, the load history is shown in the form of a load collective, as is shown in a schematically simplified manner in FIG. 4. Four different classes I - IV are shown with increasing load levels as well as a corresponding count of the load events assigned to the respective classes.

From the subsequent comparison of this load collective with the failure behavior with such a load in the third stage of the method according to the invention, an increase in the damage sum at the corresponding time interval is determined. This will be explained using a schematically simplified example, shown in FIG. 5:

The failure behavior determined in method step 1 is shown in FIG. 5 with the aid of two curves which, depending on the load level and the frequency of occurrence of such a load, approximately describe a failure probability of 1% and 99%. In addition, the load spectrum of the second process stage is added to FIG. The frequency Z _\ with a load from the highest is shown in dashed lines

Stress class IV has occurred. A repetition number ZAI for this load was determined for this class until the critical failure threshold was reached; Here, a default probability of 1% is assumed as the critical default threshold. Accordingly, the previous number of repetitions in this class can be assigned a damage sum S, which results from the

Quotient Z | / ZAI determined. Accordingly, an increase in the total damage is calculated as an increase in the frequency of exposure ΔZ | by ΔS = ΔZ | / ZAI. at The progressive curve for the failure probability shown in FIG. 5 results in the fact that smaller loads lead to a smaller increase in the damage sum, and the damage sum is increased correspondingly more with higher loads. In addition, it is conceivable to correct the failure behavior depending on different load parameters. This can take place, for example, at elevated temperature, pressure, etc. In these cases, the load spectrum and failure behavior must be evaluated separately and the results of the damage sum calculation must be combined.

As an illustrative exemplary embodiment, a gear part which is loaded by torque fluctuations and which can be replaced as a maintenance unit is considered. It is further assumed that the torque fluctuations can be determined from the operating data or that these can be estimated at least indirectly via a model of the transmission and the measurement of the drive and driven torques. First, the load is divided into classes, for example in the present case several threshold values can be set for the torque and an event is added to a load class when the torque threshold is exceeded from the lower to the higher torque. Possible finer gradations and non-equidistant

Step sizes in the classification are advantageous in areas in which the failure behavior shows large variations with small parameter changes. This is particularly the case in the vicinity of maximum loads.

The procedure for counting load events in this illustrative example is illustrated with reference to FIG. 6. The time curve of the torque and four threshold values for the torque are shown. Vertical arrows indicate that one of these threshold values has been exceeded and the corresponding assignment to a load class.

In this way, a load collective arises during operation, which corresponds to that from the first stage of the method according to the invention determined failure behavior is offset. The increase in the number of loads in a class leads to an increase in the total damage, which, as shown above, is determined by the quotient of the actual load or the actual increase in load and the average number of loads assigned to this class up to the critical failure threshold. This share of

Damage total is calculated for each class, so that the total damage total for this maintenance unit is made up of the subtotals that result for the different classes.

The damage sum of a maintenance unit can consist of different contributions and thus of weighted partial damage sums if there are different, non-interacting damage mechanisms that cannot be meaningfully combined in a uniform load pattern. This applies in particular if the load factors have significantly different time constants and thus the increase in the partial damage sums are recorded with different time intervals. A combination of permanent loads and stress events is also possible through the summation of separately determined partial damage sums to a damage sum assigned to the maintenance unit.

In addition, the method according to the invention has a fourth stage, in which a critical threshold value S for the damage sum is set on the basis of real field data in order to correct inaccuracies in the preceding method steps. According to the procedure according to steps 1 to 3 of the procedure, the abstract size of the damage sum should best represent the actual load on a maintenance unit and thus its failure behavior. Accordingly, when recording the sum of damage during operation, this should be in a typical relationship with the probability of failure, which is shown schematically in FIG. 7. Two interpolated failure probabilities for component A and component B in

Dependence of the damage sum shown. Characteristic in both cases is a largely failure-free area for damage amounts below one certain limit. The critical threshold value S _k for the damage sum is now to be selected such that it still comes to a high degree of certainty in this failure-free area or, should a certain failure risk be tolerable for a structural unit, at least just at the beginning of the increase in the failure probability. By determining the critical threshold value S _k for the

A certain amount of error correction in the previous procedural steps must be carried out. In addition, if the results from operating a large number of identical maintenance units differ significantly, a very simple correction of maintenance intervals and thus the risk of failure can be carried out by setting the critical threshold value S _k for the

Damage amount is adjusted.

Figure 1 summarizes the inventive method. It shows the four process stages for determining maintenance intervals in the sequence in which they can be implemented in a control unit of a vehicle. The control unit is connected to the sensor system of the vehicle and the control and regulation data of the maintenance units and also receives encoder signals. Additional vehicle data and load models for the maintenance units to be monitored are stored in the control unit. As shown in the flowchart, external and internal information is added

Operating data connected. In the subsequent step, which corresponds to the second stage ^"of the method according to the invention, these operating data are classified. It is possible to use both the current operating data and also those from a specific time interval for the classification. The result is the processing of the data Maintenance units belonging

Load history for each assigned load collective. These are then linked to the failure behavior of the maintenance units known from the first method step, only the corresponding information about the failure behavior being stored in the control device and the corresponding tests and correlation calculations being carried out independently of the control device. The combination of load collective and failure behavior in the control unit becomes the abstract, if possible the Determination of the amount of damage corresponding to the load history. This corresponds to the third stage of the process. The subsequent fourth process stage is used to compare the damage sum with a critical threshold value S _k . If this value is exceeded, a maintenance interval is completed and customer service with maintenance and possible replacement of the corresponding one

Maintenance unit is the result. If this is not immediately possible, at least one of the reaction options is to operate the maintenance unit carefully.

The procedural steps that are carried out in the motor vehicle itself are very simple and can usually be carried out using control devices that are already available. This is achieved by the complex procedural steps of creating a load model to estimate loads from the operating data, the procedure for classifying loads into load patterns as well as creating load collectives and assigning the loads to one

Damage total increase based on preliminary examinations, the results of which can be programmed in the control implementations. The simplicity of the processing of the method in the motor vehicle itself is thus achieved through the preparatory work before starting up. The same applies to the fourth process stage, in which the elaborate determination of a critical

Threshold is carried out regardless of the vehicle. Advantageously, the process can be verified by constant comparison with field data from customer service. It is thus possible to adapt the critical threshold values to the experiences that result from the care of a correspondingly large one with each service inspection

Vehicle fleet results.

After the critical threshold value of the damage sum has been exceeded, various options are indicated in FIG. 1. In addition, the driver or the control unit can inform the service center about the expiry of a maintenance interval via a wireless connection when the critical threshold value is reached. In an advantageous embodiment of the invention, the If the critical damage sum has been reached, operation of the component is only carried out in a gentle mode or a corresponding recommendation is signaled to the driver.

Claims

claims

1. Method for determining maintenance intervals for motor vehicle assemblies which comprise at least one maintenance unit; 1.1 each maintenance unit is assigned a sum of damage, the value of which is at

Operation of the motor vehicle increases;

1.2 the load acting on a maintenance unit is determined or estimated in a selected time interval from the operating data of the vehicle or from part of this data and divided into different classes of a load pattern;

1.3 from the comparison of the load pattern of a time interval with stored data of the failure behavior of the maintenance unit, the increase in the damage sum for this time interval is calculated;

1.4 The end of the maintenance interval for a maintenance unit is reached when the assigned amount of damage reaches a certain critical level

Threshold exceeded.

2. The method according to claim 1, characterized in that the time intervals have a regular duration in which the classification of the load acting on a maintenance unit is carried out in load patterns.

3. The method according to claim 2, characterized in that the length of the time intervals is adapted to the respective maintenance unit.

4. The method according to claim 1, characterized in that the time intervals are different, in which the division of the load acting on a maintenance unit into load patterns is carried out and its start and duration is determined by the occurrence of defined operating states.

5. The method according to claim 1, characterized in that the number of repetitions with a load corresponding to the class is stored as data on the failure behavior for each class of the load pattern until a defined failure probability is reached.

6. The method according to claim 5, characterized in that when a load event occurs, which can be assigned to a class of the load pattern, the damage sum of a maintenance unit increases by the reciprocal of the number of repetitions until a defined failure probability is reached.

7. The method according to any one of claims 1 to 6, characterized in that the damage sum of a maintenance unit consists of partial damage sums which are assigned to the individual classes of the load pattern.

8. The method according to any one of claims 1 to 7, characterized in that the critical threshold value is determined from the failure probability of the maintenance unit as a function of the assigned damage sum.

^" 9. The method according to claim 8, characterized in that the critical threshold value of the damage sum is chosen so that the probability of failure remains below a certain value.

10. The method according to any one of claims 1 to 9, characterized in that the damage sums of the maintenance units of a motor vehicle assembly are stored in a control unit which is assigned to the motor vehicle assembly.

11. The method according to any one of claims 1 to 9, characterized in that the damage sums of the maintenance units of all Motor vehicle assemblies are stored in a central control unit.

12. The method according to any one of claims 1 to 11, characterized in that the vehicle operation is adjusted depending on the damage sums.

13. The method according to any one of claims 1 to 12, characterized in that the damage sums of the maintenance units can be read out for inspections.

14. The method according to any one of claims 1 to 13, characterized in that a warning is triggered when a certain amount of damage.

15. The method according to any one of claims 1 to 14, characterized in that the method is used for maintenance units in an automatic transmission.

16. The method according to any one of claims 1 to 15, characterized in that the method is used for maintenance units in a drive system.

17. The method according to any one of claims 1 to 16, characterized in that the method is used for maintenance units in a brake system.