Method for identifying defect in system

Abstract

To provide a method for identifying a defect in a system (100).SOLUTION: A method includes detecting parameter values for a plurality of state indexes of a system (200, 210, and 220), the state indexes including the parameters measured in the system (200) and / or parameters calculated or modeled from the measured parameters for the system (220) (200, 210, and 220), identifying the current evaluation index for the defect on the basis of the comparison between the parameter values detected for the state indexes and a prescribed reference value for each state index, and determining a defect probability on at least one component of the system (100) on the basis of a combination of evaluation indexes for a prescribed group of the state indexes each of which is assigned to at least one component.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a method for identifying defects in a system, a computing unit for implementing the method, and a computer program.

Background Art

[0002] Background of the Invention To monitor the functionality in a vehicle and identify errors, various different means are used. In particular, this includes on-board diagnostics in a control device for various different functions, where the on-board diagnostics continuously monitors, models specific parameters and measured values, and checks whether the results comply with predetermined boundary values. In this case, if an error or an unexpected result is identified, a signal (for example, a warning lamp that lights up, display of information on a display) can be output to the user, or can be read out from the factory.

Summary of the Invention

Problems to be Solved by the Invention

[0003] The purpose of such diagnostics is always to identify the error source as accurately as possible (pinpointing). If the unit with a defect cannot be identified, the suspected components have to be replaced, which is time-consuming and costly. For example, the diagnostic element may indicate an increase in a specific emission value, but in this case, it is not even clear which component is causing this undesired increase. Similarly, there may be a deviation caused by the interaction of a plurality of components instead of independent errors, and thus error entries for individual components in the control device.

Means for Solving the Problems

[0004] Disclosure of the Invention According to the present invention, a method for identifying defects in a system having the features of the independent claim, and a computing unit and computer program for carrying out the method are proposed. Advantageous configurations are described in the dependent claims and the following specification.

[0005] The present invention presents an improved diagnostic means for identifying and limiting errors in systems such as vehicles. In particular, a method is first proposed for detecting parameter values ​​for multiple state indicators of a system, where the state indicators include measured parameters of the system and / or parameters for the system calculated or modeled from the measured parameters. Next, for each of these state indicators, a current evaluation index for defects is identified, where the evaluation index is identified based on a comparison of the detected parameter value for the state indicator with a predetermined threshold value. Then, based on a combination of the thus identified evaluation indexes for a predetermined group of state indicators, the probability of a defect for at least one component of the system is determined, where the group of state indicators is assigned to at least one component. By evaluating multiple state indicators in combination based on probability, errors or changes that have not yet appeared in individual threshold exceedances or error indications can also be identified. In this case, the evaluation index can initially represent a key numerical value specific to the state indicator, and here, the evaluation indexes do not need to be equivalent to one another. The evaluation index can, for example, show probability values ​​for error or defect states, but can also show intervals between parameter values ​​and boundaries, and intervals between the distribution of parameter values ​​and typical or similar key numerical values. Intervals can also be identified, for example, as a 3σ deviation from the fleet mean or as a percentile value for a specific parameter (e.g., the 97th percentile).

[0006] Furthermore, this method allows for the determination of diagnostic probabilities, which indicate the probability that any possible diagnostic steps for further inspection and / or limitation of an error or defect in a component are most relevant to the inspection or limitation of the error. A diagnostic step may be, for example, a leak test of an air system or coolant system in relation to an underlying system error or higher-level defect, or a test for boost pressure increase in a turbocharger. The determination of the diagnostic probabilities is, in this case, based on a combination of evaluation metrics for a given group of condition metrics, each assigned to at least one diagnostic step. Furthermore, this determination may also include the diagnostic cost of the possible diagnostic steps and the determined defect probability. By determining the diagnostic probabilities, the factory can specifically identify the necessary diagnostic steps for limiting an error and present these diagnostic steps in an efficient order with respect to defect probability and diagnostic cost. This enables efficient error diagnosis and, consequently, system repair.

[0007] In addition, this method also determines the causal probability of a defect or error, which indicates the probability that an underlying system error or an error that triggers a higher-level defect exists for various different defect or error possibilities. Examples of defect or error possibilities include insufficient cooling capacity or internal coolant leakage. The determination of causal probability may be based on a combination of evaluation indices for a given group of condition indices assigned to the defect or error possibilities, similar to the determination of defect probability and diagnostic probability. This allows for the identification of the cause of the system error or higher-level defect during factory repairs, thereby further narrowing the error search. This makes system repair easier. The processes described in subsequent steps for determining defect probability and / or diagnostic probability can similarly be used to determine causal probability.

[0008] Identifying the current evaluation metrics for a state indicator may, for example, include comparing a parameter value to one or more thresholds set as reference values; comparing a parameter value to reference values ​​determined for a state indicator in one or more other systems; comparing a parameter value to reference values ​​found from statistical characteristic values ​​of multiple parameter values ​​for a state indicator in multiple other systems; or comparing a parameter value to a system parameter value set as a reference value, found at one or more preceding points in time.

[0009] Identifying the current evaluation metric for each of multiple condition indicators may, in particular, involve identifying the difference between the parameter value and the baseline value, and determining the evaluation metric in relation to the identified difference. By identifying the difference, a measure is found that represents how far the measured parameter value is from a defined defect threshold, or how far the parameter value is from a value expected to be normal, and this measure can then be used to determine the evaluation metric for the defect. Parameters that are close to the boundary value for defects are more likely to indicate a current or impending defect.

[0010] Optionally, state indicators can be classified into multiple correlation groups. These correlation groups may correspond to the underlying system error or higher-level defect to which this method is applied. For example, the underlying system error may be an increase in NOx emissions measured by a NOx tailpipe sensor, and other emissions (NH3, CO, HC, etc.) may also be considered. Furthermore, the underlying system error or higher-level defect may be an increase in the consumption of operating resources or energy, an increase in noise emissions, an increase in product vibration, or any change in the product in general. In the above example, the correlation groups are emission correlation groups to NOx emissions, where each state indicator is divided into a correlation group based on its own correlation to NOx emissions. Thus, for example, it is possible to form three correlation groups (high, medium, low) to which state indicators are assigned. However, it is also possible to form only two correlation groups, or more than three. Here, each correlation group is further assigned a correlation coefficient. The correlation coefficients may have values ​​such as 1 (high), 2 / 3 (medium), and 1 / 3 (low) when dividing into three correlation groups. However, the correlation coefficients may also have heterogeneous intervals, for example, 0.95 (high), 0.50 (medium), and 0.25 (low). Furthermore, correlation coefficients with values ​​greater than 1 are also conceivable. The evaluation index is then obtained by multiplying the defined evaluation index by the correlation coefficient in relation to the correlation group of the state index. Dividing into multiple correlation groups based on the underlying system error can improve the accuracy of determining the defect probability for a component and / or the diagnostic probability for a diagnostic step.

[0011] Here, determining the probability of a defect for at least one component of the system and / or the probability of a diagnosis for at least one diagnostic step may involve identifying combinations of evaluation metrics based on existing combinations. These existing combinations may be stored, for example, on an external server or in cloud memory, and may originate from a set of data from a fleet of similar or typical systems. For example, it can be inferred from the set of data that some combinations are actually very rare or never occur, while others are likely to occur frequently. The determination of the probability of a defect for at least one component of the system and / or the probability of a diagnosis for at least one diagnostic step can then be performed based on the identified combinations of evaluation metrics. By using a set of data to determine the combinations, the combinations of evaluation metrics are limited to frequently occurring combinations, thus reducing the computational power required to determine the probability of a defect for at least one component of the system and / or the probability of a diagnosis for at least one diagnostic step, and increasing the accuracy of those determinations. Furthermore, this allows us to examine whether the identified combination of evaluation metrics and the resulting defect and / or diagnostic probabilities are appropriate for the defective component or determined by the correct diagnostic steps. This enables further adjustment of existing combinations, which in turn can lead to more precise determination of defect and / or diagnostic probabilities.

[0012] Determining the probability of a defect for at least one component of the system and / or the probability of a diagnosis for at least one diagnostic step may include weighting each evaluation metric by a weighting coefficient. The weighting coefficient may be, for example, probabilities arising from information obtained from a diagnostic feasibility analysis (DMA) or guided error search (GFS) in the factory. This information is obtained from engineering and empirical knowledge already existing before the start of system manufacturing, which may originate from older systems and / or be determined in a prior verification fleet. The probability of a defect for at least one component of the system and / or the probability of a diagnosis for at least one diagnostic step are then determined based on the weighted evaluation metric values. Here, for example, the probability of a defect or a diagnosis can be set to the maximum value below the value of the evaluation metric. Selectively, all evaluation metrics related to the component may be proportionally incorporated into the probability of a defect or a diagnosis. By weighting the evaluation metrics with empirical values ​​influencing this method that exist before the start of system manufacturing, the identification of errors or defects can be improved.

[0013] Detecting parameter values ​​may include, in particular, obtaining parameter values ​​as output signals from sensors, and / or obtaining parameter values ​​as output values ​​from a computing module configured to calculate parameter values ​​from input values, and / or receiving parameter values ​​via a communication connection. In particular, parameter values ​​are detectable in a system where defects should be identified (e.g., in a vehicle control system), while the steps described above for evaluating parameter values ​​and determining the probability of defects are performed on other devices (e.g., factory computers, servers, or computing networks) and detected via a communication connection for this purpose. Stored data (e.g., data from other vehicles or other data from the manufacture of components, data input status of control systems in vehicles) can also be received and processed from remote memory means or cloud memory means or from a remote server.

[0014] Optionally, for at least one state index, it may be checked whether a predetermined condition is met during the detection of the parameter value, and if this condition is not met, the parameter value to which it belongs may be discarded. Optionally, the condition may already be checked during the measurement or calculation of the parameter value, and if the condition is not met, the measurement is discarded or the calculation is not performed. The condition may be, for example, the existence of a suitable and meaningful operating range, such as the existence of a steady or quasi-steady operating point, or the existence of a defined operating range with certain limitations.

[0015] The state index may, in particular, include a parameter representing a measured deviation from a given normal state, and / or a parameter representing a measure of the healthy state of at least one component of the system.

[0016] Furthermore, if the probability of a defect in a component exceeds a predetermined threshold, an instruction may be output to the user. The threshold may be the same or different for each component, and may be adjusted as needed. This allows, for example, the user to be shown that inspecting a particular subsystem is meaningful in order to identify a safety-critical error early on.

[0017] Optionally, data on actual defects and / or replacements of system components and / or performed diagnostic steps can be obtained. Such data may be obtained, for example, through user input or through automatic identification of replaced components. This data can then be compared with the determined defect probabilities for the components. That is, for example, it can be checked whether the determined high defect probabilities correlated with actual defects or whether there was an error in the judgment. Furthermore, this data can also be compared with the determined diagnostic probabilities for the diagnostic steps. That is, it can be compared which diagnostic steps were performed, whether a high diagnostic probability was determined for these diagnostic steps, i.e., whether these diagnostic steps were able to contribute to limiting the error. Based on this comparison, adjustments can then be made to the identification of evaluation metrics and / or the determination of defect probabilities or diagnostic probabilities. These adjustments can be made, for example, by changing the assignment of a predetermined group to at least one state metric, by changing the calculation settings for identifying evaluation metrics, by changing the calculation settings for identifying defect probabilities and / or diagnostic probabilities, or by changing the weighting of state metrics for identifying defect probabilities and / or diagnostic probabilities. This allows for continuous improvement and adaptation of the evaluation system that persistently detects defect probabilities from parameter values, thereby enabling more accurate diagnoses.

[0018] Since the parameter values ​​for one or more state indicators are detectable, particularly over time, this progression can be incorporated to identify evaluation indicators for defects, for example, by incorporating sudden changes alongside gradual changes over long periods. Furthermore, average values ​​can be formed from the parameter values ​​detected over time, which can be used to construct evaluation indicators.

[0019] The steps described are particularly suitable for monitoring and diagnostics in vehicles, for example. In vehicles, multiple parameter values ​​are usually already detected, modeled, and processed by a control device or other unit, so at least one portion of these values ​​can be used as a state indicator for the steps described above. However, such a method can also be used in other systems where specific parameter values ​​can be inspected continuously or at predetermined intervals, such as stationary or mobile fuel cells, electric drive systems, or other drive systems. Essentially, it can be used in any system where defects can occur and information can be detected, and by extension, in many technical systems and manufacturing equipment.

[0020] Overall, the described modifications make it possible to identify defective or deteriorated components in the system and perform appropriate repairs, as well as to estimate which components are likely to fail in the future, which in turn enables spare parts planning, for example, by the manufacturer. In such methods, measures can be taken early to address potential problem areas even if defects have not been detected, for example, by adjusting the software module during control or replacing components with improved parts before errors occur.

[0021] A computing unit according to the present invention, such as a control device for an automobile, is configured to carry out the method according to the present invention, particularly by programming techniques.

[0022] It is also advantageous to implement the method according to the invention in the form of a computer program or a computer program product comprising program code for performing all the steps. This is because, especially when the control device to be executed is further utilized for other tasks and thus exists anyway, it causes only a particularly low unit price. Finally, a machine-readable storage medium in which a computer program as described above is stored is provided. Suitable storage media or data carriers for providing the computer program are, in particular, magnetic memories, optical memories and electrical memories, such as, for example, hard disks, flash memories, EEPROMs, DVDs, etc. Downloading the program via a computer network (Internet, intranet, etc.) is also possible. Such a download can, in this case, be carried out by means of a wired or cable connection or wirelessly (for example, via a WLAN network, 3G connection, 4G connection, 5G connection or 6G connection, etc.).

[0023] Further advantages and configurations of the invention will become apparent from the specification and the attached drawings.

[0024] The invention will be schematically illustrated in the drawings in accordance with embodiments and will be described hereinafter with reference to the drawings.

Brief Description of the Drawings

[0025] [Figure 1] FIG. shows an exemplary system comprising a vehicle and a processing unit for performing steps. [Figure 2] FIG. shows exemplary steps in a flowchart.

Embodiments for Carrying Out the Invention

[0026] Embodiments of the Invention The following example relates to a vehicle as a system in which various types of data are detected, evaluated, and / or transmitted. Figure 1 illustrates a vehicle as a system 100, in which data should be evaluated probabilistically in a processing unit 130. Often, for this purpose, the vehicle 100 is equipped with one or more control devices 110, which centrally or specifically control data detection (e.g., measurements at fixed intervals) with respect to a particular function. Similarly, the data detected in such control devices can be evaluated at least partially and used for calculation and modeling and for control and adjustment in the vehicle. However, the method described is generally applicable to any system in which condition diagnosis is important and various types of data are detected, such as construction machinery, automated production machines, fuel cells, feed drives, turbines, etc. In such systems as well, there is usually a corresponding control unit 110, such as a control device, microcontroller, or at least a similar element that can control the detection of various types of data. In this case, the measurement data itself can be detected by any sensors and measuring equipment 120, 122, 124, for example, by temperature sensors, pressure sensors, mass flow sensors, rotational speed meters, vibration sensors or acoustic sensors, acceleration meters, etc. Although sensors 120, 122, 124 are only schematically illustrated in Figure 1, it is obvious that in vehicles or other systems, multiple different sensors are usually used in different subsystems. Such measured values ​​can be used directly or can be included together in models and calculations that yield a variety of different parameters.

[0027] The method described herein allows for the evaluation of large amounts of data to enable system diagnosis even in complex cases. The data can be detected, processed, appropriately aggregated, and transmitted to, for example, a remote server 130 or another suitable computing unit, where it can be used to perform a probability-based diagnosis. Based on this diagnosis, similar decisions for on-site diagnosis, component replacement, or other subsequent steps can then be performed.

[0028] Here, the diagnostic results may be obtained or improved by incorporating data collected from other vehicles 140 into the evaluation. This data may be stored, for example, on a remote server or in cloud memory, so that the evaluation unit 130 can access it. This data may also be collected and aggregated in a similar or other manner. Here, comparison with fleet data 140 may be combined with the evaluation of individual vehicles. In other words, the diagnosis is influenced in particular by the appropriate selection of monitored features, their aggregation, and the combined evaluation. Here, relatively old data from individual vehicles or other vehicles may also be included in the evaluation, and the evaluation may be structured in such a way that a learning diagnostic system is created, and the results of this diagnostic system are increasingly improved over time.

[0029] The data used for evaluation is generally referred to as state indicators below. Multiple different values, computational parameters, and modeling parameters are considered state indicators. In particular, multiple parameters already detected or available in the control unit to implement other functions can also be evaluated. This generally eliminates the need to detect or identify additional parameters as state indicators. However, similarly, only specific parameters, calculations, or evaluations may be used as state indicators and therefore provided specifically for this purpose.

[0030] Figure 2 shows exemplary steps, which will be described in more detail below. Steps 200 to 240 can be performed, for example, in the control device 110 of Figure 1. Here, in step 200, parameter values ​​can first be measured, i.e., obtained, for example, by a sensor. Furthermore, in step 210, other parameter values ​​can be formed from the measured parameter values ​​and other input values ​​by calculation, modeling, and similar processes. In step 220, the parameter values ​​thus obtained can be further aggregated, thereby reducing the amount of data required for evaluating the state indicators. For example, individual discrete measurements or a predetermined number of measurements at a given time, e.g., the last measurement, the last 10 measurements, or any other number of measurements, can be taken up to form the average value of the parameter over a specific time window, or only values ​​that satisfy a specific boundary condition, or only values ​​for which such boundary conditions were satisfied at the time of measurement, can be considered. Then, in step 230, the final parameter values ​​for various different state indicators thus obtained can be stored, at least temporarily. These steps of detection, aggregation, and storage can, of course, be repeated at any frequency, and typically, parameter values ​​are detected and processed continuously or at predetermined intervals as long as the system is operating. In step 240, the parameter values ​​thus stored for the state indicators can optionally be transmitted to an appropriate evaluation unit 130, which may be, for example, a remote server or a factory computer, where the collected state indicators can finally be evaluated in step 250.

[0031] As a state indicator, in particular, a parameter having the following characteristics, namely, Parameters having a continuous or discrete format (e.g., percentage values), Parameters that require only a small amount of measurement and / or data for transmission (for example, parameters detected only once per driving cycle or once after a predetermined driving section), Parameters that are as unrelated as possible to the operating point, driving conditions, and surrounding conditions. A parameter that can be compared with other state indicators, and optionally with the state indicators of other components. These are suitable, but not limited to them.

[0032] While these characteristics offer various advantages in evaluation, they are not necessarily required, and it is possible to consider at least partial parameters that do not satisfy all of these characteristics, such as very frequently detected binary evaluations or parameters, as state indicators. In this case, parameters may be detected, for example, at very short intervals for adjustment purposes, but only a selected portion of the detected parameter values ​​are evaluated as state indicators. Similarly, individual values ​​of state indicators can be formed from multiple values ​​for a single parameter by averaging or other statistical operations, for example, by a moving average over a defined time window. Load counters, as described below, as state indicators, do not need to be independent of, for example, the operating point and driving conditions. Furthermore, data that can be indirectly detected or measured in the vehicle can also be incorporated. These can be provided, for example, by factory personnel via an input mask on the factory computer. Examples of such data include engine starting delays, namely, the engine not starting at all, the engine starting with a significant delay (more than 5 seconds), the engine starting with a delay, the engine starting with a jerk, the engine starting only occasionally, or the engine starting only when cold.

[0033] To clarify, possible state indicators can be divided into two groups. In one group, parameters that represent the deviation of components or measured values ​​from the normal state or boundary values, as measured in the vehicle, are suitable as state indicators. These state indicators will be referred to as diagnostic features below. In the other group, parameters that detect the load on components, for example, the count of load fluctuations in a component, are suitable as indicators. These state indicators will be referred to as load counters below. However, as is obvious, state indicators that cannot be classified into either of these groups can also be included in the evaluation.

[0034] As state indicators from a class of diagnostic features, several parameters that are typically available in the control device can be used.

[0035] Possible diagnostic features as state indicators could be, for example, the physical state of the components. This includes directly measured physical values, such as pressure, temperature, mass flow rate, or any other values ​​detected by appropriate sensors. However, the physical state could also be one in which parameters are determined from measurements via a physical model.

[0036] For example, the radiator ventilation level can be determined via a physical model using, for instance, measurements from temperature, pressure, and mass flow sensors, along with vehicle speed, and used as a state indicator. This value is suitable as a state indicator for any cooling system, such as an intercooler, engine radiator, exhaust gas recirculation radiator, battery radiator, fuel cell radiator, etc. Furthermore, the determination of appropriate quantities as state indicators for various types of radiators will be described in more detail below in relation to an example for a diesel engine air system.

[0037] Further physical quantities available as state indicators include, for example, the catalyst conversion efficiency determined from sensor measurements and a model, or the substrate storage capacity determined from the sensor and a model (e.g., the NH3 storage capacity of an SCR catalyst). Another example is the flow resistance characteristic value calculated from a differential pressure sensor, which can be determined, for example, via a particle filter, catalyst, exhaust gas recirculation pipeline, fuel cell stack, or other suitable part of the system.

[0038] Typically, learned values ​​for sensors and actuators, which are learned and used in control devices to correct their values, can be considered another group of state indicators that can be listed as diagnostic features, and therefore these values ​​often already exist. In particular, these may be learned values, which represent, for example, the drift characteristics of a sensor or actuator, such as the drift of an air mass sensor, or the learned deviation of a pressure sensor when compared with ambient pressure under appropriate operating conditions.

[0039] For example, learned values ​​for the injection behavior of an injector to adjust the amount of injector, or learned values ​​for the "open" and "close" stoppers of a flap or valve, can also be used, which may indicate, for example, a mechanical defect or contamination.

[0040] Similarly, counters of various functions can be used as diagnostic features and, by extension, as indicators of condition. This is especially true for counters that indicate how frequently a regeneration function for a component has been activated. Such regeneration processes frequently change the underlying system, and therefore, as the number of processes increases, the effect of regeneration weakens. Similarly, counters of regeneration processes and similar processes can be used as counters for specific load conditions of a component, and consequently, for degradation. For example, a counter for the number of regenerations of a particle filter can be used as an indicator of condition. During particle filter regeneration, the temperature is temporarily raised significantly, which burns off the particulate deposits. Therefore, such a counter may, for example, indirectly indicate increasing ashing of the filter. Furthermore, the frequency of the required regeneration process may indicate other underlying defects, such as increased soot formation in engine combustion in a diesel engine. The distinction between filter ashing and other defects resulting in increased soot formation can be determined, for example, by other diagnostic quantities or by load counters (e.g., total mileage (km) since the current filter element was installed). As another example, a counter that counts the number of flap release functions can be used as a status indicator. Such a function might utilize a short actuator impulse to release a frozen flap again, for example.

[0041] The debounce time of a function can also be used as a state indicator, or to derive a state indicator from it. The debounce time is defined as a temporal boundary value to record a change, for example, after a predetermined period has elapsed or after a predetermined duration has occurred when a state change occurs on the signal input side. For this purpose, a counter or timer may be started immediately when a particular signal, state, or value first occurs. After reaching a predetermined debounce time, the signal itself can be further processed. In this case, the debounce time can be defined for signal changes or state changes on the signal input side, as well as for changes related to a specific boundary value (e.g., an upper threshold). Therefore, a debounce time can also be defined for defect identification, so that the error itself is identified, stored, or displayed only after the corresponding debounce time has elapsed. For example, when adjusting the air mass in an engine, short-term deviations are normal and should not be avoided, especially in temporary driving conditions, whereas long-lasting deviations suggest an error. In response to this, a debounce time for the deviation signal may be defined. The debounce time is observed in several normal diagnostic functions of the control device and can therefore be easily used as a status indicator without incurring additional costs for evaluation.

[0042] For example, if a diagnostic function detects a maximum debounce time within a single driving cycle or a predetermined driving section (e.g., every 100km), this value can be used in relation to a threshold defined for the defect to determine how far each component was from the defect identification threshold.

[0043] As a further value for the status index, so-called error rate or error quotient can be used. The error rate may be the ratio of the time a component is suspected of being defective (for example, based on exceeding a threshold) to the time a component is considered to be defect-free. The error rate, alone or in relation to the error conditions used, can provide a measure of how far a component is from the defect identification threshold.

[0044] The adjustment parameters of the adjustment unit can also be used as state indicators. For example, adjustment parameters can be detected and evaluated, and in particular, the integrator component (I component) of the adjustment unit can be detected and evaluated. In a tuner, the I component is responsible for steady-state error correction. Therefore, in many functions, the I component approximates zero in a steady state when there are no defects, and conversely, deviations suggest possible drift or defects in relation to the quantity being adjusted. Since such adjustment parameters may be strongly related to the operating point, such adjustment parameters can also be detected and evaluated as a histogram or frequency distribution in relation to specific operating parameters, that is, they can be detected separately for various different operating ranges, for example.

[0045] As an example, the following proposes a variety of possible condition indicators from a range of diagnostic features that can be used for monitoring the condition of a diesel engine's air system.

[0046] As physical states of the components, for example, pressure, temperature, and mass flow rate in various ranges of an air system can be used as state indicators or to form state indicators. For example, the efficiency of various coolers within an air system can be used as a state indicator.

[0047] The efficiency of the high-pressure exhaust gas recirculation cooler can be determined, for example, from the temperature after the high-pressure exhaust gas recirculation cooler, the modeled exhaust gas temperature before exhaust gas recirculation cooling, the modeled pressure before exhaust gas recirculation cooling, the measured intake manifold pressure, the calculated mass flow rate due to high-pressure exhaust gas recirculation (which can be determined, for example, from measurements from an air mass sensor, intake manifold pressure, intake manifold temperature, and engine speed), and the coolant temperature of the high-pressure exhaust gas recirculation cooler.

[0048] For the modeled values, standard physical models can be used, for example, to adjust air systems or to monitor specific values ​​in exhaust gases.

[0049] As another example, the cooling efficiency of a low-pressure exhaust gas recirculation cooler can be determined from the temperature after the low-pressure exhaust gas recirculation cooler, the modeled exhaust gas temperature before low-pressure exhaust gas recirculation cooling, the modeled exhaust gas pressure before exhaust gas recirculation cooling, the measured or modeled pressure before the turbocharger, or the pressure or ambient pressure at an air mass sensor, the calculated mass flow rate through the low-pressure exhaust gas recirculation section (which can be determined, for example, from measurements at an air mass sensor, intake manifold pressure, intake manifold temperature, and engine speed, or selectively from differential pressure across the low-pressure exhaust gas recirculation section and temperature and pressure in the low-pressure exhaust gas recirculation section), and the coolant temperature of the low-pressure exhaust gas recirculation cooler.

[0050] Finally, the cooling efficiency of the intercooler can also be determined from the measurements, namely, for example, the measured temperature after the intercooler, the modeled temperature before the intercooler, for example the temperature after the turbocharger compressor calculated from the compression model, the pressure inside the intercooler, for example the boost pressure before or after the intercooler, the mass flow rate through the low-pressure exhaust gas recirculation section (which can be determined, for example, from measurements from an air mass sensor, intake manifold pressure, intake manifold temperature and engine speed, and optionally, if low-pressure exhaust gas recirculation is used, from the differential pressure across the low-pressure exhaust gas recirculation section and the temperature and pressure in the low-pressure exhaust gas recirculation section), the coolant temperature of the low-pressure exhaust gas recirculation cooler (in the case of a water-cooled intercooler), or the vehicle speed or the drive control of the intercooler fan (in the case of an air-cooled intercooler).

[0051] In the case of air systems, all three of these efficiencies, or just one or two of them, may be selected as exemplary indicators of performance.

[0052] In this case, the efficiency as a state indicator from these measurements can be determined, for example, by forming an average of the parameter values ​​(measured, calculated, or modeled values) over a predetermined period. For evaluation as a state indicator, it may be stipulated that only values ​​detected within a predetermined operating range are used. For example, it may be stipulated that only values ​​in stable operation, steady-state operation, or quasi-steady-state operation are used when evaluating as a state indicator. Selectively or additionally, the effective operating range for evaluation as a state indicator may be defined based on a threshold, so that, in the above example, only values ​​at operating points where, for example, the mass flow rate or coolant temperature is within a specific value range or exceeds a specific threshold may be considered. Additionally or selectively, it may be checked whether there is a sufficiently large temperature difference between the coolant temperature and the exhaust gas temperature at the inlet of the cooling system, and optionally, it may also be checked whether such a condition has already existed for a specific time, for example, by defining a period that is a threshold. If such parameters are already used for a diagnostic function and an effective operating range has been set for this purpose, the same effectiveness parameters can be used for use as a state indicator. However, selectively, different effective conditions may be selected for use as a state indicator than those for the diagnostic function.

[0053] For example, in a control device provided for controlling an air system, cooling efficiency can be provided as a parameter, and other quantities can be indicated as "true" or "false" to show whether the value to which it belongs was detected within an appropriate operating range. Therefore, this quantity for evaluating the effective operating range can be used as a trigger condition for the effectiveness of the cooler efficiency, which is a state index, and thus all determined values ​​for the cooler efficiency detected within the effective operating range can be stored as a state index and / or transmitted to the unit performing the evaluation.

[0054] In an air system, for example, the deviation of the air mass sensor value from the modeled mass flow rate through the engine when the exhaust gas recirculation valve is closed can be considered as a state index from the range of learned values ​​of the sensor and actuator. The modeled mass flow rate through the engine when the exhaust gas recirculation valve is closed is derived from the intake manifold pressure, intake manifold temperature, and engine speed. This modeled value can then be again compared with a measured value detected under predetermined effective operating conditions (e.g., a closed exhaust gas recirculation valve, steady operation, and air mass flow rate within a defined boundary) and optionally averaged. The identified deviation between the modeled air mass value and the measured air mass value can then be stored, transmitted, and evaluated as a state index. Furthermore, the currently measured air mass value can also be used as a unique state index. When multiple comparisons are performed under various different air masses, the progress of the deviation coefficient related to air mass can be evaluated. In such cases, a linear deviation, i.e., a constant coefficient over the air mass range, tends to suggest an error in the air mass sensor, while a deviation accompanied by an increasing coefficient over the air mass range tends to indicate intake manifold sooting due to deposits from, for example, the exhaust gas recirculation system or crankcase ventilation. If an unexpected decrease in the deviation coefficient occurs, an error due to a defective, severely soiled, or improperly cleaned air filter may be suspected.

[0055] Debounce time can be used in air systems, for example, when adjusting air mass, and can be evaluated as described above. Similarly, debounce time and the corresponding timer or debounce counter may be used when adjusting boost pressure and can be used as a condition indicator. Another area where debounce time may be used and a corresponding counter may exist is exhaust gas recirculation flow monitoring, which checks whether an excessively large or excessively small amount of exhaust gas is being directed through the return line, thereby allowing estimation of line suiting or valve malfunction. In this case as well, diagnostics can be performed separately for various different operating ranges, and the values ​​to which they belong can also be evaluated separately as condition indicators.

[0056] The error rate described above can be obtained, for example, from suiting identification in high-pressure exhaust gas recirculation, and this may optionally be given valid conditions for the operating range. Monitoring of the speed of the exhaust gas recirculation system can also be operated on an error rate basis, in which case, for example, positive and negative deviations of the error state can be used separately as indicators.

[0057] Finally, as adjustment parameters within the air system, parameters for boost pressure adjustment, air mass adjustment, exhaust gas recirculation adjustment, or position adjustment of various actuators (e.g., valves, throttle valves, exhaust gas flaps, turbocharger actuators, or swirl flaps) can be evaluated as state indicators.

[0058] All of the above-mentioned condition indicators for the air system can be evaluated individually and / or collectively to allow for the identification of faulty components from errors within the air system.

[0059] As is obvious, the aforementioned groups and subdivisions of state indicators, as well as the specific parameters mentioned, are merely illustrative examples. In principle, any value or parameter in a system can be used as a state indicator, and this state indicator may be influenced, at least partially, by a particular system state, and conversely, may be evaluated for the purpose of state evaluation.

[0060] As already explained, the values ​​assigned to the load counter class can also be used as state indicators.

[0061] For example, to detect the mechanical load on a component, the number of vibrational motions for various amplitudes can be detected and compared to predetermined values ​​for the component design. Such load values ​​may be indicated based on so-called rainflow diagrams or Wöhler lines, which are well known in the field. As a state indicator, the comparison with the component design can be used in the same way as the vibration counter itself. Detection of the minimum and / or maximum force acting on the component, or detection of the force acting for each defined running cycle, can also be used as a state indicator.

[0062] To detect thermal load, for example, the temperature distribution at points related to the durability of the components can be detected or modeled, and thus the temperature range over which each component operated and for what period can be detected. Similarly, the minimum and / or maximum temperatures during a running cycle can be evaluated. Similar to mechanical vibrations in the previous example, thermal oscillations, i.e., cycles consisting of heating and cooling over a specific temperature range, can also be detected. Since loads are generated on the components during such temperature fluctuations, these parameters can also be evaluated as state indicators for durability analysis.

[0063] Furthermore, quantities representing the vehicle's maintenance status, such as the detection of specific inspection processes, oil change intervals, or the performance of (partial) component replacements, are suitable as condition indicators.

[0064] For example, to monitor a turbocharger, the following load counter status indicators can be used as a basis: the duration of operation within a defined temperature range, e.g., the length of time the turbocharger operates at a temperature above a predetermined (high) threshold; similarly, the duration of operation within a defined rotational speed range, e.g., the length of time the turbocharger operates above a specific rotational speed threshold; the frequency of heavy loads occurring, such as high rotational speed or high temperature in a low-temperature turbocharger, i.e., the frequency of heavy load alternations; the number of temperature alternations exceeding a specific threshold for temperature differences; the frequency of motors stopping at high temperatures (risk of oil carbonization in bearings); the number or interval of oil changes; and the ratio of short-distance to long-distance driving for the vehicle.

[0065] As is obvious, the threshold to which the operating range belongs can be arbitrarily selected at the outset, and this threshold can be designed, for example, to include in the evaluation the operating range where particularly high loads on the components are expected. However, it is also possible to detect and evaluate the load counter values ​​described above for all operating ranges, or to use the load counter values ​​for selected high-load operating points as a general state indicator, in addition to the load counter values ​​for all operating ranges. The period to which the counter is relevant can also be appropriately selected, and in particular, the load counter can be considered over the entire duration of use of the components, but it is also possible to reset the load counter under predetermined conditions depending on the values ​​being used.

[0066] Next, the values ​​detected as state indicators can be aggregated and stored, for example, in the memory elements of the control device. In this case, data from various networked control devices can also be collected and stored centrally.

[0067] The values ​​thus collected can then be evaluated. For evaluation, in particular, the detected values ​​for state indicators can be transmitted to a suitable processing unit, which can provide sufficient computing power for the necessary evaluation of multiple parameters. Such a processing unit may include, for example, a remote server, a factory PC, a central control unit, or any other device having similar components capable of performing the steps described, such as a processor, a microcontroller, appropriate volatile and non-volatile memory elements, or similar components. Transmission to the processing unit can be done, for example, by wireless or wired communication connections, such as via a mobile wireless network and a corresponding interface available in the vehicle, or via cable connections during factory inspections. This data can be transmitted, for example, at predetermined times, at regular time intervals, or on call, or after each identification or calculation, for example, each time a corresponding diagnostic feature is calculated by the control unit. Values ​​can be locally stored as raw values ​​or in aggregated forms, for example, as minimum values, average values, and maximum values. These aggregated values ​​can then be transmitted, for example, at the end of a driving cycle or after predetermined driving sections (for example, every 100 km). The values ​​stored in the vehicle can then be optionally erased or reset after transmission.

[0068] The optimal transmission frequency and the number and selection of values ​​to be stored may also take into account the available memory capacity and available transmission means, such as the bandwidth and cost of data transmission. Furthermore, in order to determine the storage and transmission conditions, the frequency at which a particular state indicator is determined or calculated, and the level of accuracy or variability of the individual values ​​may also be taken into account. Since a particular operating state must exist over a specific period of time in order to provide valid results, state indicators that do not provide evaluable results for each running cycle may be defined. Other state indicators may provide multiple detected values ​​in a short period of time, and these values ​​may be available for evaluation as individual values ​​or in an aggregated form. Different transmission conditions may be defined for different state indicators or different subgroups of state indicators. For example, load counter transmission may occur over relatively long time intervals, while continuously detected diagnostic features may be transmitted more frequently. Furthermore, other conditions or trigger conditions may be in effect for transmission, and when an error is identified by a conventional diagnostic function within the control unit, it may be stipulated that all values ​​currently collected for a status index, or a certain number of other values ​​that may be related to this error, are immediately transmitted outside of the normal transmission cycle.

[0069] Next, in the system being evaluated, various evaluation steps from the subsequent evaluation steps can be executed individually or in combination.

[0070] In particular, state indicators relating to diagnostic features can be compared to predetermined thresholds. In this case, these thresholds may be the same thresholds used for the diagnostic function, and optionally, other boundary values ​​may be defined for the same parameters for probability-based evaluation. In addition to a simple check of whether the value is above or below the threshold, the interval from the detected value to each threshold can also be optionally identified, and for example, a weighting or evaluation index for possible defects can be derived from this interval (or the interval itself can be used as an evaluation index). If the interval to the threshold not yet reached for a defect is relatively large, it suggests that the probability of defect is relatively low, while on the other hand, if the deviation from the threshold is relatively small, it may indicate that the probability of defect is high. That is, by this direct or indirect comparison between a predetermined value and a parameter value, an evaluation index can be obtained for each state indicator based on the current parameter value, and the evaluation index may be, for example, a probability value, an interval value, or a similar key numerical value. These current evaluation indices do not yet need to be equivalent to one another and represent key numerical values ​​relating to the state indicator. The evaluation metrics can, for example, show probability values ​​for error or defect states, but they can also show intervals between parameter values ​​and boundaries, and intervals between the distribution of parameter values ​​and typical values ​​or similar major numerical values. As an interval, for example, a 3σ deviation from the fleet mean for a particular parameter can be identified. The identification of evaluation metrics may be set differently for each state metric.

[0071] It is also possible to compare the detected state indicators of individual systems, such as vehicles, with the values ​​for the same state indicators of other equivalent systems, for example, with a fleet of vehicles that similarly provide these values, at least partially. Here, since many of the systems that provide their own determined state indicators have a defect-free state, it can be assumed that the functionality or state of an individual system can be estimated from the statistical characteristic values ​​of the state indicators of several other systems (e.g., from the mean, standard deviation, frequency distribution, or other derived values ​​for each state indicator, respectively), and from comparisons with the state indicators of the individual system. For example, if one or more values ​​of an individual system are significantly deviated from the mean of the same parameter values ​​of several other systems, this may suggest a defect. Selectively, individual values ​​can also be compared with other values, and initially, the identified deviations can be evaluated using statistical methods. Here again, for example, an evaluation index for defects can be output in relation to the identified deviations between individual data and aggregated data from other systems.

[0072] By combining a defect threshold (for example, one specific to a diagnostic function or a status indicator) with a comparison to the values ​​of other systems that are considered defect-free on average, the detected values ​​can be better represented. The two comparisons allow the defect probability to be defined, for example, between 0% (a value corresponding to the average value of the remaining systems) and 100% (a value above the defect boundary of the diagnostic function), and similarly, a corresponding probability can be given for all values ​​that lie between these two points.

[0073] Furthermore, the temporal progression of one or more state indicators can also be used for evaluation. If a parameter changes slowly over time and / or over the distance traveled, this may suggest a slow drift. Such slow drifts are likely to persist and can be extrapolated accordingly, thereby identifying the probability of failure. In contrast, if a parameter changes abruptly and remains at a new value, this change can be optionally associated first with the driving conditions occurring at that point (e.g., high temperature, relatively long driving under full load). Otherwise, an unexpected abrupt change may suggest a spontaneous defect in the component, even if the original defect threshold for the diagnosis has not yet been reached.

[0074] Furthermore, various state indicators can be divided into multiple correlation groups related to the underlying system error. State indicators that have a relatively high correlation to the underlying system error are, in this case, placed in a different correlation group from state indicators that have a relatively low correlation to the underlying system error. Each correlation group is further assigned a correlation coefficient, which is larger for correlation groups with a relatively high correlation than for correlation groups with a relatively low correlation to the underlying system error. The state indicator is then multiplied by the correlation coefficient assigned to the correlation group.

[0075] The state indicators from a group of load counters can be compared to predetermined design targets for the design of the components. If the load counters have values ​​close to or above the design targets, similarly, the probability of failure may be higher compared to relatively low load counter values.

[0076] In particular, evaluating diagnostic features for components or subsystems in combination with load counters allows for improved diagnostics compared to evaluating only one of these groups.

[0077] The evaluation of state indicators can be expressed as an evaluation index for each detected value, representing an individual probability or a similar key numerical value relating to the evaluation. Then, from combinations of these evaluation indices, the overall probability for a defect in one or more components, or the overall probability for a defect in a subsystem consisting of multiple components, or the diagnostic probability for a diagnostic step can be determined. Combinations of evaluation indices can be identified, for example, from a stock of existing combinations. In this case, these existing combinations may be those obtained from evaluations of fleet data. Therefore, some combinations of evaluation indices may occur rarely or never in practice, while others may occur very frequently. Thus, the selection of evaluation index combinations may be limited to one of the frequently occurring combinations. Furthermore, individual evaluation indices may be weighted, resulting in certain deviations or errors being weighted more highly than others for a specific defect. Weighting coefficients can, in particular, be obtained from information from factory diagnostic feasibility analysis (DMA) or guided error search (GFS) that already exist before the start of system manufacturing. This is because the weighting coefficients are derived from empirical data, or they are determined in advance in a validation fleet.

[0078] Furthermore, it may be determined which state indicators are directly or indirectly related, or related to specific components, and this may be determined, for example, by grouping the state indicators into different groups, by providing weights, or by other means. The selection of these groups may be pre-defined and may be obtained based on general defect identification. For example, it may be determined which state indicators are affected by the failure of individual catalysts, in which case these state indicators can be evaluated together for catalyst defect identification. In this case, each state indicator may be assigned to any number of groups. For example, when defects or deterioration occur in several different components, changes in specific exhaust gas parameters may occur, and therefore the state indicators to which they belong are grouped according to these components. Nevertheless, these groups may differ in their overall composition, which allows for a more precise limitation of errors. Thus, changes in the parameter values ​​of state indicators unrelated to each component can be ignored.

[0079] Essentially, state indicators can also be evaluated using machine learning, for example, by one or more neural networks. Input data for the neural networks can include, for example, the detected state indicators themselves and / or evaluation metrics determined from threshold comparisons, comparisons with fleet values, and other specifics. Labeled training data for supervised machine learning can be formed by detecting state indicators for a system over a relatively long period and linking each value for a state indicator to error-free time or the effects that actually occurred. Additionally or selectively, unsupervised learning methods can be chosen to identify the structure in the deviations of multiple state indicators. In this case, the output value could be, for example, the defect probability for a particular component. Selectively or additionally, the current system state can be classified into multiple classes as an output value, in which case, for example, the error-free state, as well as any possible defects in a component or the system as a whole, might form one class.

[0080] The evaluation system can optionally incorporate further information to continuously improve the evaluation process.

[0081] For example, if further manual diagnostic steps are performed in a factory, the results of the diagnostics can be detected and fed back into the evaluation system. This allows the evaluation system's results to be reinforced, confirmed, or refuted by other indicators. In response, the defect probability in the overall evaluation or the weighting of individual condition indicators can then be adjusted. In a similar manner, defective components to be replaced can be selectively examined, and results that may include precise details of the defects can be fed back into the evaluation system. The relationship between condition indicators and defects can then be estimated, along with measurement data and condition indicators detected over time for these components.

[0082] Similarly, the values ​​of the condition indicators before the repair or inspection process can be compared with the values ​​of the condition indicators after the repair or inspection process. If the condition indicators change, i.e., if the effect of the repair is visible in the condition indicators, the assignment and weighting of the probabilistic evaluation may be reviewed. Conversely, if, for example, a component is replaced but the measured value does not change, a more detailed examination may be conducted to determine whether defects exist in other components. Furthermore, in such cases, the assignment of each condition indicator to a component or subsystem, or the weighting of the condition indicators, may be reviewed and modified.

[0083] If improvements are identified after repair, but further improvements or other indications of defects reappear within a relatively short period of time, a similar procedure may be used to check for defects in other components or complex errors in the interaction of multiple components.

[0084] Furthermore, the evaluation steps described above can essentially be performed directly on the vehicle, or on a stationary device to be monitored in this manner at the site. For example, if a sufficiently high-performance processing unit is available, such as a complex control device capable of performing such calculations, transmission to other processing units can be performed supplementarily, although this is not mandatory. When on-site evaluations are performed, any additional values ​​and parameters necessary for this, such as fleet data for comparison and updated thresholds, can be transmitted from a remote server or other unit to the on-site processing unit. For example, the collected fleet data can be evaluated on a remote server, from which current thresholds or weights for individual evaluations can be derived, and these derived thresholds or weights can then be sent to the on-site processing unit, where they can be used for evaluation.

[0085] In particular, the aforementioned know-how model, based on the experience of skilled factory workers and data determined during the verification phase, is transformed into a model or hybrid model based on know-how and fleet data. This allows for the refinement of the defect probability of individual components or diagnostic tools based on a series of data. Additionally, the temporal effects of individual condition indicators can also be incorporated into the consideration, making it possible to identify defects that first occur, for example, after many years of operation or after a certain number of kilometers, using an improved method.

[0086] Fleet data collected in similar systems and stored on remote servers or cloud memory, along with data collected in the system, can be used to divide the systems in the fleet and the systems to be examined into various groups. For example, data on emission levels for various types of exhaust gases (NOx, NH3, CO, HC, ...) can be collected, and systems can be assigned, for example, to low-emission groups and high-emission groups, particularly separately for each type of exhaust gas. If a system to be examined exhibits high emission values ​​and is therefore assigned to the high-emission system group, only fleet data from systems assigned to the same group can be used to determine the probability of defects and / or diagnostics, for example, when identifying combinations of evaluation indicators. Furthermore, by dividing into various groups, it can be determined which condition indicators or combinations of condition indicators, even if their values ​​are significant, do not affect underlying system errors or higher-level defects. For example, if a condition index or a combination of different condition indices exhibits significant values ​​in both low-emission and high-emission vehicles, it is unlikely that these condition indices or combinations are related to identifying an underlying system error or higher-level defect. Conversely, if significant values ​​of a condition index or a combination of different condition indices occur only in the high-emission vehicle group, these are particularly relevant to determining defect probabilities and / or diagnostic probabilities. Therefore, dividing the fleet's systems and the systems to be investigated into different groups, for example based on emission levels, improves the identification of the appropriate weighting of individual condition indices or combinations of evaluation indices to use.

[0087] Conversely, if sufficient fleet data is available, a model can be developed immediately, which will model, for example, emission levels and any other related system errors. Such models are created using machine learning or other AI applications. From this, for example, it is possible to identify in advance how changing values ​​of future state indicators may affect emission levels. This further allows for a better identification of relevant combinations of state indicators or evaluation indicators that determine the likelihood of defects and / or diagnostics. The temporal progression of one or more state indicators can also be extrapolated, thereby identifying future points in time when emission levels will be exceeded or system errors will occur, allowing for early warning messages to be displayed to system users.

[0088] In particular, a model is developed to identify the impact of replacing or repairing one or more components on condition indicators by incorporating data on repairs performed, i.e., the components ultimately replaced, and the impact of these repairs on, for example, emission levels or individual condition indicators. In combination with the aforementioned model for identifying emission levels based on condition indicators, it is possible to determine which one or more components should be replaced or repaired in order to shift, for example, the emission level or other system errors back into an acceptable range. In particular, to enable system repairs as quickly and at the lowest possible cost, for example, the cost of the components and / or the availability or delivery date of the components can be incorporated together.

Claims

1. A method for identifying defects in a system (100), Detecting parameter values ​​for a plurality of state indicators of the system (200, 210, 220), wherein the state indicators include measured parameters of the system (200) and / or parameters for the system (220) calculated or modeled from the measured parameters (200, 210, 220), For each of the aforementioned state indicators, the current evaluation indicator for the defect is identified based on a comparison between the detected parameter value for the state indicator and a predetermined reference value. Determining the defect probability for the at least one component of the system (100) based on a combination of the evaluation index for a predetermined group of state indices each assigned to at least one component, A method that includes this.

2. The method according to claim 1, further comprising determining a diagnostic probability for at least one diagnostic step based on a combination of the evaluation indicators for the predetermined group of the state indicators assigned to each of the at least one diagnostic step.

3. Identifying the current evaluation index for the aforementioned state index means that The parameter value is compared with one or more threshold values ​​set as the reference value. Comparing the parameter value with a reference value determined for the state index in one or more other systems, The parameter value is compared with a reference value detected from the statistical characteristic values ​​of multiple parameter values ​​for the state index in multiple other systems. The parameter value is compared with the parameter value of the system, which has been set as a reference value and detected at one or more preceding points in time. The method according to claim 1, comprising at least one of the following.

4. Identifying the current evaluation index for each of the aforementioned multiple status indicators is: Identifying the difference between the parameter value and the reference value, To define the evaluation index in relation to the identified difference, The method according to claim 1, including the method described in claim 1.

5. Identifying the current evaluation index for each of the aforementioned multiple status indicators is: The aforementioned status indicators are classified into multiple correlation groups, The evaluation index is identified by multiplying the defined evaluation index by a correlation coefficient in relation to the correlation group of the state index, The method according to claim 4, further comprising:

6. Determining the defect probability for at least one component of the system (100), and / or determining the diagnostic probability for at least one diagnostic step, Identifying combinations of the evaluation indicators based on existing combinations, Based on the identified combination of the evaluation indicators, the probability of a defect for at least one component of the system (100) and / or the probability of a diagnosis for at least one diagnostic step are determined. The method according to claim 1, including the method described in claim 1.

7. Determining the probability of a defect for at least one component of the system (100), and / or determining the probability of a diagnosis for at least one diagnostic step, Weighting each evaluation metric using weighting coefficients, Based on the weighted evaluation index, the probability of the defect for at least one component of the system (100) and / or the probability of the diagnosis for at least one diagnostic step are determined. The method according to claim 1, including the method described in claim 1.

8. Detecting the aforementioned parameter value means Obtaining the parameter value as the output signal of the sensor, The parameter value is obtained as the output value of a calculation module configured to calculate the parameter value from the input value. Receiving the parameter value via a communication connection. The method according to claim 1, comprising at least one of the following.

9. The method according to claim 1, wherein for at least one state index, a predetermined condition is checked to see if it has been met during the detection of the parameter value, and if it has not been met, the parameter value to which it belongs is discarded.

10. The method according to claim 1, wherein the state index includes a parameter representing a measured deviation from a predetermined normal state, and / or a parameter representing a measure of the load of at least one component of the system.

11. The method according to claim 1, further comprising outputting an instruction to the user when the defect probability for a component exceeds a predetermined threshold.

12. The aforementioned method, To obtain data on actual defects and / or replacements and / or diagnostic steps performed on the components of the aforementioned system, The data is compared with the determined defect probability for the component and / or the diagnostic probability for the diagnostic step, Based on the above comparison, the above specific adjustments to the evaluation index and / or the above determination of the defect probability and / or the above determination of the diagnosis probability are made. It further includes, The aforementioned adjustment is, Changing the assignment of the predetermined group to at least one status indicator, To change the calculation settings for identifying the aforementioned evaluation indicators, To change the calculation settings for identifying the aforementioned defect probability, Changing the weighting of the state indicators for identifying the defect probability, To change the calculation settings for determining the aforementioned diagnostic probability, Changing the weighting of the state indicators for determining the probability of the diagnosis. The method according to claim 1, comprising at least one of the following.

13. The method according to claim 1, wherein the parameter values ​​for one or more state indicators are detected over time and / or over time based on intervals.

14. The method according to claim 1, wherein the system is a vehicle (100).

15. A computing unit (110, 130) configured to carry out all steps of the method described in any one of claims 1 to 14.

16. A computer program that, when executed on a computing unit, causes the computing unit to perform all the steps of the method according to any one of claims 1 to 14.

17. A machine-readable storage medium storing the computer program described in claim 16.

Images