Onboard fault diagnosis procedures for vehicles

Abstract

A fault diagnosis procedure for detecting faulty components in a vehicle, wherein, for at least one identifiable fault cause of a faulty component, a system parameter of the vehicle that is influenced by the fault cause is determined and compared with a target value of the system parameter, and a determined signal deviation between the determined system parameter and the target value of the system parameter is compared with a stored fault characteristic signal deviation, and if the determined signal deviation matches the fault characteristic signal deviation, the fault cause and / or faulty component associated with the fault characteristic signal deviation is uniquely identified, wherein the entire development process for the fault diagnosis procedure essentially consists of the following six individual steps, which are executed sequentially: - Selection of faults to be diagnosed, whereby in this first step the various components of the vehicle are analyzed in such a way that the possible causes of faults are listed for each component; - Structural analysis, whereby in this second step it is checked whether there is a possibility in principle of model-based diagnostics of the system; - Creation of a nominal model, whereby in this third step, so-called nominal models are created for those system parameters that are used and evaluated for the identification of a faulty system component, in which a normal operation of the system component is depicted; - Sensitivity analysis, in which, in this fourth step, defined causes of failure are imprinted into the system and their effects on the relevant system size(s) are considered using the nominal model, thereby identifying a unique assignment of a cause of failure or a faulty system component to its effect on the relevant system size; - Creation of fault models, whereby in this fifth step the knowledge about the effect is stored in a reference source onboard in the vehicle, whereby, based on this reference source, a statement can be made at any time valid for the diagnosis as to whether this cause of fault is present or not and - Integration into a diagnostic system, whereby in this sixth step the developed diagnostic procedure is transferred to a corresponding target control unit in the vehicle.

Description

[0001] The invention relates to a fault diagnosis method for the unambiguous identification of fault causes and / or faulty components in the vehicle.

[0002] With today's systems, identifying the causes of faults or faulty components within the vehicle, i.e., "onboard," is very difficult. While functions, signal paths, and sometimes even power stages are typically monitored, the precise location of a fault cannot usually be diagnosed onboard. Therefore, complex diagnostic and testing routines are required for the unambiguous identification of faults and their causes, as well as for the necessary corrective measures.

[0003] Many faults caused by defective components, however, only occur sporadically and are often not reproducible during a workshop visit. Due to the increasing distribution of functions across various control units, clearly identifying faults (and thus the defective components) is becoming increasingly difficult, even for trained specialists. This can lead to high costs due to lengthy diagnostic procedures and potentially even the incorrect replacement of perfectly functioning components.

[0004] DE 197 42 446 B4 discloses a fault diagnosis method for detecting faulty components of a technical system. This method is based on the fact that, in the case of a faulty component, a process variable of the system, designated as relevant to the fault, changes its state from a fault-free state to a fault state. This change of state allows conclusions to be drawn about one or more system components that may be faulty. The diagnostic module can be configured to display the system components identified as potentially faulty during the diagnostic process, ordered according to their probability of failure. This enables service personnel to examine the various causes of faults according to their probability of failure. However, this method does not allow for the unambiguous identification of a single cause of a fault.

[0005] For further information on the state of the art, reference is made to DE 10 2006 031 710 A1 and DE 103 19 493 A1.

[0006] The object of the invention is to provide a fault diagnosis method which enables the unambiguous detection and identification of a faulty system component in the vehicle.

[0007] This problem is solved by a method according to claim 1 and a diagnostic system according to claim 4. Advantageous further developments result from the dependent claims.

[0008] The basic idea of ​​the invention is to provide a fault diagnosis method for vehicle fault diagnosis which, in the event of a fault, clearly locates or identifies the smallest component that needs to be replaced. This allows necessary maintenance measures to be determined and subsequently implemented via offboard diagnostics without additional testing steps.

[0009] The fault diagnosis method according to the invention for use in vehicles is characterized in that, for at least one identifiable faulty component or cause of fault, a system parameter of the vehicle (measured variable or a parameter calculated from several measured variables) that is influenced by the faulty component due to the cause of the fault is determined and compared with a target value of the system parameter. If the deviation between the determined actual system parameter and the target value of the system parameter corresponds to a previously defined and stored fault-characteristic signal deviation, the cause of the fault or the faulty component can be unambiguously identified. This method allows faulty system components to be identified solely through the existing electronics in the vehicle or the existing vehicle electrical system.No additional sensors or special, diagnostic-optimized actuators are required, thus incurring no additional costs. The identification of faulty system components is based on the evaluation of system parameters that are already available or can be derived from existing sensor or system parameters. To uniquely identify a fault cause or a faulty component, at least one fault-characteristic signal deviation—that is, a characteristic change in at least one system parameter—must be determined and stored beforehand for each fault cause and / or faulty component to be identified. This characteristic signal deviation is unique for exactly one identifiable fault cause.

[0010] Faulty components are considered identifiable when the fault or its cause leads to a change in at least one system parameter that is characteristic of that fault. If the change in the system parameter, or the deviation of the system parameter from a determined target value caused by the faulty system component, is characteristic of this fault, the corresponding fault can be uniquely identified based on this deviation.

[0011] Advantageously, the target value of the system variable is determined using a predefined nominal model, where the nominal model represents the normal operation of the system component. Signals and quantities that are independent of the presence of a fault in the system component are used as input variables for the mathematical nominal model.

[0012] The following description clarifies the considerations on which the fault diagnosis procedure is based and how a clear identification of a faulty component is made possible using this procedure.

[0013] The entire development process for the fault diagnosis method according to the invention essentially consists of six individual steps, which must be carried out sequentially. 1. Selection of the faults to be diagnosed

[0014] As a first step, the various components of the vehicle are analyzed to identify the potential causes of failure for each component. This analysis also defines which types of failures or their causes are actually considered failures. For example, in the intake manifold system of an internal combustion engine, a hole in the intake manifold could be a cause of failure. However, it is important to note that a qualitative statement about the cause of failure is insufficient. A quantitative statement about the cause of failure must also be made. For instance, a hole in the intake manifold only leads to a failure if its diameter is approximately 5 mm or greater. If the hole is smaller, it does not cause a malfunction in the system; therefore, there is no need to identify such a hole.

[0015] Further criteria for defining a fault or fault cause include the system's diagnostic capability, the probability of occurrence, and the economic impact. Diagnostic capability involves comparing the possible fault causes with the diagnostic capabilities available in the control unit. This allows for the determination of which fault causes can already be detected by the existing system and which are not yet covered. For example, engine control units typically lack a diagnostic capability to detect a hole in the intake manifold. The fault causes identified in this step can then be evaluated with regard to their probability of occurrence and the associated costs. The result of this step is a selection of defined fault causes to be diagnosed. 2. Structural Analysis

[0016] This step verifies whether the system is fundamentally capable of model-based diagnostics. This is demonstrated by determining the number of analytical redundancies available within the system. Analytical redundancy exists when a variable or system parameter can be determined via two independent paths – for example, using other system parameters. This allows for a direct comparison between the calculated results of the system parameters. For instance, the measured fresh air mass flow rate from the air mass sensor can be compared with the air mass flow rate into the intake manifold calculated from the intake manifold pressure and the reservoir model.

[0017] Following the structural analysis, an objective statement about the diagnosability of the considered causes of errors or the faulty components is possible; that is, a statement can be made as to whether a model-based diagnosis is possible or not. 3. Creation of a nominal model

[0018] In the next step, nominal models are created for the system parameters used and evaluated to identify a faulty system component. These models depict the normal operation of the system component and are required for the sensitivity analysis (see section 4). The structural analysis, which lists at least the necessary relationships between the relevant signals and system parameters, can serve as the basis for creating the nominal model. However, unlike the structural analysis, creating a nominal model requires knowledge of the underlying mathematical relationships. The mathematical equations and the relationships between the signals can then be represented in a modeling program such as MATLAB / Simulink. It is important to ensure that the significant influences are included and that the input parameters are not affected by the faults under consideration.The causes of errors depend on the model. Since a model can only ever be sufficiently accurate compared to reality, it is always necessary to check the model to determine whether a particular effect needs to be taken into account or not. 4. Sensitivity analysis

[0019] In sensitivity analysis, defined root causes of failure are imprinted on the system, and their effects on the relevant system variable(s) are examined using the nominal model. This aims to identify a clear correlation between a root cause or a faulty system component and its effect on the system or the relevant system variable. To unambiguously attribute the effects to a root cause, it is insufficient to simply detect a deviation from the nominal behavior, i.e., a deviation in the relevant system variable when a failure occurs. Rather, the changes in the system variables must exhibit a characteristic that is unique to the cause. Only in this way is it possible to make a definitive statement about a root cause. The underlying principle is that each root cause under consideration affects the individual system variables differently and exhibits its own unique characteristics. For example,If the differential pressure sensor in the intake manifold exhibits a drift in its measured value (a system parameter) from the target value, this could be due to a change in the diaphragm or a small hole in the intake manifold. The sensor reading behaves the same regardless of the cause of the fault, making a clear identification impossible when the drift is detected. However, it would be possible to compare another system parameter with the target value in addition to the differential pressure sensor reading. Based on the deviation of both actual system parameters from the target values ​​calculated using a nominal model, a characteristic deviation could be determined, thus enabling a clear identification of the fault cause. Alternatively, the characteristic deviation of an alternative parameter could also be used. 5. Creation of the error models

[0020] After the previous steps have determined the effects of the fault on the system, the next step involves storing this knowledge in an onboard reference source within the vehicle. This reference source allows for a determination at any given time relevant to the diagnosis as to whether or not the fault is present. The type of reference source can be selected depending on the chosen diagnostic method. For example, qualitative model-based diagnostics utilize fault models that depict the behavior of the faulty system. If the system's behavior matches the fault model, the fault can be clearly identified.

[0021] In contrast to qualitative model-based diagnosis, quantitative model-based diagnosis uses a nominal model to identify a characteristic deviation of the system. This deviation is a difference formed between a characteristic quantity of the system (e.g., sensor value) and the corresponding value of the nominal model – also called the residual. 6. Integration into a diagnostic system

[0022] Finally, the developed diagnostic procedure is transferred to the corresponding target control unit in the vehicle. This transfer can be done in various ways. For example, the developed algorithm can be directly translated into code readable by the control unit. Another possibility is exporting the code from a simulation program such as MatLab / Simulink and integrating the exported code into the control unit software.

[0023] The advantage of the onboard fault diagnosis method according to the invention lies in the fact that, in the event of a fault, the cause of the fault can be identified down to the smallest replaceable unit "onboard" and is quickly accessible to service personnel by means of an entry in a designated fault memory. Based on the memory entry, the defective component can be identified directly and unambiguously. This significantly reduces high warranty costs and the number of repairs required.

Claims

A fault diagnosis procedure for detecting faulty components in a vehicle, wherein, for at least one identifiable fault cause of a faulty component, a system parameter of the vehicle that is influenced by the fault cause is determined and compared with a target value of the system parameter, and a determined signal deviation between the determined system parameter and the target value of the system parameter is compared with a stored fault characteristic signal deviation, and if the determined signal deviation matches the fault characteristic signal deviation, the fault cause and / or faulty component associated with the fault characteristic signal deviation is uniquely identified, wherein the entire development process for the fault diagnosis procedure essentially consists of the following six individual steps, which are executed sequentially: - Selection of faults to be diagnosed,in this first step, the various components of the vehicle are analyzed to identify the possible causes of failure for each component; - Structural analysis, in this second step, it is checked whether model-based diagnostics of the system are possible in principle; - Creation of a nominal model, in this third step, so-called nominal models are created for those system parameters that are used and evaluated for the identification of a faulty system component, in which normal operation of the system component is depicted; - Sensitivity analysis, in this fourth step, defined causes of failure are programmed into the system and their effects on the relevant system parameter(s) are considered with the help of the nominal model.which identifies a clear assignment of a fault cause or a faulty system component to its effect on the relevant system parameter; - Creation of fault models, whereby in this fifth step the knowledge about the effect is stored in a reference source onboard in the vehicle, whereby, based on this reference source, a statement can be made at any time valid for the diagnosis as to whether this fault cause is present or not; and - Integration into a diagnostic system, whereby in this sixth step the developed diagnostic procedure is transferred to a corresponding target control unit in the vehicle. Fault diagnosis method according to claim 1, characterized in that, depending on the identified cause of the fault and / or the faulty component, a fault-specific memory entry is made in the vehicle. Fault diagnosis method according to one of the preceding claims, characterized in that for each identifiable fault cause of a faulty component exactly one unique fault-characteristic signal deviation is stored. Diagnostic system with integrated fault diagnosis procedure according to one of the preceding claims, which is transferred to a target control unit in the vehicle, whereby in the event of faults occurring the cause of the fault can be identified down to the smallest replaceable unit onboard and is quickly accessible to service personnel by means of an entry in a designated fault memory.