Method for estimating the failure of a vehicle component

Abstract

The invention relates to a method for estimating the failure of a vehicle component, wherein the component failure is estimated based on diagnostic data and type data. The invention also relates to a device and a computer program.

Description

State of the art

[0001] Vehicles send fault or diagnostic data, for example in the form of so-called Diagnostic Trouble Codes (DTCs), to backend servers, such as those operated by vehicle manufacturers. Diagnostic Trouble Codes (DTCs) are standardized codes generated by vehicles when a problem or fault is detected in the vehicle system. These codes are generated by the vehicle's on-board diagnostic (OBD) systems and serve to help mechanics and diagnostic equipment identify and diagnose problems. Each code corresponds to a specific problem or fault in the vehicle, which facilitates diagnosis and repair. The code "P0300," for example, represents a general cylinder fault in the engine. Diagnostic Trouble Codes can indicate serious problems, but are often also transient, meaning they only occur temporarily.After a certain period of time, these codes are no longer issued because the underlying problem has resolved itself, or they are even actively deleted if the cause no longer exists. Diagnostic Trouble Codes related to the mass airflow sensor or exhaust gas recirculation are particularly common and transient. Examples include "P0101," a DTC indicating a dirty mass airflow sensor or a loose connection, and "P0135," which can be caused by an aging oxygen sensor. Diagnostic Trouble Codes can be transmitted to a backend system via the vehicle's communication device or read out in workshops using diagnostic testers.

[0002] When a vehicle is repaired in a workshop, data about the repair is often entered into a database. This is particularly the case when the repair involves replacing a vehicle component. This data is also available in a backend system. Repair data is also referred to as claim data.

[0003] A systematic analysis of the failure probability of a component of a vehicle of a given vehicle type based on Diagnostic Trouble Codes is complex and often not very precise, since, especially when considering Diagnostic Trouble Codes of vehicle fleets, statistically insignificant correlations are often difficult to find in the large amount of data. Disclosure of the invention

[0004] The inventive method for estimating the failure of a vehicle component has the advantage that the component failure is estimated based on diagnostic and type data. The diagnostic data advantageously consists of fault codes, in particular Diagnostic Trouble Codes. Type data can be data that describes a vehicle type in more detail. Type data can therefore be, for example, data that, individually or in combination, describe the following attributes: vehicle type according to manufacturer's designation, engine variant, transmission variant, traction battery type, injector type, equipment variant, variant of an engine control unit, software version, national sub-variant of the vehicle type, and model year. Considering both diagnostic and type data advantageously enables significantly more precise analyses. Statistically weak correlations can thus be identified earlier.

[0005] It is advantageous if the assessment of the vehicle component's failure is based on a time series derived from the diagnostic data. This time series can, in particular, be a chronological sequence of the occurrence of fault data, especially Diagnostic Trouble Codes, within the diagnostic data. The diagnostic data is advantageously collected during diagnostic sessions.

[0006] It is advantageous if only diagnostic data correlated with predefined type data is used to create the time series. In particular, a filter can be defined before creating the time series, allowing, for example, only data that includes a predefined combination of diagnostic and type data. The mapping of diagnostic data to type data can advantageously take place in a backend, where, for example, type data from a database can be assigned to each vehicle, which can be identified, in particular, by its Vehicle Identification Number (VIN). Therefore, all desired type data can be advantageously assigned to each transmitted set of diagnostic data via the VIN and the database.

[0007] It is advantageous to assess a component failure as imminent if the time series meets a predefined criterion, particularly if the temporal density of diagnostic data within the time series exceeds a threshold. Alternatively, the temporal sequence of occurrence of various fault codes, especially Diagnostic Trouble Codes (DTCs), can be analyzed within the time series. Experience shows that certain DTCs typically appear first, before being followed by other DTCs that indicate a kind of "subsequent fault."

[0008] It is advantageous to use diagnostic data from a large number of vehicles, where the vehicles are linked via their type data. In particular, using a large number of vehicles allows for the identification of a pattern in the time series that is characteristic of an impending failure. Specifically, graded threshold values ​​can be learned from the diagnostic data of these numerous vehicles and correlated with failure probabilities.Thus, an initial density of Diagnostic Trouble Codes in a time series of an individual vehicle can be correlated with an initial probability of failure of the corresponding component, for example, within the next 500 km. A second, particularly higher, density of Diagnostic Trouble Codes in the time series of the individual vehicle can, for example, be correlated with a second probability of failure of the corresponding component within the next 500 km, whereby the second probability of failure can be particularly higher than the first probability of failure.

[0009] It is advantageous if the diagnostic data consists of Diagnostic Trouble Codes.

[0010] It is advantageous if the type data includes component data and / or location data. Location data can, in particular, specify the intended areas of operation for vehicles over a wide area. Specifically, the location data could be, for example, "Europe," "US," "Australia," or similar. Analyzing the type data advantageously allows for the significantly faster and more precise identification of a fault that occurs only in a sub-variant of the vehicle type or only in a combination of specific installed components than would be possible by considering all diagnostic data for the vehicle type.

[0011] The procedure described above can be refined, in particular, by a so-called actual / non-actual comparison. For this purpose, a data pattern is first identified that exhibits a high correlation with the replacement of a component. This is achieved by comparing fault data from diagnostic sessions associated with a component replacement—i.e., for which claim data is available—with fault data from diagnostic sessions for which no claim data is available. A data pattern exhibits a high correlation with a component replacement if the data pattern, for example, a combination of diagnostic trouble codes, occurs significantly more frequently in connection with claim data than without. The data pattern is then advantageously used as a filter when creating the time series.

[0012] An advantageous feature is a device configured to carry out the method according to the invention. A further advantage is a computer program that causes a computing unit to carry out the method according to the invention when the computer program is running on the computing unit.

[0013] An embodiment of the present invention will now be explained in more detail with reference to the accompanying drawing. The drawing shows: Brief description of the drawing Fig. 1 a schematic representation of a sequence of the process according to the invention. embodiment of the invention

[0014] Fig. Figure 1 shows a schematic representation of a sequence of steps in the method according to the invention. The presented embodiment begins in step 100.

[0015] In step 100, an actual / non-actual comparison is performed. For this purpose, diagnostic data from a large number of vehicles of the same vehicle type are compared with claim data to determine which combination of fault data, especially the Diagnostic Trouble Codes, correlates with a component replacement. For example, it can be determined that the occurrence of Diagnostic Trouble Codes "A" and "B" is significantly more frequent in temporal relation to a component replacement than the occurrence of the same Diagnostic Trouble Codes without a component replacement. The combination of Diagnostic Trouble Codes "A" and "B" is therefore a suitable indicator for determining the probability of component failure. Step 110 is then performed.

[0016] In step 110, a vehicle-type-specific filter is defined based on the actual / non-actual comparison performed in step 100. This filter is, in particular, a combination of fault data, especially Diagnostic Trouble Codes, and type data that characterizes the vehicle type. The combination of Diagnostic Trouble Codes could, for example, be a combination of Diagnostic Trouble Code "A" and Diagnostic Trouble Code "B". The type data could, for example, be a vehicle model year, such as "2019", and a specific vehicle type, such as "Vehicle Type_XYZ". Step 120 is then performed. The use of type data significantly improves the predictive power of the procedure. A fault that occurs only in connection with, for example, a specific injection system of the vehicle type can thus be isolated and identified more effectively.

[0017] In step 120, the filter defined in step 110 is used to generate a time series of error data, specifically Diagnostic Trouble Codes. Step 130 is then performed.

[0018] In step 130, threshold values ​​are derived from the time series generated in step 120. In particular, the temporal density of the occurrence of Diagnostic Trouble Codes in the time series can be used. In other words, a threshold value is set where the Diagnostic Trouble Codes occur statistically frequently within a predefined time interval. Advantageously, this results in several hierarchical threshold values ​​that characterize different failure probabilities. Step 140 then follows.

[0019] In step 140, the threshold values ​​determined in step 130 are applied to a time series of diagnostic trouble codes for an individual vehicle of the type under consideration. Here, too, the time series is created using filtering. In other words, starting in step 140, the individual vehicle is monitored for an impending component failure. Step 150 then follows.

[0020] Step 150 checks whether a threshold has been exceeded. If so, step 160 follows. If no threshold has been exceeded, step 150 is repeated. The time series can be updated during this step.

[0021] In step 160, the individual vehicle is called to a workshop due to an impending component failure. Alternatively, an error message or other notification can be issued to the vehicle's driver or a fleet manager.

Claims

[1] Method for estimating the failure of a component of a vehicle, characterized by that the assessment of the failure of the vehicle component is based on diagnostic data and type data. [2] Method according to claim 1, characterized by , that the estimation of the failure of the vehicle component is based on a time series determined from the diagnostic data. [3] Method according to claim 2, characterized by , that only diagnostic data that are correlated with predefined type data are used to create the time series. [4] Method according to claim 2 or 3, wherein a failure of the component is assessed as imminent when the time series meets a predefinable criterion, in particular when a temporal density of the diagnostic data within the time series exceeds a threshold. [5] Method according to any one of claims 1 to 4, characterized bythat diagnostic data from a large number of vehicles are used, with the large number of vehicles being linked together via the type data. [6] Method according to any one of claims 1 to 5, characterized by that the diagnostic data consists of Diagnostic Trouble Codes. [7] Method according to any one of claims 1 to 6, characterized by that the type data consists of component data and / or location data. [8] Device configured to carry out the method according to any one of claims 1 to 7. [9] Computer program that causes a computing unit to perform the method according to any one of claims 1 to 6 when the computer program is running on the computing unit.

Images