Vehicle-mounted controller fault prediction method and device, electronic equipment and medium
The vehicle controller fault prediction model, which is trained through multi-class processing, enables real-time prediction and prevention of vehicle faults, solves the problem of insufficient dynamic adaptation capability in the existing technology, and improves the safety of the vehicle controller.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- AUTOLINK INFORMATION TECHNOLOGY CO LTD
- Filing Date
- 2026-03-23
- Publication Date
- 2026-05-12
AI Technical Summary
Existing vehicle fault diagnosis solutions are difficult to adapt to vehicle configuration upgrades and new fault modes, lack dynamic adaptation capabilities, and are difficult to predict fault occurrences and take preventive measures in advance.
By processing fault data in multiple categories, a fault prediction model for the vehicle controller is trained and validated, deployed to the vehicle controller for real-time fault prediction, and processed based on the prediction results. This model adapts to various scenarios, predicts faults in advance, and takes preventative measures.
It improves dynamic adaptability, enabling early detection and preventative handling of faults to avoid their occurrence and protect the vehicle controller.
Smart Images
Figure CN122018489A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of data processing technology, and more specifically, to a method, apparatus, electronic device, and medium for predicting faults in vehicle controllers. Background Technology
[0002] The background technology for fault detection in vehicle controllers has primarily evolved around the core need of how to detect and locate anomalies in automotive electronic control units (ECUs) in real time and accurately. Its evolution is closely related to the increasing complexity of automotive electronic and electrical architecture.
[0003] Currently, fault prediction models can be constructed using big data, deep learning, and data mining to monitor vehicle faults, thereby improving the accuracy and speed of vehicle fault diagnosis and addressing the lack of experience among after-sales maintenance personnel.
[0004] However, existing vehicle fault diagnosis solutions and fault prediction models are difficult to adapt to scenarios such as vehicle configuration upgrades and new fault modes. Their dynamic adaptation capabilities are generally limited, and they are difficult to predict the occurrence of faults in advance. Furthermore, they lack preventive measures for the controller before faults occur. Summary of the Invention
[0005] In view of this, the purpose of this application is to provide a method, device, electronic device and medium for predicting vehicle controller faults. The method trains and verifies a basic vehicle controller fault prediction model by processing fault data of multiple categories to obtain a final vehicle controller fault prediction model. The model is then deployed to the vehicle controller to perform real-time fault prediction on the vehicle controller. Based on the fault prediction results, the faults are processed. This method can adapt to various scenarios, improve dynamic adaptability, and can predict the occurrence of faults in advance. Furthermore, it can prevent and process the controller in advance before the fault occurs, thereby avoiding the occurrence of faults and protecting the vehicle controller.
[0006] In a first aspect, embodiments of this application provide a method for predicting faults in an on-board controller, the method comprising: Fault data of the vehicle controller is acquired, and the fault data is processed in multiple categories to obtain processed fault data; wherein, the multiple categories of processing include at least data cleaning, data labeling, data augmentation, and data synthesis. Based on the processed fault data, the preset basic vehicle controller fault prediction model is trained and verified to obtain the final vehicle controller fault prediction model. The vehicle controller fault prediction model is sent to the vehicle controller and deployed based on a preset deployment method; The on-board controller fault prediction model is used to predict real-time fault data of the on-board controller, and the fault is processed based on the fault prediction results.
[0007] In one possible implementation, the multi-category processing of the fault data includes: Locate the missing data in the fault data, determine the filler data that matches the missing data, and replace the missing data with the filler data; Invalid data in the fault data is identified based on a preset verification mechanism, and the invalid data is removed to obtain cleaned fault data.
[0008] In one possible implementation, the multi-category processing of the fault data includes: Based on a preset classification model, the fault data is aggregated into multiple fault categories to obtain the corresponding target classification model; wherein, the target classification model represents the classification model under the multiple fault categories; The fault data is labeled based on the target classification model to align the fault data with the corresponding fault categories, thus obtaining labeled fault data.
[0009] In one possible implementation, the multi-category processing of the fault data includes: Add low-intensity Gaussian white noise to the labeled fault data and determine the target type of fault data in the fault data; Set a randomly selected range, and scale the fault data of the target type based on the randomly selected range to obtain enhanced fault data.
[0010] In one possible implementation, the multi-category processing of the fault data includes: The labeled fault data and the enhanced fault data are input into the preset first recurrent neural network model to generate the corresponding true latent vectors; The preset Gaussian white noise is input into the preset second recurrent neural network model to generate the corresponding pseudo-hidden vector; The true and false latent vectors are input into a preset third recurrent neural network model to distinguish between them, and the final true and false latent vectors are input into a preset fourth recurrent neural network model to obtain the synthesized fault data.
[0011] In one possible implementation, training and validating a preset basic vehicle controller fault prediction model based on the processed fault data includes: The synthesized fault data is randomly split into a first proportion of first fault data and a second proportion of second fault data; The basic vehicle controller fault prediction model is fine-tuned based on the first fault data. The accuracy of the vehicle controller fault prediction model is verified based on the second fault data. If the accuracy of the model is greater than the preset accuracy threshold, the second fault data is input into the vehicle controller fault prediction model for further training to obtain the final vehicle controller fault prediction model.
[0012] In one possible implementation, the method further includes: In response to receiving an update instruction from the manufacturer, it determines whether the update information and model accuracy of the current vehicle controller fault prediction model conform to the update instruction, and feeds it back to the manufacturer for secondary confirmation. In response to receiving an update confirmation instruction from the manufacturer, the vehicle controller fault prediction model is packaged into a model installation package in the target format, and the vehicle is determined to be in a dormant state within a preset update time. In response to the vehicle being online, the model installation package is downloaded via the vehicle, and an update is performed based on the model installation package.
[0013] Secondly, embodiments of this application also provide a vehicle controller fault prediction device, the device comprising: The processing module is used to acquire fault data from the vehicle controller and perform multi-category processing on the fault data to obtain processed fault data; wherein, the multi-category processing includes at least data cleaning, data labeling, data augmentation, and data synthesis. The acquisition module is used to train and verify the preset basic vehicle controller fault prediction model based on the processed fault data to obtain the final vehicle controller fault prediction model. The deployment module is used to send the vehicle controller fault prediction model to the vehicle controller and deploy it based on a preset deployment method. The prediction module is used to predict the real-time fault data of the vehicle controller through the vehicle controller fault prediction model, and to process the fault based on the fault prediction results.
[0014] In one possible implementation, the processing module is specifically used for: Locate the missing data in the fault data, determine the filler data that matches the missing data, and replace the missing data with the filler data; Invalid data in the fault data is identified based on a preset verification mechanism, and the invalid data is removed to obtain cleaned fault data.
[0015] In one possible implementation, the processing module is specifically used for: Based on a preset classification model, the fault data is aggregated into multiple fault categories to obtain the corresponding target classification model; wherein, the target classification model represents the classification model under the multiple fault categories; The fault data is labeled based on the target classification model to align the fault data with the corresponding fault categories, thus obtaining labeled fault data.
[0016] In one possible implementation, the processing module is specifically used for: Add low-intensity Gaussian white noise to the labeled fault data and determine the target type of fault data in the fault data; Set a randomly selected range, and scale the fault data of the target type based on the randomly selected range to obtain enhanced fault data.
[0017] In one possible implementation, the processing module is specifically used for: The labeled fault data and the enhanced fault data are input into the preset first recurrent neural network model to generate the corresponding true latent vectors; The preset Gaussian white noise is input into the preset second recurrent neural network model to generate the corresponding pseudo-hidden vector; The true and false latent vectors are input into a preset third recurrent neural network model to distinguish between them, and the final true and false latent vectors are input into a preset fourth recurrent neural network model to obtain the synthesized fault data.
[0018] In one possible implementation, the acquisition module is specifically used for: The synthesized fault data is randomly split into a first proportion of first fault data and a second proportion of second fault data; The basic vehicle controller fault prediction model is fine-tuned based on the first fault data. The accuracy of the vehicle controller fault prediction model is verified based on the second fault data. If the accuracy of the model is greater than the preset accuracy threshold, the second fault data is input into the vehicle controller fault prediction model for further training to obtain the final vehicle controller fault prediction model.
[0019] In one possible implementation, the device further includes: The first judgment module is used to respond to the update instruction received from the manufacturer, determine whether the update information and model accuracy of the current vehicle controller fault prediction model meet the update instruction, and feed back to the manufacturer for secondary confirmation. The second judgment module is used to package the vehicle controller fault prediction model into a model installation package in the target format in response to receiving the update confirmation instruction from the manufacturer, and to determine that the vehicle is in a dormant state within a preset update time. An update module is used to download the model installation package through the vehicle and perform an update based on the model installation package when the vehicle is online.
[0020] Thirdly, embodiments of this application provide an electronic device, including: a processor, a storage medium, and a bus, wherein the storage medium stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the storage medium via the bus, and the processor executes the machine-readable instructions to perform the steps of the vehicle controller fault prediction method as described in any of the first aspects.
[0021] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the vehicle controller fault prediction method according to any one of the first aspects.
[0022] This application provides a method, apparatus, electronic device, and medium for predicting vehicle controller faults. It acquires fault data from the vehicle controller, performs multi-category processing on the fault data to obtain processed fault data, trains and validates a preset basic vehicle controller fault prediction model based on the processed fault data, obtains a final vehicle controller fault prediction model, distributes the vehicle controller fault prediction model to the vehicle controller, and deploys it according to a preset deployment method. The vehicle controller fault prediction model predicts real-time fault data of the vehicle controller, and handles faults based on the prediction results. This application, by training and validating a basic vehicle controller fault prediction model using multi-category processed fault data to obtain a final vehicle controller fault prediction model, and then deploying the model to the vehicle controller, enables real-time fault prediction of the vehicle controller and fault handling based on the prediction results. This approach is adaptable to various scenarios, improves dynamic adaptability, and allows for early prediction of fault occurrence and preventative handling of the controller before faults occur, thereby avoiding faults and protecting the vehicle controller.
[0023] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description
[0024] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0025] Figure 1 This is a flowchart of the vehicle controller fault prediction method provided according to the embodiments of this application; Figure 2 This is a schematic diagram of the structure of the vehicle controller fault prediction device provided in the embodiments of this application; Figure 3 This is a schematic diagram of the structure of an electronic device provided according to an embodiment of this application. Detailed Implementation
[0026] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the accompanying drawings in this application are for illustrative and descriptive purposes only and are not intended to limit the scope of protection of this application. Furthermore, it should be understood that the schematic drawings are not drawn to scale. The flowcharts used in this application illustrate operations implemented according to some embodiments of this application. It should be understood that the operations in the flowcharts may not be implemented in sequence, and steps without logical contextual relationships may be reversed or implemented simultaneously. In addition, those skilled in the art, guided by the content of this application, may add one or more other operations to the flowcharts, or remove one or more operations from the flowcharts.
[0027] Furthermore, the described embodiments are merely some, not all, of the embodiments of this application. The components of the embodiments of this application described and illustrated herein can typically be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.
[0028] It should be noted that the term "comprising" will be used in the embodiments of this application to indicate the presence of the features declared thereafter, but does not exclude the addition of other features.
[0029] Considering the background technology of vehicle controller fault detection, its development has mainly revolved around the core need of how to detect and locate anomalies in automotive electronic control units (ECUs) in real time and accurately. Its evolution is closely related to the increasing complexity of automotive electronic and electrical architecture.
[0030] Currently, fault prediction models can be constructed using big data, deep learning, and data mining to monitor vehicle faults, thereby improving the accuracy and speed of vehicle fault diagnosis and addressing the lack of experience among after-sales maintenance personnel.
[0031] However, existing vehicle fault diagnosis solutions and fault prediction models are difficult to adapt to scenarios such as vehicle configuration upgrades and new fault modes. Their dynamic adaptation capabilities are generally limited, and they are difficult to predict the occurrence of faults in advance. Furthermore, they lack preventive measures for the controller before faults occur.
[0032] To address this issue, this application provides a method, device, electronic device, and medium for predicting vehicle controller faults. The method trains and validates a basic vehicle controller fault prediction model using multi-category processed fault data to obtain a final vehicle controller fault prediction model. This model is then deployed to the vehicle controller for real-time fault prediction. Based on the fault prediction results, the method handles faults, adapting to various scenarios and improving dynamic adaptability. Furthermore, it can predict fault occurrences in advance and preventative measures are taken before faults occur, thus protecting the vehicle controller.
[0033] Figure 1 This is a flowchart of a vehicle controller fault prediction method provided according to an embodiment of this application. Figure 1 As shown, the vehicle controller fault prediction method of this application embodiment may specifically include: S101. Obtain fault data from the vehicle controller and perform multi-category processing on the fault data to obtain processed fault data.
[0034] S102. Based on the processed fault data, the preset basic vehicle controller fault prediction model is trained and verified to obtain the final vehicle controller fault prediction model.
[0035] S103. The vehicle controller fault prediction model is sent to the vehicle controller and deployed based on the preset deployment method.
[0036] S104. Predict the real-time fault data of the vehicle controller using the vehicle controller fault prediction model, and process the fault based on the fault prediction results.
[0037] In the above-mentioned vehicle controller fault prediction method, the basic vehicle controller fault prediction model is trained and validated using fault data processed by multiple categories to obtain the final vehicle controller fault prediction model. The model is then deployed to the vehicle controller to perform real-time fault prediction on the vehicle controller and to handle faults based on the fault prediction results. This method can adapt to various scenarios, improve dynamic adaptability, and can predict the occurrence of faults in advance. Furthermore, it can prevent and handle the controller in advance before the fault occurs, thereby avoiding the occurrence of faults and protecting the vehicle controller.
[0038] The exemplary steps described above in the embodiments of this application are illustrated below with specific examples: S101: Obtain fault data from the vehicle controller and perform multi-category processing on the fault data to obtain processed fault data.
[0039] In this embodiment, the multi-category processing includes at least data cleaning, data labeling, data augmentation, and data synthesis. Fault data undergoes these processes to obtain processed fault data for subsequent processing. The fault data can be information from one minute prior to the fault occurrence, such as hardware information (including chip and peripheral current, voltage, temperature, and instruction and register integrity); software information (including runtime, watchdog timeout frequency, abnormal interrupt frequency, CPU load, error count, crystal oscillator frequency anomaly, task cycle timeout frequency, task anomaly recovery rate, and resource utilization); and communication information (including error frames, frame loss rate, and transmission offset).
[0040] Optionally, the system can locate missing data in the fault data, determine the filling data that matches the missing data, and replace the missing data with the filling data; it can also determine invalid data in the fault data based on a preset verification mechanism, remove the invalid data, and obtain cleaned fault data.
[0041] Specifically, faulty data is cleaned by converting it into usable data. For example, interpolation is performed to fill in the missing data information, such as current, voltage, and temperature signals, with the average value of adjacent time intervals. Invalid data is removed, for example, if a lot of information is missing at a certain moment and data errors are found during the transmission process (CRC cyclic redundancy check is used to determine whether there are errors), it is determined to be invalid data and removed. Through data cleaning, the quality of faulty data is ensured.
[0042] Optionally, fault data is aggregated into multiple fault categories based on a preset classification model to obtain the corresponding target classification model; the fault data is then labeled based on the target classification model to align the fault data with the corresponding fault categories, resulting in labeled fault data. Here, the target classification model represents the classification model under multiple fault categories.
[0043] Specifically, fault data is labeled. Data labeling aligns fault classification with fault data. A classification model (such as the K-means algorithm) can be used to aggregate fault data into multiple fault categories (e.g., 20 categories) to obtain the target classification model, namely the 20-means model. Then, the fault data is labeled using the 20-means model.
[0044] Optionally, low-intensity Gaussian white noise is added to the labeled fault data, and the fault data of the target type in the fault data is identified; a randomly selected range is set, and the fault data of the target type is scaled based on the randomly selected range to obtain the enhanced fault data.
[0045] Alternatively, the fault data of the target type can be offset from left to right along the time axis to obtain the offset fault data.
[0046] Specifically, data augmentation of fault data involves adding low-intensity Gaussian white noise to the labeled fault data and using amplitude scaling and offset methods to scale the labeled fault data information (e.g., numerical data such as voltage, current, temperature, load rate, etc.) within a small range (e.g., randomly selected range of 0.9 to 1.1 times) or offset it (the fault data is offset to the left and right along the time axis) to obtain the augmented fault data.
[0047] Optionally, the labeled fault data and the enhanced fault data are input into a preset first recurrent neural network model to generate corresponding true latent vectors; preset Gaussian white noise is input into a preset second recurrent neural network model to generate corresponding false latent vectors; the true and false latent vectors are input into a preset third recurrent neural network model to distinguish between true and false latent vectors, and the final true and false latent vectors are input into a preset fourth recurrent neural network model to obtain synthesized fault data. The recurrent neural network model is the GRU model.
[0048] Specifically, the fault data is synthesized using a time-based generative adversarial network (GRU). For example, labeled and augmented data are input into the algorithm embedding module (GRU1 model) to generate real latent vectors. Gaussian noise is input into the generator module (GRU2 model) to obtain false latent vectors. The real and false latent vectors are then input into the discriminator (GRU3 model) to distinguish between them. Finally, the real and false latent vectors are input into the restoration module (GRU4 model) to generate new time-series data, i.e., the synthesized fault data. The processed data (labeled original fault data, augmented fault data, and synthesized fault data) are then stored in the database.
[0049] S102, Based on the processed fault data, the preset basic vehicle controller fault prediction model is trained and verified to obtain the final vehicle controller fault prediction model.
[0050] In this embodiment of the application, the fault data processed in step S101 is used to train and verify the basic vehicle controller fault prediction model to obtain the final vehicle controller fault prediction model for subsequent processing.
[0051] Optionally, when training and validating the preset basic vehicle controller fault prediction model based on the processed fault data, the synthesized fault data is randomly split into a first proportion of first fault data and a second proportion of second fault data; the basic vehicle controller fault prediction model is fine-tuned based on the first fault data; the model accuracy of the vehicle controller fault prediction model is verified based on the second fault data; if the model accuracy is greater than the preset accuracy threshold, the second fault data is input into the vehicle controller fault prediction model for further training to obtain the final vehicle controller fault prediction model.
[0052] Specifically, for example, the synthesized fault data is randomly split into an 8:2 ratio. 80% of the data is used to fine-tune the basic vehicle controller fault prediction model to obtain a new model, namely the vehicle controller fault prediction model. 20% of the data is used for model validation to verify whether the accuracy of the new model is greater than a preset accuracy threshold (e.g., 90%). If the requirement is met, the 20% of the data is input into the new model to continue training to obtain the final model. Finally, the model and its parameters are stored and awaited deployment. If the model does not meet the requirements, fault data is collected again for data processing and used for model training.
[0053] S103, the vehicle controller fault prediction model is sent to the vehicle controller and deployed based on the preset deployment method.
[0054] In this embodiment of the application, the vehicle controller fault prediction model in step S102 is sent to the vehicle controller for deployment according to the set deployment method for subsequent application.
[0055] S104 predicts real-time fault data of the vehicle controller through the vehicle controller fault prediction model and processes the fault based on the fault prediction results.
[0056] In this embodiment of the application, the real-time fault data of the vehicle controller can be predicted by the deployed vehicle controller fault prediction model to obtain the fault prediction result, and the fault can be processed according to the fault prediction result.
[0057] Optionally, the confidence level of the fault data is determined; if the confidence level of the fault data is greater than a preset confidence threshold, the on-board controller responds to the fault data.
[0058] Specifically, for example, real-time data from the vehicle controller is input into the vehicle controller fault prediction model. The model outputs the changes in each signal and the corresponding faults and confidence levels within the next 5 seconds. If the confidence level is greater than 50%, the fault is determined to be a real and reliable fault, and the corresponding response method is used (e.g., for faults caused by excessive temperature or excessive load, shutting down unnecessary software; for faults caused by abnormal software interruption, restarting the relevant software, etc.) to reduce the possibility of the controller failing. If the prediction confidence level is less than 50% or the fault is not predicted, the data information is stored in the memory.
[0059] The vehicle controller fault prediction method provided in this application acquires fault data of the vehicle controller, performs multi-category processing on the fault data to obtain processed fault data, trains and validates a preset basic vehicle controller fault prediction model based on the processed fault data to obtain a final vehicle controller fault prediction model, distributes the vehicle controller fault prediction model to the vehicle controller and deploys it according to a preset deployment method, predicts real-time fault data of the vehicle controller using the vehicle controller fault prediction model, and processes the fault based on the fault prediction results. This vehicle controller fault prediction method, by training and validating a basic vehicle controller fault prediction model using multi-category processed fault data to obtain a final vehicle controller fault prediction model, and then distributing the model to the vehicle controller for deployment, enables real-time fault prediction of the vehicle controller using the vehicle controller fault prediction model and processing of the fault based on the fault prediction results. It can adapt to various scenarios, improve dynamic adaptability, and can predict fault occurrence in advance, providing preventative measures to the controller before faults occur, thereby avoiding faults and protecting the vehicle controller.
[0060] Furthermore, in response to receiving an update instruction from the manufacturer, it determines whether the update information and model accuracy of the current vehicle controller fault prediction model meet the update instruction, and feeds back to the manufacturer for secondary confirmation; in response to receiving an update confirmation instruction from the manufacturer, it packages the vehicle controller fault prediction model into a model installation package in the target format, and determines whether the vehicle is in a dormant state within a preset update time; in response to the vehicle being online, it downloads the model installation package through the vehicle and updates based on the model installation package.
[0061] Furthermore, the vehicle controller fault prediction model is dynamically updated via cloud service OTA; the updated vehicle controller fault prediction model is then deployed in the vehicle controller.
[0062] Specifically, the model deployment is carried out on the vehicle via OTA according to the manufacturer's requirements. The model deployment module can receive instructions from the manufacturer's backend to update the model via OTA, and deploy the new model that meets the requirements (e.g., accuracy > 90%) in the vehicle controller. That is, the manufacturer's backend requests an update -> the model deployment module receives the instruction, determines whether the current model has been updated and whether the accuracy meets the requirements, and provides feedback to the manufacturer -> the manufacturer needs to confirm whether to update or not, and gives the instruction to the model deployment module. If an update is required, the model is packaged into a binary format installation package. At the preset update time, for example, 3 am, the vehicle is determined to be in a dormant state. The vehicle is woken up through the cloud server, a connection is established, and then the vehicle begins to download the installation package, decompress and upgrade. After the upgrade is completed, the manufacturer and user are notified of the successful upgrade.
[0063] Therefore, this application solves the problem of insufficient fault prediction data through a series of data processing methods, and dynamically updates the model in the controller through cloud service OTA to ensure that the accuracy of model prediction is continuously updated, thereby improving the security of the controller.
[0064] Figure 2 This is a schematic diagram of the structure of the vehicle controller fault prediction device provided in the embodiments of this application; as shown below. Figure 2 As shown, the vehicle controller fault prediction device 200 of this application embodiment may specifically include: The processing module 201 is used to acquire fault data from the vehicle controller and perform multi-category processing on the fault data to obtain processed fault data; wherein, the multi-category processing includes at least data cleaning, data labeling, data augmentation, and data synthesis. The acquisition module 202 is used to train and verify the preset basic vehicle controller fault prediction model based on the processed fault data to obtain the final vehicle controller fault prediction model. The deployment module 203 is used to send the vehicle controller fault prediction model to the vehicle controller and deploy it based on a preset deployment method; The prediction module 204 is used to predict the real-time fault data of the vehicle controller through the vehicle controller fault prediction model, and to process the fault based on the fault prediction results.
[0065] In one possible implementation, the processing module is specifically used for: Locate the missing data in the fault data, identify the fill data that matches the missing data, and replace the missing data with the fill data; Invalid data in the fault data is identified based on a preset verification mechanism and removed to obtain cleaned fault data.
[0066] In one possible implementation, the processing module is specifically used for: Based on a pre-defined classification model, fault data is aggregated into multiple fault categories to obtain the corresponding target classification model; whereby the target classification model represents the classification model under multiple fault categories. The fault data is labeled based on the target classification model to align the fault data with the corresponding fault categories, thus obtaining labeled fault data.
[0067] In one possible implementation, the processing module is specifically used for: Add low-intensity Gaussian white noise to the labeled fault data and identify the target type of fault data in the fault data; Set a randomly selected range, and scale the fault data of the target type based on the randomly selected range to obtain enhanced fault data.
[0068] In one possible implementation, the processing module is specifically used for: The labeled fault data and the enhanced fault data are input into the preset first recurrent neural network model to generate the corresponding true latent vectors; The preset Gaussian white noise is input into the preset second recurrent neural network model to generate the corresponding pseudo-hidden vector; The real and fake latent vectors are input into a pre-defined third recurrent neural network model to distinguish between them, and the final real and fake latent vectors are input into a pre-defined fourth recurrent neural network model to obtain the synthesized fault data.
[0069] In one possible implementation, the acquisition module is specifically used for: The synthesized fault data is randomly split into a first proportion of first fault data and a second proportion of second fault data. Fine-tuning the basic vehicle controller fault prediction model based on the first fault data; The accuracy of the vehicle controller fault prediction model was verified based on the second fault data. If the model accuracy is greater than the preset accuracy threshold, the second fault data is input into the vehicle controller fault prediction model for further training to obtain the final vehicle controller fault prediction model.
[0070] In one possible implementation, the apparatus further includes: The first judgment module is used to respond to the update instruction received from the manufacturer, determine whether the update information and model accuracy of the current vehicle controller fault prediction model meet the update instruction, and feed back to the manufacturer for secondary confirmation. The second judgment module is used to package the vehicle controller fault prediction model into a model installation package in the target format in response to receiving the update confirmation instruction from the manufacturer, and to determine whether the vehicle is in a dormant state within a preset update time. The update module is used to download the model installation package via the vehicle and perform updates based on the model installation package when the vehicle is online.
[0071] The vehicle controller fault prediction device provided in this application acquires fault data from the vehicle controller, performs multi-category processing on the fault data to obtain processed fault data, trains and validates a preset basic vehicle controller fault prediction model based on the processed fault data, obtains a final vehicle controller fault prediction model, distributes the vehicle controller fault prediction model to the vehicle controller and deploys it according to a preset deployment method, predicts real-time fault data of the vehicle controller using the vehicle controller fault prediction model, and processes the fault based on the fault prediction results. This vehicle controller fault prediction device, by training and validating a basic vehicle controller fault prediction model using multi-category processed fault data to obtain a final vehicle controller fault prediction model, and then distributing the model to the vehicle controller for deployment, enables real-time fault prediction of the vehicle controller using the vehicle controller fault prediction model and processing of the fault based on the fault prediction results. It can adapt to various scenarios, improve dynamic adaptability, and can predict fault occurrence in advance, providing preventative measures to the controller before faults occur, thereby avoiding faults and protecting the vehicle controller.
[0072] like Figure 3 As shown in the embodiment of this application, an electronic device 300 includes a processor 301, a memory 302, and a bus. The memory 302 stores machine-readable instructions that can be executed by the processor 301. When the electronic device is running, the processor 301 communicates with the memory 302 via the bus. The processor 301 executes the machine-readable instructions to perform the steps of the above-described vehicle controller fault prediction method.
[0073] Specifically, the memory 302 and processor 301 mentioned above can be general-purpose memory and processor, without any specific limitations. When the processor 301 runs the computer program stored in the memory 302, it can execute the above-mentioned vehicle controller fault prediction method.
[0074] Corresponding to the above-described vehicle controller fault prediction method, this application embodiment also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the above-described vehicle controller fault prediction method.
[0075] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and devices described above can be referred to the corresponding processes in the method embodiments, and will not be repeated here. In the several embodiments provided in this application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed mutual coupling or direct coupling or communication connection can be through some communication interfaces; the indirect coupling or communication connection of devices or modules can be electrical, mechanical, or other forms.
[0076] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.
[0077] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.
[0078] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the deployment methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.
[0079] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.
Claims
1. A method for predicting faults in an on-board controller, characterized in that, The method includes: Fault data of the vehicle controller is acquired, and the fault data is processed in multiple categories to obtain processed fault data; wherein, the multiple categories of processing include at least data cleaning, data labeling, data augmentation, and data synthesis. Based on the processed fault data, the preset basic vehicle controller fault prediction model is trained and verified to obtain the final vehicle controller fault prediction model. The vehicle controller fault prediction model is sent to the vehicle controller and deployed based on a preset deployment method; The on-board controller fault prediction model is used to predict real-time fault data of the on-board controller, and the fault is processed based on the fault prediction results.
2. The method according to claim 1, characterized in that, The multi-category processing of the fault data includes: Locate the missing data in the fault data, determine the filler data that matches the missing data, and replace the missing data with the filler data; Invalid data in the fault data is identified based on a preset verification mechanism, and the invalid data is removed to obtain cleaned fault data.
3. The method according to claim 1, characterized in that, The multi-category processing of the fault data includes: Based on a preset classification model, the fault data is aggregated into multiple fault categories to obtain the corresponding target classification model; wherein, the target classification model represents the classification model under the multiple fault categories; The fault data is labeled based on the target classification model to align the fault data with the corresponding fault categories, thus obtaining labeled fault data.
4. The method according to claim 1, characterized in that, The multi-category processing of the fault data includes: Add low-intensity Gaussian white noise to the labeled fault data and determine the target type of fault data in the fault data; Set a randomly selected range, and scale the fault data of the target type based on the randomly selected range to obtain enhanced fault data.
5. The method according to claim 1, characterized in that, The multi-category processing of the fault data includes: The labeled fault data and the enhanced fault data are input into the preset first recurrent neural network model to generate the corresponding true latent vectors; The preset Gaussian white noise is input into the preset second recurrent neural network model to generate the corresponding pseudo-hidden vector; The true and false latent vectors are input into a preset third recurrent neural network model to distinguish between them, and the final true and false latent vectors are input into a preset fourth recurrent neural network model to obtain the synthesized fault data.
6. The method according to claim 1, characterized in that, The step of training and validating a preset basic vehicle controller fault prediction model based on the processed fault data includes: The synthesized fault data is randomly split into a first proportion of first fault data and a second proportion of second fault data; The basic vehicle controller fault prediction model is fine-tuned based on the first fault data. The accuracy of the vehicle controller fault prediction model is verified based on the second fault data. If the accuracy of the model is greater than the preset accuracy threshold, the second fault data is input into the vehicle controller fault prediction model for further training to obtain the final vehicle controller fault prediction model.
7. The method according to claim 1, characterized in that, The method further includes: In response to receiving an update instruction from the manufacturer, it determines whether the update information and model accuracy of the current vehicle controller fault prediction model conform to the update instruction, and feeds it back to the manufacturer for secondary confirmation. In response to receiving an update confirmation instruction from the manufacturer, the vehicle controller fault prediction model is packaged into a model installation package in the target format, and the vehicle is determined to be in a dormant state within a preset update time. In response to the vehicle being online, the model installation package is downloaded via the vehicle, and an update is performed based on the model installation package.
8. A vehicle controller fault prediction device, characterized in that, The device includes: The processing module is used to acquire fault data from the vehicle controller and perform multi-category processing on the fault data to obtain processed fault data; wherein, the multi-category processing includes at least data cleaning, data labeling, data augmentation, and data synthesis. The acquisition module is used to train and verify the preset basic vehicle controller fault prediction model based on the processed fault data to obtain the final vehicle controller fault prediction model. The deployment module is used to send the vehicle controller fault prediction model to the vehicle controller and deploy it based on a preset deployment method. The prediction module is used to predict the real-time fault data of the vehicle controller through the vehicle controller fault prediction model, and to process the fault based on the fault prediction results.
9. An electronic device, characterized in that, include: The device includes a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, the steps of the vehicle controller fault prediction method as described in any one of claims 1 to 7 are performed.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the vehicle controller fault prediction method as described in any one of claims 1 to 7.