A large model-based automobile fault light elimination method, device and medium

Abstract

The application provides a large model-based automobile fault light elimination method, which acquires fault light signals, vehicle real-time operation parameters and historical fault data through a vehicle sensor and a fault diagnosis system memory, constructs a fault light description information database, uses the acquired data as a learning sample, trains a fault recognition large model based on the learning sample, enables the large model to have fault light signal recognition, fault analysis and scheme generation capabilities, integrates the optimized large model into an automobile fault diagnosis system, triggers a fault light recognition and elimination process when the fault light is on, records driver driving habits and vehicle use conditions, and adjusts fault diagnosis priorities and solution recommendation logic in a targeted manner; the application realizes accurate fault determination and efficient processing through multi-source data deep fusion, twin model training and personalized adaptation optimization.

Description

Technical Field

[0001] This invention relates to the field of data analysis models and fault elimination technology, specifically to a method, device, and medium for eliminating automotive fault lights based on a large model. Background Technology

[0002] With the rapid advancement of automotive electronics and intelligence, in-vehicle electronic control systems are becoming increasingly complex. As a direct warning system for vehicle malfunctions, the accurate identification, fault location, and efficient elimination of faults directly impact driver safety and the user experience. Currently, automotive malfunction indicator lamp (MIL) diagnostic technology has evolved from traditional single fault code reading to a multi-dimensional diagnostic model combining sensor data and historical fault records. Some models have even introduced machine learning models to assist in fault analysis, attempting to improve the accuracy of fault diagnosis.

[0003] In existing technologies, fault light clearing mainly relies on manual reset after maintenance or automatic reset based on simple system threshold judgment. The core technology depends on matching fault codes with a preset fault database. While some solutions incorporate vehicle operating parameters, the data integration is low, and the correlation between data and the root cause of the fault is not fully explored, resulting in insufficient adaptability between fault diagnosis and fault light clearing logic. With consumers' increasing demands for driving safety and intelligent experience, there is an urgent need for an integrated technical solution that can achieve accurate fault identification, intelligent fault level determination, personalized solution delivery, and efficient fault light clearing.

[0004] The following drawbacks still exist in the existing technology: Low data correlation and insufficient accuracy in fault diagnosis: Existing solutions mostly rely on fault light signals or single operating parameters for diagnosis, failing to achieve deep integration of fault light signals, real-time operating parameters, and historical fault data. Furthermore, they do not quantify the contribution weight of each data modality to fault diagnosis and the root cause correlation coefficient, resulting in large deviations in fault cause analysis, inaccurate level determination, and a high risk of misjudgment or omission.

[0005] Lack of personalized adaptation: The existing fault diagnosis and light-off solutions use a uniform logic, which does not take into account the driver's driving habits and vehicle usage to adjust the diagnosis priority and solution recommendation logic. This makes it impossible to adapt to the different user scenarios and needs, resulting in a poor user experience. Summary of the Invention

[0006] This invention provides a method, device, and medium for eliminating automotive fault lights based on a large model. By deeply fusing multi-source data, training twin models, and personalized adaptation and optimization, it achieves accurate fault determination and efficient processing, thereby solving the problems in the background technology.

[0007] To achieve the above objectives, the technical solution of the present invention is as follows: S1, Data Collection Phase: Through vehicle sensors and fault diagnosis system memory, fault light signals, real-time vehicle operating parameters and historical fault data are collected to build a fault light description information database. S2, Model Training Phase: Using the fault light instruction information database as the learning samples, a large fault recognition model is trained to enable it to recognize fault light signals, analyze faults, and generate solutions. S3, System Integration Stage: The trained and optimized fault identification model is integrated into the vehicle fault diagnosis system, and coordinated with the vehicle electronic control system and the central control screen system. S4, Fault Handling Phase: When the fault light illuminates, the fault light identification and clearing process is automatically triggered. S5, the personalized adaptation stage, records the driver's driving habits and vehicle usage, and adjusts the fault diagnosis priority and solution recommendation logic accordingly.

[0008] Preferably, the specific implementation steps of S1 are as follows: S11, Deploy data acquisition terminals: Deploy data acquisition terminals in vehicle sensors and fault diagnosis system memory to establish data transmission links; S12, Multi-type data acquisition: Simultaneously acquire fault light signals, real-time vehicle operating parameters, and historical fault data through the acquisition terminal; S13, Data Cleaning and Verification: Remove outliers from the collected data, standardize the format, and verify the integrity and validity of the data. S14, Database Construction: Based on the verified data, build a fault light description information database, and input fault light diagrams, safety level classifications and corresponding parameters; S15, Data entry and storage: Classify and store the collected valid data and basic database information into the database, and complete the connection and docking between the data and the database.

[0009] Preferably, the specific implementation steps of S2 are as follows: S21, Sample processing: Extract the data collected in S1, analyze historical fault information, and generate standardized learning samples including fault light signals, operating parameters, and fault labels; S22, Sub-model 1 training: Train a multimodal fusion sub-model based on the samples to be learned, and extract feature fusion and root cause association features including fault light signals and vehicle operation data; S23, Sub-model 2 training: Based on the fusion features output by sub-model 1, train the fusion feature-root cause association weighted probability calibration sub-model to determine the fault level; S24, Joint Training of Large Models: Jointly optimize the trained twin models to generate a large fault identification model, and optimize the model output logic by combining fault handling decision criteria. S25, Model Performance Verification: Verify the large model using test samples to confirm its ability to identify fault light signals, analyze faults, and generate solutions. If it fails to meet the requirements, return to S22 for iterative training.

[0010] Preferably, the specific implementation steps of the S22 sub-model are as follows: S221, receiving the data verified by S1 includes setting the fault light signal. Real-time operating parameters Historical fault data Output a fusion feature matrix with root cause correlation. :

[0011] in: For fault light signal Modal weights, For real-time running parameters Modal weights, Historical fault data The modal weights are all obtained by statistical fitting from the historical fault data collected by S1 and by calculating the contribution ratio of each data mode in the fault determination. For fault light signal The correlation coefficient, For fault light signal The correlation coefficient, For fault light signal The correlation coefficients were all obtained through joint training from the S1 fault light description information database and historical fault data. Represented as element-wise product; S222, Integrating Feature Filtering and Standardization: For Core features are selected and then normalized to obtain a standardized fusion feature matrix. ':

[0012] in, , Output for S221 Minimum and maximum values; S223, Training Sample Reconstruction: Based on ', associate the safety level label y with the S1 fault light description information database, and reconstruct the exclusive training sample set for S23. ; S224, Sample Splitting and Input: Divide the reconstructed sample set into a training set and a validation set in a 7:3 ratio, and start training sub-model 2.

[0013] Preferably, the specific implementation steps of the S23 sub-model are as follows: S231, Model Initialization: Set the root cause triggering influence coefficient ξ, which is obtained from the historical fault data and safety level classification in S1. Define the model loss function as cross-entropy loss L. S232, Forward calculation of risk probability: Substitute the training set input from S224 into the core calibration formula, and calculate the failure risk probability p for each sample:

[0014] in, This represents the weight coefficient of the k-th data mode. Represents the correlation coefficient of the k-th data mode; S233, Loss Calculation and Backward Parameter Tuning: Substitute the P obtained from the forward calculation and the training set label y into the loss function to calculate the loss value L, and then backpropagate through the gradient descent method to fine-tune the model adaptation parameters.

[0015] Where n represents the total amount of data in the sample set, and i represents the index of the data in the sample set. This represents the grade label of the i-th sample. This represents the failure risk probability of the i-th sample; S234, Validation of validation set effect: Substitute the validation set input in S224 into the optimized model, calculate the validation set risk probability, match the validation set labels, verify the accuracy of the model's fault level determination, and converge and solidify the model after meeting the preset threshold.

[0016] Preferably, the specific implementation steps of S3 are as follows: S31, Integrated Environment Setup: Builds a dedicated integrated environment for automotive fault diagnosis systems, adapts to the computing power specifications of in-vehicle hardware, completes communication protocol integration with in-vehicle electronic control systems and central control screen systems, and reserves interfaces for large model data interaction and command transmission.

[0017] S32, Model Deployment and Loading: Solidify and deploy the large fault recognition model trained and optimized by S2 to the automotive fault diagnosis system to complete the lightweight adaptation of the model. S33, the data link is established, creating a two-way data link between the fault diagnosis system and the S1 acquisition terminal, the vehicle electronic control system, and the central control screen system. This enables real-time transmission of acquired data to the model and the issuance of model commands to the vehicle system.

[0018] S34, functional module linkage debugging, verifies the model recognition, analysis, and level determination capabilities, as well as the display of diagnostic results to the central control screen and the execution effect of control commands to the vehicle electronic control system.

[0019] Preferably, the specific implementation steps of S4 are as follows: S41, the large model automatically recognizes the fault light signals on the instrument panel and the central control screen; S42, combine real-time vehicle operating status and historical fault data to analyze possible causes of the fault; S43 generates the fault level and corresponding solution, which is then displayed to the driver via the central control screen. S44: If the fault does not require repair, a control command is generated and the fault light is automatically cleared through the vehicle electronic control system; if repair is required, the system confirms that the fault has been eliminated and clears the fault light after the driver completes the repair; by continuously analyzing vehicle operating data, potential faults are predicted and the driver is reminded in advance through the central control screen.

[0020] Preferably, the specific implementation steps of S5 are as follows: S51, Continuous data collection: The large model records the driver's driving habits and vehicle usage in real time; S52, Data Feature Extraction: Extract features from recorded data and correlate them with historical fault diagnosis and solution feedback information; S53, Adaptation Logic Adjustment: Based on feature analysis, the logic for prioritizing fault diagnosis and recommending solutions is optimized in a targeted manner; S54, Adaptation effect verification: Combined with subsequent fault handling scenarios, verify the adaptability and practicality of the adjusted logic, and optimize the personalized adaptation logic.

[0021] In another aspect, the present invention also discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the method described above.

[0022] In another aspect, the present invention also discloses a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the method described above.

[0023] As can be seen from the above technical solution compared with the prior art, the present invention has the following beneficial effects: This invention constructs a multi-dimensional data acquisition and deep fusion mechanism, combined with twin sub-mode training, to achieve full-dimensional correlation of fault light signals, real-time operating parameters, and historical fault data. It quantifies the modal weights and root cause correlation coefficients of each data mode, thereby improving the accuracy of fault identification, cause analysis, and level determination.

[0024] Non-emergency fault lights can be turned off manually, improving fault handling efficiency. Attached Figure Description

[0025] Figure 1This is a schematic diagram of the method steps in an embodiment of the present invention; Figure 2 This is a schematic diagram of the specific steps in S4 of this embodiment of the invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0027] The embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are used to illustrate the present invention, but should not be used to limit the scope of the present invention.

[0028] S1, Data Collection Phase: Through vehicle sensors and fault diagnosis system memory, fault light signals, real-time vehicle operating parameters and historical fault data are collected to build a fault light description information database. S2, Model Training Phase: Using the fault light instruction information database as the learning samples, a large fault recognition model is trained to enable it to recognize fault light signals, analyze faults, and generate solutions. S3, System Integration Stage: The trained and optimized fault identification model is integrated into the vehicle fault diagnosis system, and coordinated with the vehicle electronic control system and the central control screen system. S4, Fault Handling Phase: When the fault light illuminates, the fault light identification and clearing process is automatically triggered. S5, the personalized adaptation stage, records the driver's driving habits and vehicle usage, and adjusts the fault diagnosis priority and solution recommendation logic accordingly.

[0029] Example: like Figure 1 As shown in this embodiment, a malfunction indicator light suddenly illuminates while a vehicle is being driven, presenting problems such as the inability to determine the severity of the malfunction, the lack of precise troubleshooting guidance, and the cumbersome process of clearing the malfunction indicator light. Therefore, the present invention adopts a method for eliminating car fault lights based on a large model, and then initiates the fault handling process, automatically completes fault light identification, multi-source data fusion analysis and fault level determination, pushes the appropriate solution, automatically confirms and eliminates the fault light after the fault is resolved, and can also provide early warning of potential faults, thus completely solving the above-mentioned car use problems of users.

[0030] S11, Deploy data acquisition terminals: Deploy data acquisition terminals in vehicle sensors and fault diagnosis system memory to establish data transmission links; S12, Multi-type data acquisition: Simultaneously acquire fault light signals, real-time vehicle operating parameters, and historical fault data through the acquisition terminal; S13, Data Cleaning and Verification: Remove outliers from the collected data, standardize the format, and verify the integrity and validity of the data. S14, Database Construction: Based on the verified data, build a fault light description information database, and input fault light diagrams, safety level classifications and corresponding parameters; S15, Data entry and storage: Classify and store the collected valid data and basic database information into the database, and complete the connection and docking between the data and the database.

[0031] S21, Sample processing: Extract the data collected in S1, analyze historical fault information, and generate standardized learning samples including fault light signals, operating parameters, and fault labels; S22, Sub-model 1 training: Train a multimodal fusion sub-model based on the samples to be learned, and extract feature fusion and root cause association features including fault light signals and vehicle operation data; S23, Sub-model 2 training: Based on the fusion features output by sub-model 1, train the fusion feature-root cause association weighted probability calibration sub-model to determine the fault level; S24, Joint Training of Large Models: Jointly optimize the trained twin models to generate a large fault identification model, and optimize the model output logic by combining fault handling decision criteria. S25, Model Performance Verification: Verify the large model using test samples to confirm its ability to identify fault light signals, analyze faults, and generate solutions. If it fails to meet the requirements, return to S22 for iterative training.

[0032] The specific implementation steps of the S22 sub-model are as follows: S221, receiving the data verified by S1 includes setting the fault light signal. Real-time operating parameters Historical fault data Output a fusion feature matrix with root cause correlation. :

[0033] in: For fault light signal Modal weights, For real-time running parameters Modal weights, Historical fault data The modal weights are all obtained by statistical fitting from the historical fault data collected by S1 and by calculating the contribution ratio of each data mode in the fault determination. For fault light signal The correlation coefficient, For fault light signal The correlation coefficient, For fault light signal The correlation coefficients were all obtained through joint training from the S1 fault light description information database and historical fault data. Represented as element-wise product; S222, Integrating Feature Filtering and Standardization: For Core features are selected and then normalized to obtain a standardized fusion feature matrix. ':

[0034] in, , Output for S221 Minimum and maximum values; S223, Training Sample Reconstruction: Based on ', associate the safety level label y with the S1 fault light description information database, and reconstruct the exclusive training sample set for S23. ; S224, Sample Splitting and Input: Divide the reconstructed sample set into a training set and a validation set in a 7:3 ratio, and start training sub-model 2.

[0035] This application outputs fusion features with root cause correlation through sub-model 1, and sub-model 2 performs risk probability calibration based on these features, realizing strong linkage between the two models' data. This ensures that fault level determination has clear data support and mathematical basis, avoids the subjective threshold judgment bias of traditional solutions, and improves the reliability of fault level determination.

[0036] S231, Model Initialization: Set the root cause triggering influence coefficient ξ, which is obtained from the historical fault data and safety level classification in S1. Define the model loss function as cross-entropy loss L. S232, Forward calculation of risk probability: Substitute the training set input from S224 into the core calibration formula, and calculate the failure risk probability p for each sample:

[0037] in, This represents the weight coefficient of the k-th data mode. Represents the correlation coefficient of the k-th data mode; S233, Loss Calculation and Backward Parameter Tuning: Substitute the P obtained from the forward calculation and the training set label y into the loss function to calculate the loss value L, and then backpropagate through the gradient descent method to fine-tune the model adaptation parameters.

[0038] Where n represents the total amount of data in the sample set, and i represents the index of the data in the sample set. This represents the grade label of the i-th sample. This represents the failure risk probability of the i-th sample; This application constructs a multi-dimensional data acquisition and deep fusion mechanism, combined with twin model training, to achieve full-dimensional correlation of fault light signals, real-time operating parameters, and historical fault data. It quantifies the modal weights and root cause correlation coefficients of each data mode, thereby improving the accuracy of fault identification, cause analysis, and level determination.

[0039] S234, Validation of validation set effect: Substitute the validation set input in S224 into the optimized model, calculate the validation set risk probability, match the validation set labels, verify the accuracy of the model's fault level determination, and converge and solidify the model after meeting the preset threshold.

[0040] S31, Integrated Environment Setup: Builds a dedicated integrated environment for automotive fault diagnosis systems, adapts to the computing power specifications of in-vehicle hardware, completes communication protocol integration with in-vehicle electronic control systems and central control screen systems, and reserves interfaces for large model data interaction and command transmission.

[0041] S32, Model Deployment and Loading: Solidify and deploy the large fault recognition model trained and optimized by S2 to the automotive fault diagnosis system to complete the lightweight adaptation of the model. S33, the data link is established, creating a two-way data link between the fault diagnosis system and the S1 acquisition terminal, the vehicle electronic control system, and the central control screen system. This enables real-time transmission of acquired data to the model and the issuance of model commands to the vehicle system.

[0042] S34, functional module linkage debugging, verifies the model recognition, analysis, and level determination capabilities, as well as the display of diagnostic results to the central control screen and the execution effect of control commands to the vehicle electronic control system.

[0043] like Figure 2 As shown, the specific steps of S4 include: S41, the large model automatically recognizes the fault light signals on the instrument panel and the central control screen; S42, combine real-time vehicle operating status and historical fault data to analyze possible causes of the fault; S43 generates the fault level and corresponding solution, which is then displayed to the driver via the central control screen. S44: If the fault does not require repair, a control command is generated and the fault light is automatically cleared through the vehicle electronic control system; if repair is required, the system confirms that the fault has been eliminated and clears the fault light after the driver completes the repair; by continuously analyzing vehicle operating data, potential faults are predicted and the driver is reminded in advance through the central control screen.

[0044] This invention integrates a large model with the vehicle electronic control system and the central control screen system, enabling full-link data transmission and command interaction. It achieves full-process automation of "identification-analysis-judgment-push-light extinguishing" after a fault is triggered, and can complete the extinguishing of non-emergency faults without manual intervention, thus improving fault handling efficiency.

[0045] S51, Continuous data collection: The large model records the driver's driving habits and vehicle usage in real time; S52, Data Feature Extraction: Extract features from recorded data and correlate them with historical fault diagnosis and solution feedback information; S53, Adaptation Logic Adjustment: Based on feature analysis, the logic for prioritizing fault diagnosis and recommending solutions is optimized in a targeted manner; S54, Adaptation effect verification: Combined with subsequent fault handling scenarios, verify the adaptability and practicality of the adjusted logic, and optimize the personalized adaptation logic.

[0046] This invention continuously records driver habits and vehicle usage, extracts core features, and optimizes diagnostic priorities and solution recommendation logic to achieve personalized fault handling, meeting the needs of different users, improving user driving safety and intelligent experience. Pre-set verification mechanisms and iterative optimization logic ensure that the large model and twin models meet performance standards, and continuous data collection enables dynamic model optimization, guaranteeing the stability and adaptability of the solution, making it suitable for different vehicle models and fault scenarios.

[0047] In another aspect, the present invention also discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the method described above.

[0048] In another aspect, the present invention also discloses a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the method described above.

[0049] In another embodiment provided in this application, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to execute any of the large-model-based automotive fault light elimination methods described in the above embodiments.

[0050] It is understood that the systems, devices, and storage media provided in the embodiments of the present invention correspond to the methods provided in the embodiments of the present invention, and the explanations, examples, and beneficial effects of the relevant content can be referred to the corresponding parts of the above methods.

[0051] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another.

[0052] For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access, or a data storage device such as a server or data center that integrates one or more available media.

[0053] The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid state disks (SSDs)).

[0054] It should be noted that in this paper, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0055] Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0056] The various embodiments in this specification are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0057] The embodiments of the present invention are given for the purposes of illustration and description. Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for eliminating automotive fault lights based on a large model, characterized in that, Perform the following steps using a computer device: S1, Data Collection Phase: Through vehicle sensors and fault diagnosis system memory, fault light signals, real-time vehicle operating parameters and historical fault data are collected to build a fault light description information database. S2, Model Training Phase: Using the fault light instruction information database as the learning samples, a large fault recognition model is trained to enable it to recognize fault light signals, analyze faults, and generate solutions. S3, System Integration Stage: The trained and optimized fault identification model is integrated into the vehicle fault diagnosis system, and coordinated with the vehicle electronic control system and the central control screen system. S4, Fault Handling Phase: When the fault light illuminates, the fault light identification and clearing process is automatically triggered. S5, the personalized adaptation stage, records the driver's driving habits and vehicle usage, and adjusts the fault diagnosis priority and solution recommendation logic accordingly.

2. The method for eliminating automotive fault lights based on a large model as described in claim 1, characterized in that: The specific implementation steps of S1 are as follows: S11, Deploy data acquisition terminals: Deploy data acquisition terminals in vehicle sensors and fault diagnosis system memory to establish data transmission links; S12, Multi-type data acquisition: Simultaneously acquire fault light signals, real-time vehicle operating parameters, and historical fault data through the acquisition terminal; S13, Data Cleaning and Verification: Remove outliers from the collected data, standardize the format, and verify the integrity and validity of the data; S14, Database Construction: Based on the verified data, build a fault light description information database, and input fault light diagrams, safety level classifications and corresponding parameters; S15, Data entry and storage: Classify and store the collected valid data and basic database information into the database, and complete the connection and integration between the data and the database.

3. The method for eliminating automotive fault lights based on a large model as described in claim 1, characterized in that: The specific implementation steps of S2 are as follows: S21, Sample processing: Extract the data collected in S1, analyze historical fault information, and generate standardized learning samples including fault light signals, operating parameters, and fault labels; S22, Sub-model 1 training: Train a multimodal fusion sub-model based on the samples to be learned, and extract feature fusion and root cause association features including fault light signals and vehicle operation data; S23, Sub-model 2 training: Based on the fusion features output by sub-model 1, train the fusion feature-root cause association weighted probability calibration sub-model to determine the fault level; S24, Joint Training of Large Models: Jointly optimize the trained twin models to generate a large fault identification model, and optimize the model output logic by combining fault handling decision criteria. S25, Model Performance Verification: Verify the large model using test samples to confirm its ability to identify fault light signals, analyze faults, and generate solutions. If it fails to meet the requirements, return to S22 for iterative training.

4. The method for eliminating automotive fault lights based on a large model as described in claim 3, characterized in that: The specific implementation steps of the S22 sub-model are as follows: S221, receiving the data verified by S1 includes setting the fault light signal. Real-time operating parameters Historical fault data Output a fusion feature matrix with root cause correlation. : in: For fault light signal Modal weights, For real-time running parameters Modal weights, Historical fault data The modal weights are all obtained by statistical fitting from the historical fault data collected by S1 and by calculating the contribution ratio of each data mode in the fault determination. For fault light signal The correlation coefficient, For fault light signal The correlation coefficient, For fault light signal The correlation coefficients were all obtained through joint training from the S1 fault light description information database and historical fault data. Represented as element-wise product; S222, Integrating Feature Filtering and Standardization: For Core features are selected and then normalized to obtain a standardized fusion feature matrix. ': in, , Output for S221 Minimum and maximum values; S223, Training Sample Reconstruction: Based on ', associate the safety level label y with the S1 fault light description information database, and reconstruct the exclusive training sample set for S23. ; S224, Sample Splitting and Input: Divide the reconstructed sample set into a training set and a validation set in a 7:3 ratio, and start training sub-model 2.

5. The method for eliminating automotive fault lights based on a large model as described in claim 4, characterized in that: The specific implementation steps of the S23 sub-model are as follows: S231, Model Initialization: Set the root cause triggering influence coefficient ξ, which is obtained from the historical fault data and safety level classification in S1. Define the model loss function as cross-entropy loss L. S232, Forward calculation of risk probability: Substitute the training set input from S224 into the core calibration formula, and calculate the failure risk probability p for each sample: in, This represents the weight coefficient of the k-th data mode. Represents the correlation coefficient of the k-th data mode; S233, Loss Calculation and Backward Parameter Tuning: Substitute the P obtained from the forward calculation and the training set label y into the loss function to calculate the loss value L, and then backpropagate through the gradient descent method to fine-tune the model adaptation parameters. Where n represents the total amount of data in the sample set, and i represents the index of the data in the sample set. This represents the grade label of the i-th sample. This represents the failure risk probability of the i-th sample; S234, Validation of validation set effect: Substitute the validation set input in S224 into the optimized model, calculate the validation set risk probability, match the validation set labels, verify the accuracy of the model's fault level determination, and converge and solidify the model after meeting the preset threshold.

6. The method for eliminating automotive fault lights based on a large model as described in claim 1, characterized in that: The specific implementation steps of S3 are as follows: S31, Integrated Environment Setup: Build a dedicated integrated environment for the automotive fault diagnosis system, adapt to the computing power specifications of vehicle hardware, complete the communication protocol docking with the vehicle electronic control system and central control screen system, and reserve interfaces for large model data interaction and command transmission. S32, Model Deployment and Loading: Solidify and deploy the large fault recognition model trained and optimized by S2 to the automotive fault diagnosis system to complete the lightweight adaptation of the model. S33, data link is established, establishing a two-way data link between the fault diagnosis system and the S1 acquisition terminal, vehicle electronic control system, and central control screen system, which enables real-time transmission of acquired data to the model and the issuance of model commands to the vehicle system; S34, functional module linkage debugging, verifies the model recognition, analysis, and level determination capabilities, as well as the display of diagnostic results to the central control screen and the execution effect of control commands to the vehicle electronic control system.

7. The method for eliminating automotive fault lights based on a large model as described in claim 1, characterized in that: The specific implementation steps of S4 are as follows: S41, the large model automatically recognizes the fault light signals on the instrument panel and the central control screen; S42, combine real-time vehicle operating status and historical fault data to analyze possible causes of the fault; S43 generates the fault level and corresponding solution, which is then displayed to the driver via the central control screen. S44: If the fault does not require repair, a control command is generated and the fault light is automatically cleared through the vehicle electronic control system; if repair is required, the system confirms that the fault has been eliminated and clears the fault light after the driver completes the repair; by continuously analyzing vehicle operating data, potential faults are predicted and the driver is reminded in advance through the central control screen.

8. The method for eliminating automotive fault lights based on a large model as described in claim 1, characterized in that: The specific implementation steps of S5 are as follows: S51, Continuous data collection: The large model records the driver's driving habits and vehicle usage in real time; S52, Data Feature Extraction: Extract features from recorded data and correlate them with historical fault diagnosis and solution feedback information; S53, Adaptation Logic Adjustment: Based on feature analysis, the logic for prioritizing fault diagnosis and recommending solutions is optimized in a targeted manner; S54, Adaptation effect verification: Combined with subsequent fault handling scenarios, verify the adaptability and practicality of the adjusted logic, and optimize the personalized adaptation logic.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it causes the processor to perform the steps of the method as described in any one of claims 1 to 8.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the computer program is executed by the processor, it causes the processor to perform the steps of the method as described in any one of claims 1 to 8.

Images