Vehicle fault diagnosis method and device, electronic equipment and storage medium

Abstract

The application provides a vehicle fault diagnosis method and device, electronic equipment and storage medium, which are suitable for the technical field of vehicle fault diagnosis. The method comprises the following steps: acquiring vehicle fault information; adopting a preset core knowledge model to diagnose the vehicle fault information, and acquiring a diagnosis knowledge tuple; the diagnosis knowledge tuple comprises a multi-dimensional and logically associated structured knowledge object; and adopting a preset localization adaptation model to perform target language adaptation processing on the diagnosis knowledge tuple, and acquiring a target diagnosis result. Through the double-AI collaborative architecture of the "core knowledge model" and the "localization adaptation model", the dynamic intelligent generation and adaptation mode of the double-AI model collaboration is adopted, so that a solution with higher diagnosis quality and stronger automation degree is provided for the globalization of AI diagnosis.

Description

Technical Field

[0001] This application relates to the technical field of vehicle fault diagnosis, and in particular to a vehicle fault diagnosis method, device, electronic equipment and storage medium. Background Technology

[0002] In the global wave of intelligent and connected vehicles, artificial intelligence has become a core driving force for improving the efficiency and accuracy of vehicle diagnostics. Currently, AI (Artificial Intelligence) models trained on large-scale diagnostic data can quickly interpret fault codes and provide preliminary repair guidance, significantly lowering the technical threshold. However, when these advanced diagnostic AI models need to serve the global market, a fundamental contradiction immediately emerges: on the one hand, in the source language (e.g., Chinese) market, AI models rely on massive amounts of high-quality data, and their diagnostic suggestions have reached a high level of professionalism and completeness; on the other hand, when deployed to overseas markets, due to the extreme scarcity of professional diagnostic corpora in the target language, the output quality of the model drops sharply, resulting in lower diagnostic accuracy and making it difficult for local repair technicians to effectively trust and use it.

[0003] Moreover, traditional methods that rely on teams of human translation experts or general machine translation tools also suffer from low efficiency, making it difficult for AI models to be widely adopted due to language issues, and their core value cannot be fully realized globally. Summary of the Invention

[0004] In view of this, embodiments of this application provide a vehicle fault diagnosis method, apparatus, electronic device, and storage medium to solve the problem that existing AI models for vehicle fault diagnosis are difficult to promote due to language issues.

[0005] The first aspect of this application provides a vehicle fault diagnosis method, including: Obtain vehicle fault information; The system uses a pre-defined core knowledge model to diagnose vehicle fault information and obtain diagnostic knowledge tuples. The diagnostic knowledge tuples include multi-dimensional and logically related structured knowledge objects. A pre-defined localization adaptation model is used to adapt the diagnostic knowledge tuples to the target language to obtain the target diagnostic results.

[0006] In one possible implementation, after using a pre-defined localization adaptation model to adapt the diagnostic knowledge tuples to the target language and obtaining the target diagnostic results, the following steps are also included: A localized adaptation model is used to verify the target diagnostic results and obtain a quality confidence score; the quality confidence score is used to characterize the reliability of the target diagnostic results.

[0007] In one possible implementation, after verifying the target diagnostic results using a localized adaptation model and obtaining a quality confidence score, the following steps are also included: If the quality confidence score is lower than the preset threshold, the target diagnostic result will be sent to the calibration workbench. In response to receiving calibration results from the calibration workbench for the target diagnostic results, the core knowledge model and / or localized adaptation model are adjusted based on the calibration results.

[0008] In one possible implementation, the core knowledge model and / or the localization adaptation model are adjusted based on the calibration results, including: Based on the calibration results, at least one correction information is determined; Based on each correction information, determine the correction type corresponding to that correction information; Based on each correction type and the corresponding correction information, the core knowledge model or localization adaptation model is adjusted.

[0009] In one possible implementation, after using a pre-defined localization adaptation model to adapt the diagnostic knowledge tuples to the target language and obtaining the target diagnostic results, the following steps are also included: In response to a labeling operation on the target diagnostic result, obtain the target diagnostic result with the labeling information; Send the target diagnostic results with tagged information to the calibration workbench; In response to receiving calibration results from the calibration workbench for target diagnostic results with tagged information, the core knowledge model and / or localized adaptation model are adjusted based on the calibration results.

[0010] In one possible implementation, a pre-defined localization adaptation model is used to adapt the diagnostic knowledge tuples to the target language to obtain the target diagnostic results, including: A localization adaptation model is used to convert the diagnostic knowledge tuples into the target language to obtain the diagnostic results to be processed. Insert compliance requirement information into the pending diagnostic results to obtain the target diagnostic results; the compliance requirement information is obtained through the relevant regulations of the region corresponding to the target language.

[0011] In one possible implementation, vehicle fault information includes fault codes, and the method also includes: When updating the diagnostic logic data for any fault code in the core knowledge base, an incremental update package for the fault code in multiple languages ​​is generated. The incremental update package includes the updated diagnostic logic data for the fault code, and the diagnostic logic data in the core knowledge base is used for training the core knowledge model.

[0012] A second aspect of this application provides a vehicle fault diagnosis device, comprising: The acquisition module is used to acquire vehicle fault information; The first diagnostic module is used to diagnose vehicle fault information using a preset core knowledge model and obtain diagnostic knowledge tuples; the diagnostic knowledge tuples include multi-dimensional and logically related structured knowledge objects; The second diagnostic module is used to adapt the diagnostic knowledge tuples to the target language using a preset localization adaptation model to obtain the target diagnostic results.

[0013] A third aspect of this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method of the first aspect.

[0014] A fourth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the method of the first aspect.

[0015] Compared with the prior art, the embodiments of this application have at least the following technical effects: The vehicle fault diagnosis method of the first aspect of this application can acquire vehicle fault information, then diagnose the vehicle fault information using a preset core knowledge model to obtain a diagnostic knowledge tuple; subsequently, it uses a preset localization adaptation model to adapt the diagnostic knowledge tuple to a target language to obtain a target diagnostic result. Since the diagnostic knowledge tuple includes multi-dimensional and logically related structured knowledge objects, and is rich in logical relationships (such as causality, sequence, and subordination), the target diagnostic result output in the target language after the localization adaptation model adapts the diagnostic knowledge tuple is more accurate. Therefore, this application embodiment, through a dual-AI collaborative architecture of "core knowledge model" and "localization adaptation model," and employing a dynamic intelligent generation and adaptation method of dual AI model collaboration, provides a solution with higher diagnostic quality and greater automation for the globalization of AI diagnosis.

[0016] It is understood that the beneficial effects of the second to fourth aspects mentioned above can be found in the relevant descriptions in the first aspect mentioned above, and will not be repeated here. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a vehicle fault diagnosis method provided in an embodiment of this application; Figure 2 This is a flowchart of another vehicle fault diagnosis method provided in the embodiments of this application; Figure 3 This is a schematic diagram of the structure of a vehicle fault diagnosis device provided in an embodiment of this application; Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0019] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0020] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0021] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0022] In the description of this application, unless otherwise stated, the " / " used in this specification and appended claims indicates that the related objects are in an "or" relationship. For example, A / B can mean A or B. The "and / or" in this application merely describes the relationship between the related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. Furthermore, in the description of this application, unless otherwise stated, "multiple" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can represent: a, b, c, a and b, a and c, b and c, or a, b, and c. Here, a, b, and c can be single or multiple.

[0023] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0024] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0025] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0026] Research has revealed that current mainstream AI diagnostic solutions largely follow a "single-model" paradigm. Regardless of scale, their knowledge boundaries and language capabilities are limited by the data corpus used during training. For less common languages ​​or specific regions, the cost and time investment required to collect sufficient structured text data—including fault codes, deep diagnostic logic, repair procedures, and safety regulations—to support the training of a large-scale professional diagnostic model is prohibitive for most companies. Therefore, the industry generally faces a critical bottleneck: how to efficiently, faithfully, and compliantly translate proven professional diagnostic knowledge from a specific language region (such as China) into high-quality content applicable to other language regions. Traditional methods relying on human translation expert teams or general machine translation tools are not only inefficient but also fail to address deeper issues such as consistency of terminology, preservation of diagnostic logic context, and local adaptation. This severely hinders the globalization of diagnostic AI, preventing its core value from being fully realized globally.

[0027] The vehicle fault diagnosis method, device, electronic equipment, and storage medium provided in this application are intended to solve the above-mentioned technical problems of the prior art.

[0028] The technical solution of this application and how it solves the above-mentioned technical problems are described in detail below with specific embodiments. It should be noted that the following embodiments can be referenced, borrowed, or combined with each other, and the same terms, similar features, and similar implementation steps in different embodiments will not be described again.

[0029] See Figure 1 As shown, this application provides a flowchart of a vehicle fault diagnosis method. Figure 1 As shown, the vehicle fault diagnosis method includes steps S101 to S103.

[0030] S101. Obtain vehicle fault information.

[0031] Vehicle fault information refers to information related to vehicle malfunctions, which may include fault codes. Fault codes are unique digital identifiers (UDIC) used by onboard diagnostic systems to recognize fault conditions and help technicians quickly locate and resolve vehicle problems.

[0032] S102. Use a preset core knowledge model to diagnose vehicle fault information and obtain diagnostic knowledge tuples; the diagnostic knowledge tuples include multi-dimensional and logically related structured knowledge objects.

[0033] Among them, the core knowledge model is a pre-trained AI model used for vehicle fault diagnosis.

[0034] Optionally, the core knowledge model can transform unstructured diagnostic experience into machine-understandable and transferable structured knowledge objects. As a "domain expert," when receiving a fault code query request, the core knowledge model does not simply retrieve text, but performs deep reasoning and structured organization. It analyzes and outputs a "structured knowledge object" containing multi-dimensional information and logical connections. For example, it associates the "P0300 - random / multi-cylinder misfire" fault with multiple possible cause chains such as "ignition coil aging," "fuel injector blockage," and "vacuum leak," and binds corresponding diagnostic steps and key data flow checkpoints to each cause chain.

[0035] Specifically, the core knowledge model can achieve cross-dimensional correlations based on a deep understanding of the fault mechanism. For example, for the "ABS (Anti-lock Braking System) malfunction indicator light is on", the core knowledge model can not only output possible causes, but also correlate them with "affected systems" (braking system, wheel speed sensor circuit), "corresponding data streams that must be checked" (four wheel speed signals, lateral acceleration sensor values), and "necessary post-repair operations" (conducting a road test to activate and match the ABS module).

[0036] S103. Adapt the diagnostic knowledge tuples to the target language using a preset localization adaptation model to obtain the target diagnostic results.

[0037] The localization adaptation model is a pre-trained AI model that can convert diagnostic knowledge tuples into target diagnostic results in the target language. The target language is the language of a predefined region, i.e., the localized language.

[0038] In some embodiments, a preset localization adaptation model is used to adapt the diagnostic knowledge tuples to the target language to obtain the target diagnostic results, including: A localization adaptation model is used to convert the diagnostic knowledge tuples into the target language to obtain the diagnostic results to be processed. Insert compliance requirement information into the pending diagnostic results to obtain the target diagnostic results; the compliance requirement information is obtained through the relevant regulations of the region corresponding to the target language.

[0039] In some embodiments, after adapting the diagnostic knowledge tuples to the target language using a preset localization adaptation model and obtaining the target diagnostic results, the method further includes: A localized adaptation model is used to verify the target diagnostic results and obtain a quality confidence score; the quality confidence score is used to characterize the reliability of the target diagnostic results.

[0040] The localization adaptation model in this application ensures that the professionalism, security, and usability of knowledge are not degraded when it crosses languages ​​and cultures. The intelligent translation and deep verification functions of the localization adaptation model include: 1. Generation of expressions that conform to local technical culture: The model will apply the technical rhetoric habits of the target market. For example, the more general phrase "check the circuit" in Chinese will be translated into the specific and rigorous phrase "use a multimeter to measure the resistance of the wires and the voltage to ground, and check whether the plug pins are corroded or deformed" in German.

[0041] 2. Secure dynamic compliance verification: The model's embedded rule engine scans maintenance actions within structured knowledge objects and matches them with relevant regulations for the target region. For example, when a knowledge object contains "disposing of waste engine oil," the output for certain regions will automatically include a mandatory clause stating that "it must be handled by a licensed waste oil collector."

[0042] 3. Terminology Consistency Management: When generating lengthy diagnostic guidelines, the model ensures that the same component or concept uses completely consistent terminology throughout the text to avoid confusion and enhance the professionalism of the content.

[0043] The localization adaptation model, acting as a "localization expert," receives "structured knowledge objects" from upstream sources. Its task is not simple translation, but rather "knowledge re-creation." Based on the rules and terminology of the target market, it performs the following tasks: 1) Technical context translation: converting professional descriptions into expressions familiar to local technicians; 2) Compliance embedding: automatically inserting relevant safety warnings (such as "high-voltage power must be disconnected first") and environmental requirements into appropriate locations; 3) Logical fluency polishing: ensuring the generated text conforms to the reading logic of local technical documents. Finally, it self-verifies the output content and generates a localization quality confidence score.

[0044] Based on steps S101 to S103 above, the vehicle fault diagnosis method of this application embodiment can acquire vehicle fault information, then use a preset core knowledge model to diagnose the vehicle fault information and obtain a diagnostic knowledge tuple; furthermore, it uses a preset localization adaptation model to adapt the diagnostic knowledge tuple to the target language and obtain the target diagnostic result. Since the diagnostic knowledge tuple includes multi-dimensional and logically related structured knowledge objects, and is rich in logical relationships (such as causality, sequence, and subordination), the target diagnostic result output in the target language after the localization adaptation model adapts the diagnostic knowledge tuple is more accurate. Therefore, this application embodiment, through a dual-AI collaborative architecture of "core knowledge model" and "localization adaptation model," adopts a dynamic intelligent generation and adaptation method of dual AI model collaboration, providing a solution with higher diagnostic quality and greater automation for the globalization of AI diagnosis.

[0045] The technical solution of this application does not aim to reconstruct a fully functional large model on the target language side, but rather creates an architecture of "specialized division of labor and collaboration". A first large model (core knowledge model) performs deep understanding and structured encapsulation on the source language side, while a second large model (localization adaptation model) performs deep adaptation and regeneration on the target language side, thereby bypassing the dependence on massive amounts of labeled target data. Therefore, this application embodiment can overcome the barrier of data scarcity, achieve high-fidelity cross-language transfer of professional knowledge, and overcome the fundamental dilemma of scarce target language corpora that prevent the training of local specialized models.

[0046] Furthermore, this application embodiment ensures that the knowledge units output from the source are structurally complete and logically autonomous through a core knowledge model. A localization adaptation model enables accurate translation based on the technical context, maintaining the coherence of diagnostic logic and proactively injecting mandatory content that conforms to the technical habits and security standards of the target market, thus resolving the issues of professional distortion, broken logical chains, and compliance blind spots caused by traditional methods.

[0047] This application presents a paradigm of "dynamic intelligent generation and adaptation based on the collaboration of two large AI models." The first large model (core knowledge model) is not a simple information extractor, but a "domain expert" with a deep understanding of the diagnostic knowledge system and the ability to express it in a structured manner. Its output is not a block of text to be translated, but a diagnostic knowledge tuple rich in logical relationships (such as causality, sequence, and subordination). The second large model (localization adaptation model) is a "localization expert" proficient in the target language's technical context. Its task is not word-for-word translation, but rather, based on the received structured knowledge, to "restate" it in accordance with local technical expression habits and proactively inject compliance clauses. The entire process is a bidirectional, verifiable, and iterative intelligent system.

[0048] In short, existing solutions attempt to solve language conversion problems using engineering methods, while the embodiments of this application use artificial intelligence methods to solve the problem of cross-cultural knowledge transfer and re-creation. The former deals with "words" and "sentences," while the latter deals with "meaning" and "context." The latter provides a new solution for the globalization of diagnostic AI that is of higher quality, more automated, and has continuous learning capabilities.

[0049] Furthermore, this application embodiment constructs a system capable of automatically, faithfully, and compliantly generating and adapting vehicle diagnostic knowledge across languages ​​through a dual AI collaborative architecture of a "core knowledge model" and a "localization adaptation model." This overcomes a series of shortcomings of existing technologies, such as dependence on target language data, low output quality, lack of compliance, difficulty in evolution, and fragile intellectual property management, truly empowering the global and efficient deployment of diagnostic AI services.

[0050] In some embodiments, after verifying the target diagnostic results using a localized adaptation model and obtaining a quality confidence score, the method further includes: If the quality confidence score is lower than the preset threshold, the target diagnostic result will be sent to the calibration workbench. In response to receiving calibration results from the calibration workbench for the target diagnostic results, the core knowledge model and / or localized adaptation model are adjusted based on the calibration results.

[0051] In some embodiments, after adapting the diagnostic knowledge tuples to the target language using a preset localization adaptation model and obtaining the target diagnostic results, the method further includes: In response to a labeling operation on the target diagnostic result, obtain the target diagnostic result with the labeling information; Send the target diagnostic results with tagged information to the calibration workbench; In response to receiving calibration results from the calibration workbench for target diagnostic results with tagged information, the core knowledge model and / or localized adaptation model are adjusted based on the calibration results.

[0052] This application embodiment features a "calibration center" and a continuously evolving "driving engine" to ensure the reliability of AI output in complex scenarios. This application embodiment automatically pushes output content with low quality confidence scores in the collaborative processing layer, or content actively marked as "questionable" by the user, to the expert calibration workbench. Domain experts (such as senior bilingual technicians) can see the AI's original output, the reasons for low confidence scores (such as "a certain security clause mismatch"), and can make corrections, confirmations, or additions in the workbench.

[0053] Then, all manual calibration operations and their results are automatically categorized as high-quality labeled data and fed back to different parts of the system: corrections to the diagnostic logic are used to optimize the "core knowledge model"; corrections to language and compliance are used to optimize the "localized adaptation model" and the "localized rule base". This forms a reinforced closed loop that drives the continuous improvement of dual-model capabilities.

[0054] See Figure 2 As shown, this application provides a flowchart of another vehicle fault diagnosis method. Figure 2 As shown, the vehicle fault diagnosis method includes steps S101 to S103.

[0055] S201. Obtain vehicle fault information.

[0056] S202. Use a preset core knowledge model to diagnose vehicle fault information and obtain diagnostic knowledge tuples; the diagnostic knowledge tuples include multi-dimensional and logically related structured knowledge objects.

[0057] S203. Use a preset localization adaptation model to adapt the diagnostic knowledge tuples to the target language and obtain the target diagnostic results.

[0058] Steps S201 to S203 in this embodiment are implemented in the same principle as steps S101 to S103 in this embodiment, and will not be repeated here.

[0059] S204. The target diagnostic results are validated using a localized adaptation model to obtain a quality confidence score; the quality confidence score is used to characterize the reliability of the target diagnostic results.

[0060] S205. If the quality confidence score is lower than the preset threshold, the target diagnostic result will be sent to the calibration workbench.

[0061] In practical applications, the implementation of precise calibration and feedback attribution functions in human-machine collaboration can include the following: 1. Intelligent task assignment based on quality confidence: The calibration workbench assigns tasks to the most suitable experts based on the type of problem and its quality confidence level. For example, German content involving the interpretation of complex circuit diagrams will be automatically pushed to German experts with an electrical engineering background.

[0062] 2. Structured Correction and Feedback Attribution: When experts make modifications on the calibration interface, they need to select the "Correction Type" through drop-down menus or labels, such as "Diagnostic Step Error," "Inaccurate Terminology," "Missing Safety Warning," or "Unconventional Expression." Based on the label, the correction type is determined, and the modified example is automatically categorized into the "Core Knowledge Optimization" or "Localization Optimization" training queue, achieving precise utilization of feedback.

[0063] S206. In response to receiving the calibration results returned by the calibration workbench for the target diagnostic results, adjust the core knowledge model and / or the localized adaptation model based on the calibration results.

[0064] In some embodiments, adjustments are made to the core knowledge model and / or the localization adaptation model based on the calibration results, including: Based on the calibration results, at least one correction information is determined; Based on each correction information, determine the correction type corresponding to that correction information; Based on each correction type and the corresponding correction information, the core knowledge model or localization adaptation model is adjusted.

[0065] This application also provides a physical or logical system for implementing the above method. The system includes: a first server carrying a core knowledge model, a second server carrying a localized adaptation model, and a collaborative control module for managing user feedback and optimization instructions. Particularly important to protect is the method by which these components communicate and collaborate through a specific structured data interface protocol.

[0066] The system's knowledge source and data foundation are responsible for the centralized storage, management, and updating of core diagnostic knowledge assets. The system also includes: Core Knowledge Base: Stores structured and standardized vehicle diagnostic fault code data, covering fault code descriptions, root causes, affected systems, standard diagnostic procedures, repair recommendations, and safety precautions. The core knowledge base provides the data foundation for training and querying the "core knowledge model."

[0067] Localization Rules and Terminology Library: This library stores localization rule sets for each target market, including mappings to safety standard provisions, industry-standard technical terms and expressions, and local common names for vehicle models / parts. This library provides the basis for validating and refining the "localization adaptation model."

[0068] User feedback data pool: Continuously collects and stores evaluations, correction suggestions, and usage data from global end users (repair technicians) on multilingual diagnostic content, providing fuel for the system's optimization loop.

[0069] This application embodiment can also design an efficient intermediate representation and interaction protocol between the two models, enabling the entire system to quickly respond to updates to the knowledge source and achieve scalable, automated pipeline output. A collaborative optimization closed loop based on user feedback is established. Feedback from overseas technicians is accurately attributed (whether it's a knowledge error or poor localization) and used for targeted fine-tuning of the two models, giving the system the ability to continuously improve itself. This solves the problems of high iteration costs and inability to learn in static systems, constructing a dynamic, evolvable, and highly efficient intelligent collaborative system.

[0070] In some embodiments, the vehicle fault information includes fault codes, and the method further includes: When updating the diagnostic logic data for any fault code in the core knowledge base, an incremental update package for the fault code in multiple languages ​​is generated. The incremental update package includes the updated diagnostic logic data for the fault code, and the diagnostic logic data in the core knowledge base is used for training the core knowledge model.

[0071] This application embodiment ensures that global users can obtain the latest and most accurate diagnostic support in real time. The dynamic publishing and updating function of the multilingual knowledge service in this application embodiment includes: incremental updates and real-time synchronization; when the diagnostic logic of a fault code is updated in the core knowledge base, the system can automatically trigger a regeneration process for all supported language content for that fault code, and push it to the terminal application in the form of an "incremental update package" through the publishing system, achieving rapid global knowledge synchronization.

[0072] This application embodiment can construct an automated transformation engine from unstructured operation sequences to structured, executable knowledge. The "delivery terminal" produced by the system is responsible for providing the processed, high-quality, multilingual diagnostic knowledge to global users in an efficient and convenient manner, including the following methods: Knowledge Service Portal: Provides standard interface services for overseas versions of diagnostic software, repair information portals, or technician community apps via Web API (Application Programming Interface) or integrated SDK (Software Development Kit). The service interface supports precise queries based on vehicle model, fault code, language, and other criteria.

[0073] Content Management and Publishing System: Supports version management, review and release, and A / B testing of generated multilingual content. Ensures the synchronization and consistency of knowledge updates; for example, when the core knowledge base is updated with diagnostic methods for a new electric vehicle, updated content in all supported languages ​​can be quickly triggered and released.

[0074] Based on the above technical solutions, the core technical problem that this application embodiment can solve is: how to overcome the obstacle of scarce target language professional data in the global deployment of vehicle diagnostic AI, realize the high-quality, compliant and localized automatic conversion and output of diagnostic knowledge from the source language to the target language, and build a sustainable, optimized, and centrally managed multilingual knowledge supply system with core intellectual property rights.

[0075] Specifically, this can be broken down into the following four progressive technical challenges: 1. The problem of “structured lossless extraction and transformation” of complex diagnostic knowledge from core models to target languages.

[0076] Diagnostic knowledge tuples are highly structured, multi-dimensional, and complex knowledge entities (description, cause, system, method, direction, security). Traditional methods (such as calling general translation APIs) can disrupt their inherent logical structure and consistency with technical terminology. This application aims to address how to design the output interface and protocol of the first large model (core knowledge model) to "package" and output multi-dimensional, strongly correlated knowledge units surrounding a specific fault code from the internal knowledge base in a clearly structured, well-defined, and terminologically consistent manner. This provides high-quality, complete "knowledge raw materials" for subsequent deep localization, rather than merely text fragments.

[0077] 2. The issue of "deep localization adaptation and compliance embedding" of diagnostic knowledge.

[0078] Simply translating structured Chinese diagnostic knowledge directly into the target language still results in issues such as technical terminology that is out of touch with local context and operational procedures that do not conform to relevant standards. This application's embodiments aim to address how to utilize a second major model (a localization adaptation model) to go beyond literal translation and achieve "accurate translation within the technical context," "adaptive refinement based on regional industry practices," and "dynamic verification and embedding of safety clauses." This requires the localization adaptation model to possess a deep understanding of the target market's automotive repair standards, language habits, and technical culture, and to perform logical judgments based on rules and knowledge to ensure that the output content is not only readable in the local context but also directly, safely, and legally usable.

[0079] 3. The problem of "high-fidelity knowledge transfer and efficiency" under the dual-model professional collaboration.

[0080] The collaborative operation of the two independent large models in this application embodiment faces the risk of knowledge distortion and bottlenecks in flow efficiency during knowledge transfer. This application embodiment needs to address how to design an efficient intermediate representation and interaction protocol to ensure that the structured knowledge output by the core knowledge model can be unambiguously understood and processed by the localized model. Simultaneously, it needs to optimize the calling process and responsibility boundaries between the two models, forming a standardized "extraction-conversion-verification" pipeline. This will enable the automation and scalability of multilingual content generation while ensuring output quality, supporting rapid deployment across multiple regions simultaneously.

[0081] 4. The issue of "precise collaborative optimization of dual models" based on user feedback.

[0082] The initially constructed dual-model system struggles to cover all complex and edge-of-the-road fault scenarios and localization details. This application's embodiments aim to address how to establish a human-machine collaborative optimization closed loop driven by user feedback: how to collect and analyze quality feedback from overseas maintenance technicians regarding multilingual diagnostic content (e.g., "inaccurate terminology," "invalid steps"); how to accurately determine whether the root cause of the problem should be attributed to the core knowledge model (diagnostic logic error) or the localization model (translation or adaptation error); and how to use the attributed feedback data for targeted fine-tuning and iteration of both models, thereby forming a reinforcing loop that allows the diagnostic knowledge base and localization capabilities to continuously co-evolve.

[0083] Therefore, the core technical problem of this application's embodiments lies in: addressing the scarcity of multilingual professional knowledge data in the field of vehicle diagnostics, designing an architecture and process that combines "core knowledge extraction + deep localization adaptation" with dual AI models to solve the challenge of automated conversion from a single-language knowledge base to high-quality, compliant multilingual output. Furthermore, through a human-machine feedback loop, the system achieves self-improvement, ultimately providing reliable, easy-to-use, and legal diagnostic decision support for repair technicians in different regions worldwide, while ensuring the centralization and security of core technology assets. Furthermore, the embodiments of this application address the core challenge of the global export of diagnostic knowledge. This is specifically reflected in the following four aspects: 1. Architectural innovation: The division of roles between domain experts and localization experts.

[0084] The core idea of ​​this application's embodiments is to decompose a complex task and have it completed collaboratively by two specialized models. The first large model acts as a "vehicle diagnostics expert," whose core capability is not translation, but rather deep understanding, reasoning, and extraction of structured diagnostic knowledge tuples from massive amounts of Chinese data. The second large model acts as a "target market localization expert," whose core capability is not generating diagnostic logic, but rather intelligently translating, refining, and enhancing the compliance of the received structured knowledge to conform to the local technical context and expression habits. This division of labor allows each model to achieve optimal performance in its respective area of ​​expertise.

[0085] 2. Data Flow Innovation: Standardized Structured Knowledge Representation and Interfaces.

[0086] To achieve seamless collaboration in the aforementioned division of labor, this application defines a "smart contract" for interaction between the two models, a machine-readable, language-neutral structured data format. This format mandates that the core knowledge model output not a paragraph of text, but a standard knowledge tuple that encapsulates multi-dimensional knowledge such as fault descriptions, root causes, diagnostic steps, and safety considerations, along with their internal logical relationships (e.g., causal and sequential). This ensures that the logic is not lost and the structure is not disintegrated during knowledge transmission, laying a solid foundation for high-quality localization.

[0087] 3. Process innovation: A closed-loop system that includes dual verification and continuous evolution.

[0088] This application embodiment constructs a dynamically optimized intelligent system. Internally, the localization adaptation model performs self-checks on terminology consistency, logical coherence, and compliance with keywords before output. Externally, the system establishes a mechanism for collecting and attributing user feedback, which can determine whether quality problems stem from diagnostic logic errors (requiring optimization of the core model) or poor localization (requiring optimization of the adaptation model). This drives the two models to undergo targeted and differentiated co-evolution, achieving continuous improvement in system performance.

[0089] 4. Deployment model innovation: a globalization model of "centralized control and distributed adaptation".

[0090] This application supports a strategic deployment approach. Core diagnostic knowledge assets (core knowledge models) can be centrally deployed, uniformly updated, and maintained, effectively ensuring the security and consistency of core technology intellectual property rights. Multiple localized adaptation models can be distributed and deployed in different regions. Both communicate through standardized interfaces, achieving an efficient and secure architecture of "one set of core knowledge serving the global market."

[0091] Furthermore, the embodiments of this application are based on a dual-AI large-scale model for multilingual output of DTC (Diagnostic Trouble Code), the core advantage of which stems from the architectural innovation of "specialized division of labor and collaboration". Compared with traditional single-model or simple translation solutions, the embodiments of this application have the following four significant advantages: 1. Fundamentally break through the bottleneck of scarcity of high-quality multilingual diagnostic data.

[0092] This application's embodiments eliminate the need for investing heavily in collecting and labeling massive amounts of diagnostic data in the target market to train a completely new, fully functional model. Through division of labor, the core knowledge model can deeply utilize mature, systematic, high-quality diagnostic databases, ensuring the authority and completeness of the knowledge source; the localization adaptation model focuses on its strongest task, "language conversion and local adaptation," and the difficulty and cost of acquiring its training data requirements (such as bilingual comparative corpora and local standard texts) are far lower than those of complete diagnostic data. This significantly reduces the data barriers and time costs associated with the global deployment of diagnostic AI technology.

[0093] 2. It has achieved "high fidelity" and "strong compliance" in the output of diagnostic knowledge.

[0094] Traditional methods often lead to the "distortion" of professional knowledge in translation. However, the embodiments of this application ensure the quality of the output content through dual safeguards: In terms of professionalism: the structured knowledge package output by the core knowledge model retains the integrity of the diagnostic logic (such as causal chains and step sequence), preventing information fragmentation from the source.

[0095] In terms of localization: the localization model's contextual refinement ensures that terminology and expressions conform to the reading habits of local technicians, and its embedded compliance checks automatically incorporate key local safety and environmental requirements into maintenance recommendations. The final output is no longer a rigid translation, but a guideline that can be directly and safely used in local maintenance practices.

[0096] 3. We have built an efficient and scalable global knowledge delivery pipeline.

[0097] The dual-model architecture and standardized interface of this application embodiment form an efficient "knowledge refinement and assembly line".

[0098] Highly efficient collaboration: Once the core knowledge model outputs a standardized knowledge structure, it can be quickly transferred to the localized model of any target language for processing, achieving the scalability effect of "analysis once, output in multiple languages".

[0099] Easy to maintain and expand: Updates to the core knowledge base (such as adding fault logic for new vehicle models) only need to be performed once on the core model side, and can be synchronized to all language versions. To expand into new language markets, the main task is to deploy or train a new localized adaptation model, without having to rebuild the entire diagnostic knowledge system, which greatly improves the system's scalability and maintenance efficiency.

[0100] 4. A virtuous cycle of sustainable evolution and centralized control of core intellectual property rights has been formed.

[0101] Continuous system evolution: Through a closed-loop user feedback system, the system continuously collects genuine evaluations from terminal repair technicians. By intelligent attribution (distinguishing between diagnostic logic issues and localization problems), feedback data can be precisely used to optimize corresponding models, enabling the entire system to continuously improve itself and enhance service quality.

[0102] The security and controllability of knowledge assets: The core diagnostic logic and knowledge graph are encapsulated in a centrally deployed core knowledge model, always remaining within a controllable range. Only the final "knowledge conclusions" are exported overseas, not the original, reverse-engineerable core rules and data. This provides high-quality localization services while maximizing the protection of the enterprise's core intellectual property assets, achieving a balance between value output and asset security.

[0103] In summary, the advantages of the embodiments of this application are interconnected, which not only solves the urgent problems of low quality and lack of compliance, but also builds long-term advantages at the strategic level of efficiency and cost, evolutionary capability and asset security.

[0104] See Figure 3 As shown in the diagram, this application provides a structural schematic diagram of a vehicle fault diagnosis device 30. Figure 3 As shown, the vehicle fault diagnosis device 30 includes: an acquisition module 301, a first diagnosis module 302, and a second diagnosis module 303.

[0105] The acquisition module 301 is used to acquire vehicle fault information.

[0106] The first diagnostic module 302 is used to diagnose vehicle fault information using a preset core knowledge model and obtain diagnostic knowledge tuples; the diagnostic knowledge tuples include multi-dimensional and logically related structured knowledge objects.

[0107] The second diagnostic module 303 is used to perform target language adaptation processing on the diagnostic knowledge tuples using a preset localization adaptation model to obtain the target diagnostic results.

[0108] Optionally, the second diagnostic module 303 is used to verify the target diagnostic results using a localized adaptation model and obtain a quality confidence score; the quality confidence score is used to characterize the reliability of the target diagnostic results.

[0109] Optionally, the vehicle fault diagnosis device 30 also includes a sending module and an adjustment module; The sending module is used to send the target diagnostic results to the calibration workbench if the quality confidence score is lower than a preset threshold.

[0110] The adjustment module is used to adjust the core knowledge model and / or localized adaptation model based on the calibration results returned by the calibration workbench for the target diagnostic results.

[0111] Optionally, the adjustment module is used to determine at least one correction information based on the calibration results; determine the correction type corresponding to each correction information based on each correction information; and adjust the core knowledge model or localization adaptation model based on each correction type and the correction type corresponding to the correction information.

[0112] Optionally, the acquisition module is used to acquire the target diagnostic result with tagging information in response to the tagging operation on the target diagnostic result. The sending module is used to send the target diagnostic result with tagging information to the calibration workbench. The adjustment module is used to adjust the core knowledge model and / or the localized adaptation model based on the calibration result returned by the calibration workbench for the target diagnostic result with tagging information in response to the calibration result received.

[0113] Optionally, the second diagnostic module 303 is used to convert the diagnostic knowledge tuples into the target language using a localization adaptation model to obtain the diagnostic result to be processed; to insert compliance requirement information into the diagnostic result to be processed to obtain the target diagnostic result; the compliance requirement information is obtained through the relevant normative content of the region corresponding to the target language.

[0114] Optionally, the vehicle fault information includes fault codes, and the vehicle fault diagnosis device 30 also includes a generation module. The generation module is used to generate an incremental update package of the fault code in multiple languages ​​when updating the diagnostic logic data of any fault code in the core knowledge base. The incremental update package includes the diagnostic logic data after the fault code is updated, and the diagnostic logic data of the core knowledge base is used for training the core knowledge model.

[0115] In application, the modules in the vehicle fault diagnosis device 30 can be software program modules, or they can be implemented by different logic circuits integrated in the processor, or they can be implemented by multiple distributed processors.

[0116] The vehicle fault diagnosis device 30 of this application embodiment can execute the method provided in this application embodiment. The implementation principle is similar. The actions performed by each module in the vehicle fault diagnosis device 30 of each embodiment of this application correspond to the steps in the method of each embodiment of this application. For detailed functional descriptions of each module of the vehicle fault diagnosis device 30, please refer to the descriptions in the corresponding methods shown above, which will not be repeated here.

[0117] See Figure 4 As shown, this application provides a schematic diagram of the structure of an electronic device 40. Figure 4 As shown, the electronic device 40 of this application embodiment includes: a memory 42, a processor 41, and a computer program 43 stored in the memory 42 and executable on the processor 41. When the processor 41 executes the computer program, it implements the steps of the methods of the various embodiments of this application.

[0118] Electronic device 40 can be a computing device such as a desktop computer, laptop, handheld computer, or cloud server. Electronic device 40 may include, but is not limited to, a processor 41 and a memory 42. Those skilled in the art will understand that electronic device 40 may also include more or fewer components, or combinations of certain components, or different components, such as input / output devices, network access devices, etc.

[0119] The processor 41 can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0120] In some embodiments, memory 42 may be an internal storage unit, such as a hard disk or RAM. Memory 42 may be a removable / non-removable, volatile / non-volatile computer system storage medium; for example, memory 42 may be a non-volatile memory used for reading and writing non-volatile magnetic media. In other embodiments, memory 42 may be an external storage device, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., provided on electronic device 40. Memory 42 is used to store operating systems, applications, bootloaders, data, and other programs, such as program code for computer programs. Memory 42 may also be used to temporarily store data that has been output or will be output.

[0121] It should be noted that the information interaction and execution process between the above-mentioned devices / units are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, and they will not be repeated here.

[0122] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments 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. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0123] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps in the above-described method embodiments.

[0124] If the integrated units described above are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying computer program code to a device / terminal equipment, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks.

[0125] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk drive (HDD), or solid-state drive (SSD), etc. The storage medium can also include combinations of the above types of memory.

[0126] This application provides a computer program product that, when run on a processor, enables the processor to execute the steps described in the various method embodiments above.

[0127] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0128] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0129] In the embodiments provided in this application, it should be understood that the disclosed apparatus / network devices and methods can be implemented in other ways. For example, the apparatus / network device embodiments described above are merely illustrative. For instance, the division of modules or units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0130] The units described above as separate components may or may not be physically separate. The components shown as units 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.

[0131] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A vehicle fault diagnosis method, characterized in that, include: Obtain vehicle fault information; The vehicle fault information is diagnosed using a preset core knowledge model to obtain diagnostic knowledge tuples; the diagnostic knowledge tuples include multi-dimensional and logically related structured knowledge objects. The diagnostic knowledge tuples are adapted to the target language using a pre-defined localization adaptation model to obtain the target diagnostic results.

2. The vehicle fault diagnosis method according to claim 1, characterized in that, After adapting the diagnostic knowledge tuples to the target language using a preset localization adaptation model to obtain the target diagnostic results, the method further includes: The localized adaptation model is used to verify the target diagnostic results and obtain a quality confidence score; the quality confidence score is used to characterize the reliability of the target diagnostic results.

3. The vehicle fault diagnosis method according to claim 2, characterized in that, After verifying the target diagnostic results using the localized adaptation model and obtaining the quality confidence score, the process further includes: If the quality confidence score is lower than a preset threshold, the target diagnostic result will be sent to the calibration workbench. In response to receiving the calibration result returned by the calibration workbench for the target diagnostic result, the core knowledge model and / or the localization adaptation model are adjusted based on the calibration result.

4. The vehicle fault diagnosis method according to claim 3, characterized in that, The adjustment of the core knowledge model and / or the localization adaptation model based on the calibration results includes: Based on the calibration results, at least one correction information is determined; Based on each piece of correction information, determine the correction type corresponding to the correction information; Based on each correction type and the correction type corresponding to the correction information, the core knowledge model or the localization adaptation model is adjusted.

5. The vehicle fault diagnosis method according to any one of claims 1-4, characterized in that, After adapting the diagnostic knowledge tuples to the target language using a preset localization adaptation model to obtain the target diagnostic results, the method further includes: In response to a labeling operation on the target diagnostic result, a target diagnostic result with labeling information is obtained; Send the target diagnostic results with tagged information to the calibration workbench; In response to receiving the calibration result returned by the calibration workbench for the target diagnostic result with tagged information, the core knowledge model and / or the localization adaptation model are adjusted based on the calibration result.

6. The vehicle fault diagnosis method according to any one of claims 1-4, characterized in that, The step of adapting the diagnostic knowledge tuples to the target language using a preset localization adaptation model to obtain the target diagnostic results includes: The localization adaptation model is used to convert the diagnostic knowledge tuples into the target language to obtain the diagnostic results to be processed. Compliance requirement information is inserted into the diagnostic results to be processed to obtain the target diagnostic results; the compliance requirement information is obtained through the relevant regulations of the region corresponding to the target language.

7. The vehicle fault diagnosis method according to any one of claims 1-4, characterized in that, The vehicle fault information includes fault codes, and the method further includes: When updating the diagnostic logic data of any fault code in the core knowledge base, an incremental update package in multiple languages ​​for the fault code is generated; the incremental update package includes the updated diagnostic logic data of the fault code, and the diagnostic logic data of the core knowledge base is used for training the core knowledge model.

8. A vehicle fault diagnosis device, characterized in that, include: The acquisition module is used to acquire vehicle fault information; The first diagnostic module is used to diagnose the vehicle fault information using a preset core knowledge model and obtain diagnostic knowledge tuples; the diagnostic knowledge tuples include multi-dimensional and logically related structured knowledge objects. The second diagnostic module is used to perform target language adaptation processing on the diagnostic knowledge tuples using a preset localization adaptation model to obtain the target diagnostic results.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 7.

Images