Fault diagnosis methods, devices, electronic equipment and storage media

Abstract

This application discloses a fault diagnosis method, apparatus, electronic device, and storage medium, comprising: acquiring a fault case of a vehicle to be diagnosed; extracting a fault phenomenon description from the fault case; matching the fault phenomenon description with a pre-created fault database; determining the fault phenomenon matching the fault phenomenon description and the target fault diagnosis model corresponding to the fault phenomenon; if the fault case contains a diagnostic fault code, determining the diagnosis time range of the fault case based on the diagnostic fault code; otherwise, determining the diagnosis time range of the fault case based on historical diagnostic fault codes associated with the fault case; acquiring vehicle operating data within the diagnosis time range; and performing fault diagnosis on the vehicle operating data using the target fault diagnosis model to obtain a fault diagnosis result. This application automatically selects a fault diagnosis model and locks the diagnosis time range based on the diagnostic fault code, thereby improving the efficiency and accuracy of fault diagnosis.

Description

Technical Field

[0001] This application belongs to the field of vehicle technology, and specifically relates to a fault diagnosis method, device, electronic device and storage medium. Background Technology

[0002] As the level of intelligence in automobiles continues to increase, the complexity of vehicle systems is growing daily. At the same time, a massive number of vehicle-related market failure cases are continuously accumulating through various channels such as direct user feedback and repair station reports. Therefore, how to diagnose and review these failure cases has become a core technical challenge for the automotive industry to improve product quality and optimize user experience.

[0003] Currently, when a fault case reflecting a vehicle market problem is received, the case is usually an unstructured text containing information such as the vehicle market problem and user description. It is first judged by experienced after-sales experts. In order to locate the anomaly from the massive amount of data uploaded by the vehicle, it is necessary to determine the time range of data analysis. Since the time of the fault occurrence is often unknown or ambiguous, in order to ensure that no key anomalies are missed, the widest possible time window is generally adopted, and the fault case is manually associated with one or more preset fault diagnosis models. The fault diagnosis model analyzes the vehicle data within a wide time period. However, this process relies entirely on the personal experience and knowledge of the after-sales expert team. The fault diagnosis model decided manually and the wide time window are difficult to accurately cover the anomalies reflected by the fault case, resulting in a waste of computing resources and affecting the execution efficiency of diagnostic analysis. Summary of the Invention

[0004] The purpose of this application is to provide a fault diagnosis method, device, electronic device, and storage medium that can solve the problems of current fault diagnosis models that rely on manual decision-making and have a wide time window, which make it difficult to accurately cover the abnormal problems reflected by fault cases, resulting in a waste of computing resources and affecting the execution efficiency of diagnostic analysis.

[0005] To solve the above-mentioned technical problems, this application is implemented as follows: In a first aspect, embodiments of this application provide a fault diagnosis method, the method comprising: Obtain vehicle fault cases to be diagnosed, and extract fault phenomenon descriptions from the fault cases; The fault phenomenon description is matched with a pre-created fault database to determine the fault phenomenon that matches the fault phenomenon description and the target fault diagnosis model corresponding to the fault phenomenon; wherein, the fault database includes the mapping relationship between natural language fault phenomenon descriptions and fault diagnosis models. If the fault case contains a diagnostic fault code, the diagnostic time range of the fault case is determined based on the diagnostic fault code; otherwise, the diagnostic time range of the fault case is determined based on the historical diagnostic fault codes associated with the fault case. The vehicle operation data within the diagnostic time range is acquired, and the target fault diagnosis model is used to diagnose the fault in the vehicle operation data to obtain the fault diagnosis result.

[0006] Based on the above technical means, semantic analysis is performed on fault cases to automatically select fault phenomena and fault diagnosis models that match the fault cases. The diagnosis time range is accurately locked based on the diagnostic fault codes. There is no need for manually decided fault diagnosis models and wide time windows, which reduces the cost of manual analysis and computing resources. The matching fault diagnosis model is used to automatically perform diagnosis within the diagnosis time range, so as to achieve accurate and comprehensive coverage of the abnormal problems reflected by the fault cases and greatly improve the efficiency of fault diagnosis.

[0007] Optionally, the fault database is created in the following manner: Obtain the tree structure information of the fault diagnosis model used to perform fault diagnosis; wherein, the fault diagnosis model includes a root node and multiple leaf nodes under the root node; For each leaf node, the diagnostic path information from each leaf node to the root node is input into a preset large language model to obtain one or more natural language fault phenomenon descriptions of the leaf node corresponding to the diagnostic path output by the large language model. The natural language description of the fault phenomenon is converted into a first vector; Establish a mapping relationship between the first vector, the natural language description of the fault phenomenon, and the corresponding fault diagnosis model, and store it to construct a fault database.

[0008] Based on the above technical means, the natural language description of the fault phenomenon is generated by using a large language model, which improves the accuracy of the fault phenomenon description. The natural language description of the fault phenomenon is associated with the fault diagnosis model and stored in a vector database, so as to quickly and automatically find the fault phenomenon most similar to the new fault case and the corresponding fault diagnosis model, thereby improving the efficiency of fault diagnosis and reducing the cost of manual intervention.

[0009] Optionally, the step of matching the fault phenomenon description with a pre-created fault database to determine the fault phenomenon that matches the fault phenomenon description and the target fault diagnosis model corresponding to the fault phenomenon includes: Obtain the first vector of the natural language description of the fault phenomena in the pre-created fault database; The fault phenomenon description is converted into a second vector, the similarity between the first vector and the second vector is calculated, and the fault phenomenon that matches the fault phenomenon description is determined. Based on the mapping relationships stored in the fault database, a target fault diagnosis model corresponding to the fault phenomenon that matches the fault phenomenon description is determined.

[0010] Based on the above technical means, semantic analysis is performed on fault cases, and fault phenomena and fault diagnosis models that match the fault cases are automatically selected. The fault phenomena and corresponding fault diagnosis models that are most similar to newly added fault cases are quickly and automatically found, thereby improving the efficiency of fault diagnosis.

[0011] Optionally, if the fault case contains a diagnostic fault code, the diagnostic time range of the fault case is determined based on the diagnostic fault code; otherwise, the diagnostic time range of the fault case is determined based on historical diagnostic fault codes associated with the fault case, including: If the fault case contains a diagnostic fault code, then the first diagnostic time range is determined based on the time window centered on the trigger timestamp of the diagnostic fault code; Otherwise, obtain at least one historical diagnostic fault code associated with the fault case within the target historical time period, and determine the second diagnostic time range based on the at least one historical diagnostic fault code; The first diagnostic time range or the second diagnostic time range shall be used as the diagnostic time range for the fault case.

[0012] Based on the above technical means, the diagnostic time range is determined based on the diagnostic fault codes, accurately locating the time of fault occurrence and avoiding the problem of the time range being too large or too small. For fault cases that do not contain explicit diagnostic fault codes, the diagnostic time range is determined through historical diagnostic fault codes, ensuring the fault diagnosis needs of cases without DTCs. The automated determination of the diagnostic time range reduces the need for manual intervention and significantly improves the efficiency of fault diagnosis.

[0013] Optionally, obtaining at least one historical diagnostic fault code associated with the fault case within the target historical time period, and determining the second diagnostic time range based on the at least one historical diagnostic fault code, includes: The target historical time period is determined based on the aforementioned fault cases; Obtain at least one historical diagnostic fault code triggered within the target historical time period; At least one historical diagnostic fault code and a description of the fault phenomenon of the fault case are used as prompt words and input into a preset large language model to obtain the valid diagnostic fault code of the fault case. The second diagnostic time range is determined based on the time window centered on the trigger timestamp of the valid diagnostic fault code.

[0014] Based on the above technical means, the diagnostic time range can be accurately locked based on the diagnostic fault codes, without the need for manual decision-making of a wide time window. This allows the matching fault diagnosis model to automatically perform diagnosis within the diagnostic time range, achieving accurate and comprehensive coverage of the abnormal problems reflected in the fault cases.

[0015] Optionally, determining the target historical time period based on the fault cases includes: If the fault case contains time information of first precision, then the target historical time period is determined to be the duration range centered on the time information of first precision. If the fault case contains time information with second precision, then the target historical time period is determined to be the date range containing time information with second precision; If the fault case does not contain time information, the target historical time period is determined to be the time period between the vehicle's previous fault diagnosis time and the time when the fault case was reported.

[0016] Based on the aforementioned technical means, the target historical time period can be flexibly determined according to the accuracy of the time information contained in the fault cases, and it can be ensured that fault cases without time information can meet the diagnostic requirements, thereby improving the accuracy of the diagnostic time range.

[0017] Optionally, the method further includes: The fault time, fault phenomenon, and fault conclusion from the fault diagnosis results are added as new fields and updated to the data record of the fault case. and / or The fault diagnosis results are used as training samples to update and train the target fault diagnosis model.

[0018] Based on the aforementioned technical means, by improving fault cases and updating and optimizing the fault diagnosis model, the richness of fault cases and the accuracy of the fault diagnosis model are enhanced to adapt to new fault cases, thus significantly improving the efficiency and accuracy of fault diagnosis.

[0019] Secondly, embodiments of this application provide a fault diagnosis device, the device comprising: The data acquisition module is used to acquire vehicle fault cases to be diagnosed and extract fault phenomenon descriptions from the fault cases; The model matching module is used to match the fault phenomenon description with a pre-created fault database to determine the fault phenomenon that matches the fault phenomenon description and the target fault diagnosis model corresponding to the fault phenomenon; wherein, the fault database includes the mapping relationship between fault phenomena and fault diagnosis models. The time determination module is used to determine the diagnostic time range of the fault case based on the diagnostic fault code if the fault case contains a diagnostic fault code; otherwise, it determines the diagnostic time range of the fault case based on the historical diagnostic fault codes associated with the fault case. The fault diagnosis module is used to acquire vehicle operation data within the diagnostic time range, and to perform fault diagnosis on the vehicle operation data using the target fault diagnosis model to obtain fault diagnosis results.

[0020] Thirdly, embodiments of this application provide an electronic device including a processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein the program or instructions, when executed by the processor, implement the steps of the fault diagnosis method as described in the first aspect.

[0021] Fourthly, embodiments of this application provide a readable storage medium storing a program or instructions that, when executed by a processor, implement the steps of the fault diagnosis method as described in the first aspect.

[0022] The fault diagnosis method provided in this application involves acquiring a fault case of a vehicle to be diagnosed, extracting a fault phenomenon description from the fault case, matching the fault phenomenon description with a pre-created fault database, determining the fault phenomenon that matches the fault phenomenon description, and the target fault diagnosis model corresponding to the fault phenomenon. The fault database includes a mapping relationship between natural language fault phenomenon descriptions and fault diagnosis models. If the fault case contains a diagnostic fault code, the diagnosis time range of the fault case is determined based on the diagnostic fault code; otherwise, the diagnosis time range of the fault case is determined based on historical diagnostic fault codes associated with the fault case. Vehicle operating data within the diagnosis time range is acquired, and the target fault diagnosis model is used to diagnose the fault in the vehicle operating data to obtain the fault diagnosis result. This application embodiment performs semantic analysis on fault cases, automatically selects fault phenomena and fault diagnosis models that match the fault cases, and accurately locks the diagnosis time range based on the diagnostic fault codes. It eliminates the need for manual decision-making on fault diagnosis models and broad time windows, reducing the cost of manual analysis and computing resources. By using the matched fault diagnosis model, it automatically performs diagnosis within the diagnosis time range, achieving accurate and comprehensive coverage of the abnormal problems reflected in the fault cases. This greatly improves the efficiency of fault diagnosis, enhances the accuracy of the fault diagnosis model, and adapts to new fault cases, significantly improving the efficiency and accuracy of fault diagnosis.

[0023] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, specific embodiments of this application are given below. Attached Figure Description

[0024] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of this application. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 This is a flowchart illustrating the steps of a fault diagnosis method provided in an embodiment of this application; Figure 2 This is a flowchart of another fault diagnosis method provided in an embodiment of this application; Figure 3 This is a schematic diagram of the mapping relationship in a fault diagnosis method provided in an embodiment of this application; Figure 4 This is a schematic flowchart of a fault diagnosis method provided in an embodiment of this application; Figure 5 This is a schematic diagram of the structure of a fault diagnosis device provided in an embodiment of this application; Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0025] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0026] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0027] The fault diagnosis method, apparatus, electronic device, and storage medium provided in this application will be described in detail below with reference to the accompanying drawings and through specific embodiments and application scenarios.

[0028] Example 1 Reference Figure 1The diagram illustrates a flowchart of a fault diagnosis method provided in an embodiment of this application. The method may include: Step 110: Obtain the vehicle's fault cases to be diagnosed, and extract the fault phenomenon description from the fault cases.

[0029] In this embodiment, to achieve vehicle fault diagnosis and analysis, fault cases reflecting market problems are automatically reviewed. The fault cases and the fault diagnosis model themselves are then corrected, supplemented, and optimized, thereby improving the richness of fault cases and the accuracy of the fault diagnosis model, ultimately achieving automatic and efficient fault diagnosis. It should be noted that the executing entity in this embodiment can be a platform, system, or cloud that receives fault cases of market problems, constructing a system for automated review of market fault cases and executing the fault diagnosis method. This embodiment does not specifically limit this aspect.

[0030] In this embodiment, the system first acquires vehicle fault cases to be diagnosed. These fault cases are records of faults and abnormalities that occur during vehicle operation. Fault cases may include descriptions of the vehicle's fault symptoms, possible Diagnostic Trouble Codes (DTCs), vehicle operating status, user operation records, or feedback. In this embodiment, the fault symptom description is extracted from the fault cases and used as input for subsequent fault diagnosis, enabling the automatic matching of the fault diagnosis model to the market problems reflected in the fault cases. It should be noted that the fault symptom description originates from the fault cases and is a textual description of the vehicle's fault symptoms. This description can even be extended to user questions. For example, a fault case may contain descriptions such as "the sunroof cannot be opened while driving" or "the vehicle suddenly brakes automatically while driving," and the extracted fault symptom description corresponds to vehicle sunroof malfunction, brake malfunction, etc.

[0031] Step 120: Match the fault phenomenon description with a pre-created fault database to determine the fault phenomenon that matches the fault phenomenon description and the target fault diagnosis model corresponding to the fault phenomenon; wherein, the fault database includes the mapping relationship between natural language fault phenomenon description and fault diagnosis model.

[0032] In this embodiment, fault phenomenon descriptions extracted from fault cases are matched with a pre-created fault database to determine the fault phenomena that match the fault phenomenon descriptions and the corresponding target fault diagnosis models. The fault database contains various mapping relationships between natural language fault phenomenon descriptions and fault diagnosis models. By matching fault phenomenon descriptions with the corresponding fault phenomena in the fault database, the fault phenomenon most closely matching the fault phenomenon description is identified, and the corresponding target fault diagnosis model is found. The target fault diagnosis model is the fault diagnosis model that matches the fault phenomenon description and is used for fault diagnosis analysis. It should be noted that the fault database is a pre-created database that stores the mapping relationships between natural language fault phenomenon descriptions and fault diagnosis models. The fault database includes natural language fault phenomenon descriptions (i.e., natural language descriptions of fault phenomena) and corresponding fault diagnosis models.

[0033] In this embodiment, based on the natural language fault phenomenon descriptions in a pre-created fault database, the fault phenomenon descriptions of fault cases are iterated and matched against the natural language fault phenomenon descriptions in the fault database to determine the fault phenomena that match the fault phenomenon descriptions. Then, according to the mapping relationships stored in the fault database, the target fault diagnosis model corresponding to the fault phenomenon matching the fault phenomenon description is determined. Specifically, the fault phenomenon description can be input into the fault database, and the fault phenomenon most closely matching the fault phenomenon description can be found through similarity calculation; details will not be elaborated further here.

[0034] Step 130: If the fault case contains diagnostic fault codes, determine the diagnostic time range of the fault case based on the diagnostic fault codes; otherwise, determine the diagnostic time range of the fault case based on the historical diagnostic fault codes associated with the fault case.

[0035] In this embodiment, to determine the time range for the fault diagnosis model, so that a broad time range can be selected at the input while shortening the actual analysis time of the model, the diagnostic time range is determined based on whether the fault case contains a Diagnostic Trouble Code (DTC). Specifically, if the fault case contains a clear DTC, the time range of the fault occurrence is directly determined based on the DTC, thus determining the diagnostic time range of the fault case. If the fault case does not contain a clear DTC, the DTC is filtered out from the historical DTC data associated with the fault case using a large language model to determine the time range of the fault occurrence, thus determining the diagnostic time range of the fault case. A DTC is a fault code generated by the vehicle fault diagnosis system to identify a specific fault phenomenon or cause; the diagnostic time range is the time interval used for fault diagnosis.

[0036] Step 140: Obtain vehicle operation data within the diagnostic time range, and use the target fault diagnosis model to perform fault diagnosis on the vehicle operation data to obtain fault diagnosis results.

[0037] In this embodiment, based on a determined diagnostic time range, vehicle operation data within the diagnostic time range is acquired, and a target fault diagnosis model is used to diagnose the vehicle operation data to obtain fault diagnosis results. The vehicle operation data consists of data generated by various actuators of the vehicle during operation. The fault diagnosis model analyzes the vehicle operation data and fault phenomenon descriptions to derive fault diagnosis results for the fault case. The fault diagnosis results include the fault time, fault phenomenon, and final fault conclusion. The fault time is the exact time when the fault occurred, the fault phenomenon is the phenomenon that caused the fault, and the fault conclusion is the conclusion drawn by the fault diagnosis model after locating the fault, such as "sunroof motor fault" or "brake pad fault". The specific process of fault diagnosis by the fault diagnosis model in this embodiment will not be described in detail.

[0038] This application embodiment performs semantic analysis on fault cases, automatically selects fault phenomena and fault diagnosis models that match the fault cases, and accurately locks the diagnosis time range based on the diagnostic fault codes. It eliminates the need for manual decision-making on fault diagnosis models and broad time windows, reducing the cost of manual analysis and computing resources. By using the matched fault diagnosis model, it automatically performs diagnosis within the diagnosis time range, achieving accurate and comprehensive coverage of the abnormal problems reflected in the fault cases. This greatly improves the efficiency of fault diagnosis, enhances the accuracy of the fault diagnosis model, and adapts to new fault cases, significantly improving the efficiency and accuracy of fault diagnosis.

[0039] Example 2 Reference Figure 2 The flowchart illustrates another fault diagnosis method provided in this application embodiment, wherein the fault database is created in the following manner: Step 210: Obtain the tree structure information of the fault diagnosis model used to perform fault diagnosis; wherein, the fault diagnosis model includes a root node and multiple leaf nodes under the root node.

[0040] In this embodiment, to efficiently diagnose newly added fault cases, a fault diagnosis model for analyzing the fault cases is automatically matched. Therefore, a fault database including natural language fault phenomenon descriptions and fault diagnosis models is pre-constructed to quickly and accurately match fault diagnosis models suitable for the fault cases. Specifically, the tree structure information of the fault diagnosis model used to perform fault diagnosis is obtained. The fault diagnosis model is a tree-structured model, including a root node and multiple leaf nodes under the root node. The tree structure information includes the hierarchical structure of the model, the connection relationship between nodes, and the specific content of each node. The root node represents the entry point of the model, and the leaf nodes represent various fault causes or fault phenomena. There may be multiple leaf nodes between the root node and the leaf nodes. These leaf nodes are used to represent information such as location, bus signal, and logs.

[0041] It should be noted that a fault diagnosis model is a logical model used to diagnose vehicle faults. It is usually represented by a tree structure, containing a root node and multiple leaf nodes, which are used to describe the path and cause of the fault. The root node is the top-level node of the tree structure, representing the entry point of the fault diagnosis model and reflecting the overall fault phenomenon or initial state of the vehicle. The leaf nodes are the bottom-level nodes of the tree structure, representing the specific fault causes or fault conclusions, and are the final output of the fault diagnosis model. The nodes between the root node and the leaf nodes can represent intermediate states or conditions in the fault diagnosis process.

[0042] Step 220: For each leaf node, input the diagnostic path information from each leaf node to the root node into the preset large language model to obtain one or more natural language fault phenomenon descriptions of the diagnostic path corresponding to the leaf node output by the large language model.

[0043] In this embodiment, for each leaf node, the diagnostic path information from each leaf node to the root node is input into a preset Large Language Model (LLM). The LLM outputs one or more natural language descriptions of fault phenomena corresponding to the leaf node's diagnostic path. The diagnostic path information from each leaf node to the root node includes end-to-end information such as fault location, fault code, bus signals, and log information, describing the fault's path and cause. The LLM generates one or more natural language descriptions of fault phenomena corresponding to each leaf node based on this diagnostic path information. By combining the tree structure information of the fault diagnosis model with the LLM, detailed natural language descriptions of fault phenomena are generated, supporting the diagnosis and analysis of complex faults. Automated generation of natural language descriptions of fault phenomena reduces manual costs in the fault diagnosis process.

[0044] It should be noted that the Large Language Model (LLM) used in this embodiment is a deep learning-based natural language processing model capable of generating natural language descriptions based on input text information. In this embodiment, it is used to generate natural language descriptions of fault phenomena based on the diagnostic path information of the fault diagnosis model. Natural language fault phenomenon descriptions are descriptions of fault phenomena expressed in natural language, such as "the sunroof cannot be opened." Specifically, each leaf node of the fault diagnosis model is traversed, and the diagnostic path information from the leaf node to the root node is taken as input and passed to the Large Language Model. The Large Language Model generates one or more natural language fault phenomenon descriptions corresponding to the leaf nodes. For example, for the diagnostic path information "Sunroof subsystem - Sunroof - Sunroof parameter error - Sunroof parameter not learned," the Large Language Model can generate natural language descriptions of fault phenomena such as "the sunroof cannot be opened" or "the sunroof operation is unresponsive."

[0045] Step 230: Convert the natural language description of the fault phenomenon into a first vector.

[0046] Step 240: Establish the mapping relationship between the first vector, the natural language fault phenomenon description and the corresponding fault diagnosis model, and store it to build a fault database.

[0047] In this embodiment, the natural language description of the fault phenomenon is converted into a first vector, and a mapping relationship is established between the first vector, the natural language description of the fault phenomenon, and the corresponding fault diagnosis model. This mapping relationship is then stored to construct a fault database. Through vectorization, the natural language description of the fault phenomenon is converted into a processable numerical form, facilitating automatic matching of fault diagnosis models for subsequent fault cases. Specifically, an embedding algorithm can be used to convert the natural language description of each fault phenomenon into a first vector. The first vector, the mapping relationship between the natural language description of the fault phenomenon and the corresponding fault diagnosis model are stored together in the fault database. The first vector is a numerical vector converted from the natural language description of the fault phenomenon using the embedding algorithm, representing the semantic features of the fault phenomenon. The embedding algorithm is an algorithm that converts natural language into numerical vectors. The fault database in this embodiment can be a database that stores vector data for efficient vector retrieval and matching.

[0048] In this embodiment, based on the one or more natural language fault phenomenon descriptions corresponding to each leaf node of the fault diagnosis model, a mapping relationship between the natural language fault phenomenon descriptions and the fault diagnosis model is generated. This mapping relationship associates the natural language description of each fault phenomenon with the corresponding fault diagnosis model, forming a correspondence between possible vehicle fault phenomena and fault diagnosis models, facilitating subsequent matching of the fault diagnosis model. The natural language fault phenomenon description corresponding to each leaf node is associated with the fault diagnosis model containing that leaf node to generate a mapping relationship. For example, a mapping relationship is formed between the fault phenomenon of a sunroof not opening and the fault diagnosis model for "Vehicle window system - sunroof - sunroof parameter error - sunroof parameter not learned".

[0049] Reference Figure 3 This diagram illustrates the mapping relationship in a fault diagnosis method provided in an embodiment of this application. The fault diagnosis model can diagnose various fault phenomena. The mapping relationship between fault phenomena and fault diagnosis models can be many-to-one or many-to-many. For example, fault phenomena 1, 2, and 3 all correspond to fault diagnosis model 1, and fault phenomenon n corresponds to fault diagnosis model k. Based on the mapping relationship between fault phenomena and fault diagnosis models, once the fault phenomenon matching the fault case is determined, the fault diagnosis model that diagnoses the fault phenomenon in the fault case can be matched according to the mapping relationship between the fault phenomenon and the fault diagnosis model.

[0050] This application's embodiments utilize a large language model to generate natural language fault phenomenon descriptions, improving the accuracy of fault phenomenon descriptions. The natural language fault phenomenon descriptions are associated with fault diagnosis models and stored in a vector database, enabling quick and automatic retrieval of the fault phenomenon most similar to a newly added fault case and the corresponding fault diagnosis model, thereby improving the efficiency of fault diagnosis and reducing the cost of manual intervention.

[0051] Example 3 In some embodiments of this application, step 120 matches the fault phenomenon description with a pre-created fault database to determine the fault phenomenon that matches the fault phenomenon description and the target fault diagnosis model corresponding to the fault phenomenon. Specifically, this may include the following steps: S11, Obtain the first vector of the natural language fault phenomenon description in the pre-created fault database; S12, convert the fault phenomenon description into a second vector, calculate the similarity between the first vector and the second vector, and determine the fault phenomenon that matches the fault phenomenon description; S13, Based on the mapping relationship stored in the fault database, determine the target fault diagnosis model corresponding to the fault phenomenon that matches the fault phenomenon description.

[0052] In this embodiment, the fault phenomenon description is matched with fault phenomena in a pre-created fault database to determine the fault phenomenon that matches the fault phenomenon description and the corresponding target fault diagnosis model. Specifically, the fault database is a database that stores the mapping relationship between natural language fault phenomenon descriptions and fault diagnosis models, including natural language fault phenomenon descriptions and corresponding fault diagnosis models. A first vector of the natural language fault phenomenon description is obtained from the pre-created fault database. The first vector is a numerical vector obtained by converting the natural language fault phenomenon description through an embedding algorithm, used to represent the semantic features of the fault phenomenon. The fault phenomenon description to be diagnosed is converted into a second vector, and the similarity between the second vector and the first vector in the fault database is calculated, thereby determining the fault phenomenon that matches the fault phenomenon description.

[0053] Specifically, the second vector can also be converted from the fault phenomenon description of the fault case to be diagnosed into a numerical vector using an embedding algorithm, representing the semantic features of the fault phenomenon in the fault case. Similarity calculation can be achieved by calculating the similarity between the two vectors, such as cosine similarity, to determine the semantic similarity between the fault phenomenon description and the natural language fault phenomenon descriptions in the fault database. The fault phenomenon corresponding to the natural language fault phenomenon description with the highest similarity is identified as the fault phenomenon matching the description. Therefore, based on the mapping relationship stored in the fault database, the target fault diagnosis model corresponding to the matched fault phenomenon is determined. The mapping relationship is the correspondence between the natural language fault phenomenon descriptions and fault diagnosis models stored in the fault database, used to associate fault phenomena with corresponding fault diagnosis models. The target fault diagnosis model is the fault diagnosis model corresponding to the matched fault phenomenon, used for diagnostic analysis of the fault case.

[0054] This application embodiment performs semantic analysis on fault cases, automatically selects fault phenomena and fault diagnosis models that match the fault cases, and quickly and automatically finds the fault phenomena and corresponding fault diagnosis models that are most similar to newly added fault cases, thereby improving the efficiency of fault diagnosis.

[0055] In some embodiments of this application, step 130 involves determining the diagnostic time range of the fault case based on the diagnostic fault code if the fault case contains a diagnostic fault code; otherwise, determining the diagnostic time range of the fault case based on the historical diagnostic fault codes associated with the fault case. Specifically, this may include the following steps: S21, If ​​the fault case contains a diagnostic fault code, then a first diagnostic time range is determined based on the time window centered on the trigger timestamp of the diagnostic fault code; S22, otherwise, obtain at least one historical diagnostic fault code associated with the fault case within the target historical time period, and determine the second diagnostic time range based on the at least one historical diagnostic fault code; S23, take the first diagnostic time range or the second diagnostic time range as the diagnostic time range of the fault case.

[0056] In this embodiment, to determine the time range for the fault diagnosis model, so that when executing the fault analysis model, a wide time range can be selected at the input end while shortening the actual analysis time range of the model, effectively improving fault diagnosis efficiency and reducing computational resources, the determination of the time range for the fault diagnosis model is divided into two cases based on whether the fault case contains diagnostic fault codes: if the fault case contains diagnostic fault codes, the diagnosis time range of the fault case is determined according to the diagnostic fault codes; otherwise, the diagnosis time range of the fault case is determined according to the historical diagnostic fault codes associated with the fault case.

[0057] In practice, the diagnostic time range for the fault diagnosis model to perform diagnosis on the fault case is determined based on whether the fault case contains a Diagnostic Trouble Code (DTC). A Diagnostic Trouble Code is a fault code generated by the vehicle fault diagnosis system to identify a specific fault phenomenon or fault cause. If the fault case contains a clear Diagnostic Trouble Code (DTC), the diagnostic time range for the fault case is determined directly based on the trigger timestamp of the DTC. The trigger timestamp of the DTC indicates the time point when the diagnostic fault code was triggered and recorded. If the fault case does not contain a clear Diagnostic Trouble Code (DTC), the diagnostic time range for the fault case is determined based on the historical diagnostic fault codes associated with the fault case.

[0058] In this embodiment, if the fault case contains a diagnostic fault code (DTC), a first diagnostic time range is determined based on a time window centered on the trigger timestamp of the DTC. The time window is a time interval shifted forward or backward by a certain amount of time, such as ±2 minutes, centered on the trigger timestamp of the DTC. The first diagnostic time range is the time range determined based on the trigger timestamp of the DTC and is used for fault diagnosis. If the fault case does not contain a specific diagnostic fault code, at least one historical diagnostic fault code associated with the fault case within the target historical time period is obtained, and a second diagnostic time range is determined based on the historical diagnostic fault code. The target historical time period is the time period related to the fault case, which can be determined based on the content of the fault case or the actual situation of the vehicle; details are omitted here. The historical diagnostic fault code is the diagnostic fault code triggered by the vehicle within the target historical time period. The second diagnostic time range is the time range determined based on the historical diagnostic fault code and is used for fault diagnosis. Depending on whether the fault case contains a specific diagnostic fault code (DTC), either the first or second diagnostic time range is selected as the final diagnostic time range for performing diagnosis on the fault case.

[0059] This application's embodiments determine the diagnostic time range based on diagnostic fault codes, accurately pinpointing the time of fault occurrence and avoiding issues of excessively large or small time ranges. For fault cases without explicit diagnostic fault codes, the diagnostic time range is determined through historical diagnostic fault codes, ensuring the diagnostic needs of faults without DTC cases are met. Automated determination of the diagnostic time range reduces the need for manual intervention and significantly improves the efficiency of fault diagnosis.

[0060] In some embodiments of this application, step S22, obtaining at least one historical diagnostic fault code associated with the fault case within the target historical time period, and determining a second diagnostic time range based on the at least one historical diagnostic fault code, may specifically include the following steps: Sub-step 01: Determine the target historical time period based on the aforementioned fault cases; Sub-step 02: Obtain at least one historical diagnostic fault code triggered within the target historical time period; Sub-step 03: Input at least one historical diagnostic fault code and the fault phenomenon description of the fault case as prompt words into a preset large language model to obtain the valid diagnostic fault code of the fault case. Sub-step 04: Determine the second diagnostic time range based on the time window centered on the trigger timestamp of the valid diagnostic fault code.

[0061] In this embodiment, a target historical time period is determined based on fault cases. This target historical time period is used to clearly define the time period for filtering historical diagnostic fault codes related to the fault cases. All diagnostic fault codes triggered by the vehicle are obtained from the target historical time period, and the diagnostic fault codes are deduplicated to obtain at least one historical diagnostic fault code. Diagnostic fault codes are fault codes generated by the vehicle fault diagnosis system and are used to identify specific fault phenomena or fault causes. By removing duplicate diagnostic fault codes, it is ensured that each diagnostic fault code appears only once.

[0062] In this embodiment, historical diagnostic fault codes and descriptions of fault phenomena in the current fault case to be diagnosed are used as prompt words, which are input into a preset Large Language Model (LLM). The LLM then filters out the most relevant valid diagnostic fault codes for the fault case. The prompt words are the text information input into the LLM, used to guide the model in generating or filtering outputs. The LLM is a deep learning-based natural language processing model used to generate or filter outputs based on the input text information. In this embodiment, the valid diagnostic fault codes are those most relevant to the fault case, used to determine the diagnostic time range. Specifically, a second diagnostic time range is determined based on a time window centered on the trigger timestamp of the valid diagnostic fault code. This time window shifts forward and backward by a certain amount of time, centered on the trigger timestamp of the valid diagnostic fault code.

[0063] It should be noted that if the diagnostic fault code of the fault case cannot be obtained or the historical diagnostic fault code related to the fault case cannot be obtained, the time range can be narrowed down by using the diagnostic fault code (DTC). In order to ensure that the fault model can perform fault diagnosis on the fault case, this embodiment can select the longest online time of the system before the fault case is reported as a fallback mechanism. For example, the maximum input time allowed by the cloud can be used as the upper limit to determine the diagnostic time range. No specific limitation is made here.

[0064] The embodiments of this application accurately pinpoint the diagnostic time range based on diagnostic fault codes, eliminating the need for manual decision-making of a broad time window. This allows for the automatic execution of diagnoses within the diagnostic time range using a matching fault diagnosis model, achieving accurate and comprehensive coverage of the abnormal issues reflected in fault cases.

[0065] As a specific implementation of an embodiment of this application, sub-step 01, determining the target historical time period based on the fault case, may specifically include: If the fault case contains time information of first precision, then the target historical time period is determined to be the duration range centered on the time information of first precision. If the fault case contains time information with second precision, then the target historical time period is determined to be the date range containing time information with second precision; If the fault case does not contain time information, the target historical time period is determined to be the time period between the vehicle's previous fault diagnosis time and the time when the fault case was reported.

[0066] In this embodiment, the target historical time period is used to clearly define the time period for filtering historical diagnostic fault codes related to the fault case. Therefore, the target historical time period directly affects the accuracy of obtaining historical diagnostic fault codes. In this embodiment, the target historical time period is determined based on the precision of the time information contained in the fault case. The specific implementation methods include the following three cases: Scenario 1: If the fault case contains time information of first precision, then the target historical time period is determined as the duration range centered on the time information of first precision. Here, the time information of first precision refers to the precise time information contained in the fault case, accurate to the minute or second level. The duration range is the time interval formed by shifting a certain amount of time before and after the time information of first precision. For example, if the fault case contains time information of first precision such as 2025-10-01 14:35:42, then shifting a certain amount of time before and after this time information, such as ±2 minutes, the target historical time period is determined to be from 2025-10-01 14:33:42 to 2025-10-01 14:37:42. This time period is used to obtain historical diagnostic fault codes related to the fault case.

[0067] Scenario 2: If the fault case contains second-precision time information, the target historical time period is determined as the date range containing this second-precision time information. Second-precision time information refers to coarser time information contained in the fault case, accurate to the hour, half-day, or date level. The date range is the time interval formed by the date containing the second-precision time information and its adjacent dates, such as ±1 day. For example, if the fault case contains second-precision time information such as 2025-10-01 14:00 or 2025-10-01, the target historical time period is determined as the date range containing this time information. For instance, if the time information is 2025-10-01, then 2025-09-30 to 2025-10-02 is selected as the target historical time period to obtain all diagnostic fault codes triggered by the vehicle within this time period.

[0068] Scenario 3: If the fault case does not contain time information, the target historical time period is determined to be the period between the vehicle's last fault diagnosis time and the fault case reporting time. The last fault diagnosis time is the time when the vehicle was last diagnosed, usually recorded in the vehicle's maintenance records or historical data. The fault case reporting time is the time when the fault case was reported or recorded. If the fault case does not contain any time information, the target historical time period is determined to be the period between the vehicle's last fault diagnosis time and the fault case reporting time. For example, if the last fault diagnosis time is 2025-09-28 10:00:00 and the fault case reporting time is 2025-10-01 15:00:00, the target historical time period is determined to be from 2025-09-28 10:00:00 to 2025-10-01 15:00:00.

[0069] The embodiments of this application flexibly determine the target historical time period based on the accuracy of the time information contained in the fault cases, and ensure that fault cases without time information can meet the diagnostic requirements, thereby improving the accuracy of the diagnostic time range.

[0070] In some embodiments of this application, the method further includes: S31, update the fault time, fault phenomenon and fault conclusion in the fault diagnosis result as new fields to the data record of the fault case; and / or S32, the fault diagnosis results are used as training samples to update and train the target fault diagnosis model.

[0071] In this embodiment, based on the fault diagnosis results, the data records of fault cases are updated, and / or the target fault diagnosis model is updated to improve the richness of fault cases and the accuracy of the fault diagnosis model. Specifically, the fault time, fault phenomenon, and fault conclusion in the fault diagnosis results are added as new fields and updated to the data records of the fault cases, and / or the fault diagnosis results are used as training samples to update and train the target fault diagnosis model to improve the accuracy and applicability of the model. The specific execution of updating and training the target fault diagnosis model is not limited in this embodiment.

[0072] In practice, after the fault diagnosis model performs the diagnosis, two scenarios can occur: successful fault location or failure. For successfully located fault cases, firstly, the precise time of the final fault location is added to and updated in the fault case. The fault time, fault symptoms, and fault conclusions from the fault diagnosis results are stored in the fault case's fast retrieval database, and the review process for this fault case is closed. For fault cases that fail to locate the fault, they can enter the expert manual analysis stage. The possibilities for failure are as follows: incorrect diagnosis time range, incorrect matching of the fault diagnosis model, incomplete fault diagnosis model, or reliance on analysis logic. In the case of an incorrect diagnosis time range, after expert analysis and location, the correct diagnosis time range is added to and updated in the fault case. In the case of an incorrectly matched fault diagnosis model, after expert analysis and matching to the correct fault diagnosis model, the mapping relationship between the fault symptoms and the fault diagnosis model is added to the fault database in the form of QA pairs. In the case of an incomplete fault diagnosis model or reliance on analysis logic, after expert analysis, the fault diagnosis model is expanded or modified according to the analysis logic. Then, the corrected fault time, fault symptoms, and fault conclusions are stored in the fault case's fast retrieval database, and the review process is closed.

[0073] To facilitate understanding of the fault diagnosis method described in the above embodiments by those skilled in the art, please refer to... Figure 4 This document illustrates a flowchart of a fault diagnosis method provided in an embodiment of this application. Based on market issues related to vehicles, a vehicle fault case to be diagnosed is received, and similar fault cases are quickly queried. It should be noted that the fault phenomenon corresponding to the fault case and the fault diagnosis model corresponding to the fault phenomenon can be matched from the fault database. Fault cases reflecting market issues are generated using the fault diagnosis model or expert analysis. Then, the fault case review analysis is performed using the fault diagnosis model corresponding to the fault case. The fault time, fault phenomenon, and fault conclusion in the fault diagnosis results are stored in the fault case quick retrieval database, and the fault diagnosis model is expanded or modified by analysis logic.

[0074] The embodiments of this application improve the richness of fault cases and the accuracy of fault diagnosis models by perfecting fault cases and updating and optimizing fault diagnosis models, so as to adapt to new fault cases and significantly improve the efficiency and accuracy of fault diagnosis.

[0075] Reference Figure 5 The diagram shows a structural schematic of a fault diagnosis device 30 provided in an embodiment of this application. The device includes: The data acquisition module 310 is used to acquire vehicle fault cases to be diagnosed and extract fault phenomenon descriptions from the fault cases. The model matching module 320 is used to match the fault phenomenon description with a pre-created fault database to determine the fault phenomenon that matches the fault phenomenon description and the target fault diagnosis model corresponding to the fault phenomenon; wherein, the fault database includes the mapping relationship between natural language fault phenomenon descriptions and fault diagnosis models. The time determination module 330 is used to determine the diagnostic time range of the fault case based on the diagnostic fault code if the fault case contains a diagnostic fault code; otherwise, it determines the diagnostic time range of the fault case based on the historical diagnostic fault codes associated with the fault case. The fault diagnosis module 340 is used to acquire vehicle operation data within the diagnostic time range, and to perform fault diagnosis on the vehicle operation data using the target fault diagnosis model to obtain fault diagnosis results.

[0076] Optionally, the device further includes: The model acquisition module is used to acquire the tree structure information of the fault diagnosis model used to perform fault diagnosis; wherein, the fault diagnosis model includes a root node and multiple leaf nodes under the root node; The processing module is used to input the diagnostic path information from each leaf node to the root node into a preset large language model for each leaf node, and obtain one or more natural language fault phenomenon descriptions of the leaf node corresponding to the diagnostic path output by the large language model. A conversion module is used to convert the natural language description of the fault phenomenon into a first vector; The storage module is used to establish the mapping relationship between the first vector, the natural language fault phenomenon description and the corresponding fault diagnosis model, and store them to build a fault database.

[0077] Optionally, the model matching module 320 includes: The first acquisition submodule is used to acquire the first vector of the natural language fault phenomenon description in the pre-created fault database; The first matching submodule is used to convert the fault phenomenon description into a second vector, calculate the similarity between the first vector and the second vector, and determine the fault phenomenon that matches the fault phenomenon description. The second matching submodule is used to determine the target fault diagnosis model corresponding to the fault phenomenon that matches the fault phenomenon description based on the mapping relationship stored in the fault database.

[0078] Optionally, the time determination module 330 includes: The first determining submodule is used to determine a first diagnostic time range based on a time window centered on the trigger timestamp of the diagnostic fault code if the fault case contains a diagnostic fault code. The second determining submodule is used to otherwise obtain at least one historical diagnostic fault code associated with the fault case within the target historical time period, and determine the second diagnostic time range based on the at least one historical diagnostic fault code. The third determining submodule is used to take the first diagnostic time range or the second diagnostic time range as the diagnostic time range of the fault case.

[0079] Optionally, the second determining submodule includes: The first determining unit is used to determine the target historical time period based on the fault cases; The acquisition unit is used to acquire at least one historical diagnostic fault code triggered within the target historical time period; The processing unit is used to input at least one historical diagnostic fault code and the fault phenomenon description of the fault case as prompt words into a preset large language model to obtain the valid diagnostic fault code of the fault case. The second determining unit is used to determine the second diagnostic time range based on the time window centered on the trigger timestamp of the valid diagnostic fault code.

[0080] Optionally, the first determining unit includes: The first determining subunit is used to determine the target historical time period as a duration range centered on the time information of the first precision if the fault case contains time information of the first precision. The second determining subunit is used to determine the target historical time period as the date range containing the second precision time information if the fault case contains time information of the second precision. The third determining subunit is used to determine the target historical time period as the time period between the vehicle's previous fault diagnosis time and the time when the fault case was reported if the fault case does not contain time information.

[0081] Optionally, the device further includes: The first update module is used to update the fault time, fault phenomenon and fault conclusion in the fault diagnosis result as new fields to the data record of the fault case; and / or The second update module is used to update and train the target fault diagnosis model by using the fault diagnosis results as training samples.

[0082] The fault diagnosis device provided in this application obtains a fault case of a vehicle to be diagnosed, extracts a fault phenomenon description from the fault case, matches the fault phenomenon description with a pre-created fault database, determines the fault phenomenon that matches the fault phenomenon description and the target fault diagnosis model corresponding to the fault phenomenon; wherein, the fault database includes a mapping relationship between natural language fault phenomenon descriptions and fault diagnosis models; if the fault case contains a diagnostic fault code, the diagnosis time range of the fault case is determined according to the diagnostic fault code; otherwise, the diagnosis time range of the fault case is determined according to the historical diagnostic fault codes associated with the fault case; the vehicle operation data within the diagnosis time range is obtained, and the target fault diagnosis model is used to diagnose the fault in the vehicle operation data to obtain the fault diagnosis result. This application embodiment performs semantic analysis on fault cases, automatically selects fault phenomena and fault diagnosis models that match the fault cases, and accurately locks the diagnosis time range based on the diagnostic fault codes. It eliminates the need for manual decision-making on fault diagnosis models and broad time windows, reducing the cost of manual analysis and computing resources. By using the matched fault diagnosis model, it automatically performs diagnosis within the diagnosis time range, achieving accurate and comprehensive coverage of the abnormal problems reflected in the fault cases. This greatly improves the efficiency of fault diagnosis, enhances the accuracy of the fault diagnosis model, and adapts to new fault cases, significantly improving the efficiency and accuracy of fault diagnosis.

[0083] This application also provides an electronic device 40, please refer to... Figure 6 It includes a processor 410 and a memory 420, wherein the memory 410 is used to store computer programs; the processor 420 is used to execute the programs stored in the memory 410 to implement the fault diagnosis method described in any embodiment of this application.

[0084] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the fault diagnosis method described in any embodiment of this application.

[0085] In this application, "multiple" refers to two or more.

[0086] In this application, unless otherwise expressly defined, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0087] The terms “first,” “second,” “third,” “fourth,” etc., in this application (if present) are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0088] In this application, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, in this application, the character "" generally indicates that the preceding and following related objects have an "or" relationship.

[0089] Unless otherwise specified, all steps in this application may be performed sequentially or randomly. For example, if the method includes steps A and B, it means that the method may include steps A and B performed sequentially, or it may include steps B and A performed sequentially. For example, if the method may also include step C, it means that step C may be added to the method in any order. For example, the method may include steps A, B, and C, or it may include steps A, C, and B, or it may include steps C, A, and B, etc.

[0090] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A fault diagnosis method, characterized in that, The method includes: Obtain vehicle fault cases to be diagnosed, and extract fault phenomenon descriptions from the fault cases; The fault phenomenon description is matched with a pre-created fault database to determine the fault phenomenon that matches the fault phenomenon description and the target fault diagnosis model corresponding to the fault phenomenon; wherein, the fault database includes the mapping relationship between natural language fault phenomenon descriptions and fault diagnosis models. If the fault case contains a diagnostic fault code, the diagnostic time range of the fault case is determined based on the diagnostic fault code; otherwise, the diagnostic time range of the fault case is determined based on the historical diagnostic fault codes associated with the fault case. The vehicle operation data within the diagnostic time range is acquired, and the target fault diagnosis model is used to diagnose the fault in the vehicle operation data to obtain the fault diagnosis result.

2. The method according to claim 1, characterized in that, The fault database is created in the following manner: Obtain the tree structure information of the fault diagnosis model used to perform fault diagnosis; wherein, the fault diagnosis model includes a root node and multiple leaf nodes under the root node; For each leaf node, the diagnostic path information from each leaf node to the root node is input into a preset large language model to obtain one or more natural language fault phenomenon descriptions of the leaf node corresponding to the diagnostic path output by the large language model. The natural language description of the fault phenomenon is converted into a first vector; Establish a mapping relationship between the first vector, the natural language description of the fault phenomenon, and the corresponding fault diagnosis model, and store it to construct a fault database.

3. The method according to claim 2, characterized in that, The step of matching the fault phenomenon description with a pre-created fault database to determine the fault phenomenon that matches the fault phenomenon description and the target fault diagnosis model corresponding to the fault phenomenon includes: Obtain the first vector of the natural language description of the fault phenomena in the pre-created fault database; The fault phenomenon description is converted into a second vector, the similarity between the first vector and the second vector is calculated, and the fault phenomenon that matches the fault phenomenon description is determined. Based on the mapping relationships stored in the fault database, a target fault diagnosis model corresponding to the fault phenomenon that matches the fault phenomenon description is determined.

4. The method according to claim 1, characterized in that, If the fault case contains a diagnostic fault code, the diagnostic time range of the fault case is determined based on the diagnostic fault code; otherwise, the diagnostic time range of the fault case is determined based on historical diagnostic fault codes associated with the fault case, including: If the fault case contains a diagnostic fault code, then the first diagnostic time range is determined based on the time window centered on the trigger timestamp of the diagnostic fault code; Otherwise, obtain at least one historical diagnostic fault code associated with the fault case within the target historical time period, and determine the second diagnostic time range based on the at least one historical diagnostic fault code; The first diagnostic time range or the second diagnostic time range shall be used as the diagnostic time range for the fault case.

5. The method according to claim 4, characterized in that, The step of obtaining at least one historical diagnostic fault code associated with the fault case within the target historical time period, and determining the second diagnostic time range based on the at least one historical diagnostic fault code, includes: The target historical time period is determined based on the aforementioned fault cases; Obtain at least one historical diagnostic fault code triggered within the target historical time period; At least one historical diagnostic fault code and a description of the fault phenomenon of the fault case are used as prompt words and input into a preset large language model to obtain the valid diagnostic fault code of the fault case. The second diagnostic time range is determined based on the time window centered on the trigger timestamp of the valid diagnostic fault code.

6. The method according to claim 5, characterized in that, The determination of the target historical time period based on the fault cases includes: If the fault case contains time information of first precision, then the target historical time period is determined to be the duration range centered on the time information of first precision. If the fault case contains time information with second precision, then the target historical time period is determined to be the date range containing time information with second precision; If the fault case does not contain time information, the target historical time period is determined to be the time period between the vehicle's previous fault diagnosis time and the time when the fault case was reported.

7. The method according to claim 1, characterized in that, The method further includes: The fault time, fault phenomenon, and fault conclusion from the fault diagnosis results are added as new fields and updated to the data record of the fault case. and / or The fault diagnosis results are used as training samples to update and train the target fault diagnosis model.

8. A fault diagnosis device, characterized in that, The device includes: The data acquisition module is used to acquire vehicle fault cases to be diagnosed and extract fault phenomenon descriptions from the fault cases; The model matching module is used to match the fault phenomenon description with a pre-created fault database to determine the fault phenomenon that matches the fault phenomenon description and the target fault diagnosis model corresponding to the fault phenomenon; wherein, the fault database includes the mapping relationship between natural language fault phenomenon descriptions and fault diagnosis models. The time determination module is used to determine the diagnostic time range of the fault case based on the diagnostic fault code if the fault case contains a diagnostic fault code; otherwise, it determines the diagnostic time range of the fault case based on the historical diagnostic fault codes associated with the fault case. The fault diagnosis module is used to acquire vehicle operation data within the diagnostic time range, and to perform fault diagnosis on the vehicle operation data using the target fault diagnosis model to obtain fault diagnosis results.

9. An electronic device, characterized in that, It includes a processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein the program or instructions, when executed by the processor, implement the steps of the fault diagnosis method as described in any one of claims 1-7.

10. A readable storage medium, characterized in that, The readable storage medium stores a program or instructions that, when executed by a processor, implement the steps of the fault diagnosis method as described in any one of claims 1-7.

Images