Vehicle fault diagnosis method and device and storage medium

Abstract

The invention provides a vehicle fault diagnosis method and apparatus, and a storage medium. The method comprises the steps of obtaining abnormal fault data and a preset knowledge graph of a vehicle; determining a fault reason similarity according to the abnormal fault data and a preset knowledge graph; according to the method, the limitation of a single fault point is broken through, the fault reason similarity is determined according to the multi-fault coupling effect so as to comprehensively cover fault reasons of different probabilities related to abnormal fault data, and a plurality of guiding troubleshooting schemes are sequentially executed according to the fault reason similarity so as to diagnose the fault of the vehicle. And the pertinence and efficiency of vehicle fault diagnosis are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicles, and in particular, to a fault diagnosis method, device, and storage medium for a vehicle. Background Art

[0002] Traditional maintenance and troubleshooting methods rely on fault maintenance manuals and the long-term accumulated experience and intuition of technicians, and conduct troubleshooting according to the fixed guiding troubleshooting steps in the maintenance manuals. In the face of complex and changeable fault situations, this method is prone to misdiagnosis, missed diagnosis, or low troubleshooting efficiency, and lacks expert-style fault guidance.

[0003] In the related art, a fault maintenance knowledge graph and an AI (Artificial Intelligence) diagnosis model are used to analyze fault information to generate fault causes, and according to the fault causes, a solution associated with the obtained fault information and fault causes is determined from the fault maintenance knowledge graph.

[0004] However, in the troubleshooting process, there is a lack of correlation analysis and intelligent convergence and focusing. The current vehicle intelligent diagnosis system is limited to the isolated analysis of a single fault point and lacks the ability to comprehensively examine the correlations between systems, components, and even data. Summary of the Invention

[0005] The present invention aims to at least solve one of the technical problems existing in the prior art.

[0006] To this end, an object of the present invention is to propose a fault diagnosis method for a vehicle. This method breaks through the limitation of a single fault point, determines the similarity of fault causes according to the multi-fault coupling effect, so as to comprehensively cover different probability fault causes related to abnormal fault data, and sequentially execute multiple guiding troubleshooting schemes according to the similarity of fault causes to diagnose the faults of the vehicle, improving the pertinence and efficiency of vehicle fault diagnosis.

[0007] To this end, a second object of the present invention is to propose a fault diagnosis device for a vehicle.

[0008] To this end, a third object of the present invention is to propose a computer-readable storage medium.

[0009] To achieve the above object, an embodiment of the first aspect of the present invention proposes a fault diagnosis method for a vehicle, the method comprising: obtaining abnormal fault data of the vehicle and a preset knowledge graph; determining the similarity of fault causes according to the abnormal fault data and the preset knowledge graph; diagnosing the faults of the vehicle according to the similarity of fault causes.

[0010] The fault diagnosis method of a vehicle according to an embodiment of the present invention captures abnormal fault data in the vehicle running state in real time by obtaining a preset knowledge graph, reads the fault features corresponding to the abnormal fault data in the preset knowledge graph, performs intelligent convergence and focusing according to the interaction between the fault features, deeply mines the correlation between the fault phenomenon and the fault cause, breaks through the limitation of a single fault point, determines the similarity of fault causes according to the multi-fault coupling effect, so as to comprehensively cover different probability fault causes related to the abnormal fault data, and sequentially executes a plurality of guiding troubleshooting schemes according to the similarity of fault causes to diagnose the faults of the vehicle, thereby improving the pertinence and efficiency of vehicle fault diagnosis.

[0011] In some embodiments, determining the similarity of fault causes according to the abnormal fault data and the preset knowledge graph includes: obtaining the fault cause number in the abnormal state data and the common fault phenomena in the preset knowledge graph; determining the similarity of the fault causes according to the fault cause number and the common fault phenomena.

[0012] In some embodiments, when determining the similarity of the fault causes according to the fault cause number and the common fault phenomena, substituting the fault cause number and the common fault phenomena into the following formula: , where, is the similarity of the fault causes, v is the common fault phenomenon, is the number of fault phenomena caused by the fault cause m, is the number of fault phenomena caused by the fault cause n, is the number of fault phenomena of the common fault phenomenon.

[0013] In some embodiments, diagnosing the faults of the vehicle according to the similarity of the fault causes includes: determining the interest value of the fault phenomenon in the fault cause according to the similarity of the fault causes; diagnosing the faults of the vehicle according to the interest value of the fault phenomenon in the fault cause.

[0014] In some embodiments, when determining the interest value of the fault phenomenon in the fault cause according to the similarity of the fault causes, substituting the similarity of the fault causes into the following formula: , where, is the interest value of the common fault phenomenon in the fault cause n, is the set of the k fault causes most similar to the fault cause n, is the similarity of the fault causes, is the interest value of the common fault phenomenon in the fault cause m.

[0015] In some embodiments, diagnosing a fault of the vehicle based on the interest value of the fault cause according to the fault phenomenon includes: determining the priority order of a preset guiding troubleshooting plan according to the interest value; and sequentially executing the preset guiding troubleshooting plan on the vehicle according to the priority order to diagnose the fault of the vehicle.

[0016] In some embodiments, sequentially executing the guiding troubleshooting plan on the vehicle according to the priority order includes: if the guiding troubleshooting plan diagnoses the fault of the vehicle, determining a repair measure corresponding to the guiding troubleshooting plan; if the guiding troubleshooting plan does not diagnose the fault of the vehicle, determining an actual repair measure according to user fault information.

[0017] In some embodiments, obtaining the preset knowledge graph includes: determining a fault code of the vehicle according to an identification code of the vehicle; determining preset fault data in a preset fault knowledge base according to the fault code; extracting semantic parameters of the preset fault data, where the semantic parameters include entity parameters, relationship parameters, and attribute parameters; determining a knowledge triple according to the entity parameters, the relationship parameters, and the attribute parameters; and determining the preset knowledge graph according to the knowledge triple.

[0018] To achieve the above object, an embodiment of the second aspect of the present invention provides a vehicle fault diagnosis device, where the device includes: an acquisition module, configured to acquire abnormal fault data of the vehicle and a preset knowledge graph; a determination module, configured to determine a fault cause similarity according to the abnormal fault data and the preset knowledge graph; and a diagnosis module, configured to diagnose the fault of the vehicle according to the fault cause similarity.

[0019] According to the vehicle fault diagnosis device of the embodiments of the present invention, by acquiring a preset knowledge graph and capturing abnormal fault data in the real-time operation state of the vehicle, reading fault features corresponding to the abnormal fault data in the preset knowledge graph, performing intelligent convergence and focusing according to the interaction between the fault features, deeply mining the correlation between the fault phenomenon and the fault cause, breaking through the limitation of a single fault point, determining the fault cause similarity according to the multi-fault coupling effect, so as to comprehensively cover different probability fault causes related to the abnormal fault data, and sequentially executing a plurality of guiding troubleshooting plans according to the fault cause similarity to diagnose the fault of the vehicle, thereby improving the pertinence and efficiency of vehicle fault diagnosis.

[0020] To achieve the above object, an embodiment of the third aspect of the present invention provides a computer-readable storage medium, on which a vehicle fault diagnosis program is stored, and when the vehicle fault diagnosis program is executed by a processor, the vehicle fault diagnosis method as described in the above embodiments is implemented.

[0021] Additional aspects and advantages of the present invention will be given in part in the following description, become apparent in part from the following description, or be learned by practice of the present invention. Brief Description of the Drawings

[0022] The above and / or additional aspects and advantages of the present invention will become apparent and be readily understood from the following description of embodiments in conjunction with the accompanying drawings, in which: Figure 1 is a flowchart of a vehicle fault diagnosis method according to an embodiment of the present invention; Figure 2 is a framework diagram of a vehicle fault diagnosis method according to an embodiment of the present invention; Figure 3 is a framework diagram of a vehicle fault diagnosis method according to another embodiment of the present invention; Figure 4 is a flowchart of a vehicle fault diagnosis method according to a specific embodiment of the present invention; Figure 5 is a structural block diagram of a vehicle fault diagnosis device according to a specific embodiment of the present invention.

[0023] Reference Signs: Vehicle fault diagnosis device 2; Acquisition module 21; Determination module 22; Diagnosis module 23. Detailed Description of the Embodiments

[0024] Embodiments of the present invention will be described in detail below. The embodiments described with reference to the accompanying drawings are exemplary. Embodiments of the present invention will be described in detail below.

[0025] In the related art, a fault diagnosis method, system, device, and storage medium analyze fault information using a fault repair knowledge graph and an AI diagnosis model to generate fault causes. According to the fault causes, a solution associated with the acquired fault information and fault causes is determined and obtained from the fault repair knowledge graph. However, this solution cannot update the recommended repair solution based on user feedback to provide the optimal solution for the user.

[0026] Alternatively, an artificial intelligence vehicle automatic question-answering method and system map the generated entities and the relationships between entities to a knowledge graph vehicle question-answering knowledge base, and generate results through the knowledge graph vehicle question-answering knowledge base and feedback them to the vehicle owner. This solution can achieve interactive question-answering with users but lacks intelligent guidance for users to troubleshoot faults.

[0027] However, the above diagnostic steps are rigid and do not make full use of the latest vehicle fault data. Existing vehicle intelligent diagnostic systems adopt a preset diagnostic process. Although this process is constructed based on widely collected fault cases and expert experience, it cannot change the recommended repair plan in a timely manner based on user feedback and then give the optimal troubleshooting order. Moreover, in the face of rapidly evolving vehicle technologies and emerging new fault modes, the system fails to integrate and apply the latest vehicle fault data in real time, resulting in a lag in diagnostic strategies and an inability to accurately respond to new situations, thus limiting the further improvement of diagnostic efficiency and accuracy.

[0028] The degree of automation in organizing maintenance cases is low and the labor cost is high. After each maintenance and troubleshooting is completed, the existing system mostly relies on manual writing of maintenance cases. Human factors lead to missing or inaccurate information. Manually organized cases are difficult to ensure uniform formats and data specifications, which is not conducive to subsequent data mining and analysis.

[0029] The following combines Figures 1 - 4 to illustrate by example the vehicle fault diagnosis method of the embodiments of the present invention. The vehicle fault diagnosis method is used for vehicles.

[0030] As Figure 1 shown, the vehicle fault diagnosis method of the embodiments of the present invention at least includes step S1 - step S3.

[0031] Step S1, obtain the abnormal fault data and the preset knowledge graph of the vehicle.

[0032] In the embodiment, as Figure 2 shown, monitor the operating state of the vehicle. When the vehicle is in a normal driving or using state, once the vehicle encounters problems during operation, the sensors installed on the vehicle will monitor the state of the vehicle in real time to obtain the abnormal fault data of the vehicle, such as instrument fault light monitoring, DTC (Diagnostic Trouble Code) fault code monitoring, etc., and transmit abnormal fault data such as CAN (Controller Area Network) signals and fault codes to the built-in fault model system to be able to identify potential fault modes; once the fault model system detects abnormal fault data, it starts the fault prediction function and records these prediction information to provide basic data for subsequent fault analysis and achieve the active discovery of faults. At the same time, obtain the preset knowledge graph corresponding to the vehicle model, which includes the latest maintenance cases, expert experience and other latest data.

[0033] Step S2, determine the similarity of fault causes according to the abnormal fault data and the preset knowledge graph.

[0034] In an embodiment, by using the read abnormal fault data, the fault features corresponding to the abnormal fault data in a preset knowledge graph are read. By introducing a collaborative filtering algorithm such as the ItemCF-IUF algorithm, convergence and focusing are performed according to the interaction between the fault features, and the correlation between the fault phenomenon and the fault cause is deeply mined to determine the similarity of the fault cause, thereby solving the problem of lack of correlation analysis and intelligent convergence and focusing in the current intelligent diagnosis technology of existing vehicles.

[0035] Step S3: Diagnose the vehicle fault according to the similarity of the fault cause.

[0036] In an embodiment, after determining the similarity of the fault cause, multiple guiding troubleshooting schemes are sorted according to the numerical value of the similarity of the fault cause. The guiding troubleshooting schemes include possible fault causes and maintenance suggestions. When the user performs an interactive operation and obtains the fault feedback result of the undiagnosed vehicle, the guiding troubleshooting schemes are re-recommended, that is, the fault causes and maintenance steps are re-sorted until the problem is solved for the user, improving the pertinence and efficiency of vehicle fault diagnosis.

[0037] For example, the remote guiding troubleshooting module uses an interactive guiding mode, including automatically generating multiple guiding troubleshooting schemes according to the similarity of the fault cause. The user performs step-by-step operations according to the guiding troubleshooting schemes, that is, the preliminary guiding troubleshooting scheme determined according to the largest similarity of the fault cause, guiding the technician to perform the next operation according to the preliminary guiding troubleshooting scheme, and gradually feeding back the troubleshooting results so that the system can timely understand the progress of the fault inspection. If the vehicle fault is diagnosed according to the preliminary guiding troubleshooting scheme, it is evaluated that the current scheme is effective; if the vehicle fault is not diagnosed according to the preliminary guiding troubleshooting scheme, the recommendation engine will automatically adjust the sorting of the candidate guiding troubleshooting schemes and push them to the customer, and then find the optimal troubleshooting scheme to achieve rapid fault resolution.

[0038] According to the vehicle fault diagnosis method of the embodiment of the present invention, by obtaining a preset knowledge graph and real-time capturing abnormal fault data in the vehicle running state, reading the fault features corresponding to the abnormal fault data in the preset knowledge graph, performing intelligent convergence and focusing according to the interaction between the fault features, deeply mining the correlation between the fault phenomenon and the fault cause, breaking through the limitation of a single fault point, determining the similarity of the fault cause according to the multi-fault coupling effect to comprehensively cover different probability fault causes related to the abnormal fault data, and sequentially executing multiple guiding troubleshooting schemes according to the similarity of the fault cause to diagnose the vehicle fault, improving the pertinence and efficiency of vehicle fault diagnosis.

[0039] In some embodiments, determining the similarity of fault causes based on abnormal fault data and a preset knowledge graph includes: obtaining the fault cause numbers in the abnormal state data and the common fault phenomena in the preset knowledge graph; and determining the similarity of fault causes based on the fault cause numbers and the common fault phenomena.

[0040] In an embodiment, after obtaining the abnormal fault data and the preset knowledge graph of the vehicle, the fault cause numbers in the abnormal state data are extracted, such as fault cause m, fault cause n, etc., and the common fault phenomena in the preset knowledge graph ; The above parameters are input into the intelligent recommendation analysis model for fault causes based on the ItemCF-IUF algorithm, and then the similarity of fault causes is calculated to generate a similarity matrix. The intelligent recommendation analysis model for fault causes uses intelligent association analysis technology to deeply mine and cross-compare fault information, accurately focus on the root cause of the fault, reduce misjudgment and missed judgment. By constructing a complex fault model and a preset knowledge graph, intelligent association analysis can quickly identify the potential connections between faults, provide a clear fault path for maintenance personnel, and accelerate the fault location and repair process.

[0041] In some embodiments, when determining the similarity of fault causes based on the fault cause numbers and the common fault phenomena, substitute the fault cause numbers and the common fault phenomena into the following formula: , where, is the similarity of fault causes, v is the common fault phenomenon, is the number of fault phenomena caused by fault cause m, is the number of fault phenomena caused by fault cause n, is the number of fault phenomena of the common fault phenomenon.

[0042] In an embodiment, Jaccard similarity or cosine similarity can be selected to calculate the similarity of fault causes. The Jaccard similarity calculation formula is as follows: , In the formula, is the similarity of the fault causes, is the number of fault phenomena caused by fault cause m, is the number of fault phenomena caused by fault cause n, is the number of fault phenomena jointly caused by fault cause m and fault cause n. There is a problem in calculating the similarity of fault causes through the Jaccard similarity calculation formula, that is, if a certain fault cause is relatively common, it is easy to cause the data value of the similarity of fault causes to be very large.

[0043] The cosine similarity calculation formula is as follows: 。

[0044] When calculating the similarity of fault causes using cosine similarity, the main improvement is made to the denominator of the Jaccard similarity calculation formula. That is, the number of fault phenomena caused by a common fault cause is used as the denominator, which alleviates to a certain extent the problem that the fault causes leading to many fault phenomena are similar to a common fault cause. However, there is still a problem with cosine similarity. Since each fault phenomenon and the cause list contribute to the interest in the cause, but the interest in the common fault phenomenon and the cause is obviously not as concentrated as that of the less frequently occurring fault phenomenon, an IUF algorithm is introduced to correct the calculation of the fault cause similarity, making it reasonable that the contribution of the common fault phenomenon to the interest in the cause is lower than that of the less frequently occurring fault phenomenon. At this time, the IUF similarity calculation formula is as follows: 。

[0045] In some embodiments, diagnosing the vehicle's fault according to the fault cause similarity includes: determining the interest value of the fault phenomenon in the fault cause according to the fault cause similarity; diagnosing the vehicle's fault according to the interest value of the fault phenomenon in the fault cause.

[0046] In some embodiments, to determine the interest value of the fault phenomenon in the fault cause according to the fault cause similarity, substitute the fault cause similarity into the following formula: , where, is the interest value of the common fault phenomenon in the fault cause n, is the set of the k fault causes most similar to the fault cause n, is the fault cause similarity, is the interest value of the common fault phenomenon in the fault cause m.

[0047] In the embodiment, after obtaining the fault cause similarity, the ItemCF-IUF algorithm calculates the interest of the fault phenomenon in the fault cause. For the implicit feedback data set, if the common fault phenomenon v is caused by the fault cause m, then let = 1, and calculate the cause more similar to the cause that the fault phenomenon was interested in historically through the above formula.

[0048] In some embodiments, diagnosing the vehicle's fault according to the interest value of the fault phenomenon in the fault cause includes: determining the priority ranking of the preset guided troubleshooting plan according to the interest value; sequentially executing the preset guided troubleshooting plan on the vehicle according to the priority ranking to diagnose the vehicle's fault.

[0049] In the embodiment, as Figure 2As shown, after determining the interest value, the priority order of the preset guided troubleshooting plan is determined according to the size of the interest value. The larger the interest value, the more likely it is to obtain a relatively high ranking in the recommended list of the preset guided troubleshooting plan, so as to realize the intelligent recommended analysis of the cause of the fault. Before recommending troubleshooting cases, obtain the corresponding repair steps for the most likely cause of the fault recommended by the recommendation system. For example, the Cypher language of the Neo4j graph database can be used to query the data stored in the knowledge graph to obtain the repair steps corresponding to the cause of the fault, and form a troubleshooting case. When the user feedbacks that the problem is solved, archive the troubleshooting case; otherwise, the system re-recommends the cause of the fault and the repair steps.

[0050] After determining the priority order of the preset guided troubleshooting plan, through remote guided troubleshooting or near-field guided troubleshooting, through the newly added user interaction function, according to the real-time feedback results after the user sequentially executes the preset guided troubleshooting plan, that is, whether the feedback result diagnoses the current fault of the vehicle, continuously optimize and adjust the subsequent diagnostic strategy to ensure that the diagnostic process is more in line with the actual needs of the user, effectively improving the flexibility and scientificity of the fault troubleshooting process. By continuously learning the latest fault data, the diagnostic process can automatically introduce new diagnostic logics and rules to improve the comprehensiveness and accuracy of the diagnosis, and solve the problem of fixed and rigid diagnostic steps in the current existing vehicle intelligent diagnostic technology.

[0051] It can be understood that during the execution of the guided troubleshooting plan, multi-platform automatic linkage is carried out. For example, it can be automatically linked with the diagnostic device. The diagnostic device is connected to the vehicle's controller area network, that is, the CAN bus, to read the data stream of each part of the vehicle in real time for automatic troubleshooting, helping technicians to more accurately locate the problem. And to assist technicians to better complete the repair work, multi-modal resources such as detailed circuit diagrams, detection guides, and relevant teaching videos are provided.

[0052] In some embodiments, the guided troubleshooting plan is sequentially executed on the vehicle according to the priority order, including: if the guided troubleshooting plan diagnoses the fault of the vehicle, determine the repair measures corresponding to the guided troubleshooting plan; if the guided troubleshooting plan does not diagnose the fault of the vehicle, determine the actual repair measures according to the user's fault information.

[0053] In the embodiment, such as Figure 2As shown in the figure, when the troubleshooting is carried out according to the guided troubleshooting plan until the end, it is oriented by whether the user's problem is solved according to the troubleshooting results. If the guided troubleshooting plan diagnoses the vehicle's fault, determine the corresponding repair measures of the guided troubleshooting plan, and automatically generate a repair case. If the guided troubleshooting plan does not diagnose the vehicle's fault and it is considered that the current fault of the vehicle cannot be diagnosed through the existing plan, then manual intervention will be carried out. Professional technicians need to give the correct repair steps, supplement and modify the deficiencies in each solution and then file them, record the real actual repair measures for further analysis and use, so as to improve the processing efficiency of similar problems in the future. Manual intervention includes but is not limited to remote OTA (Over-the-Air Technology) upgrade, on-site repair by service vehicle, vehicle repair in the store, etc.

[0054] Whether the above repair cases solved according to the guidance or not solved according to the guidance, and the repair cases after manual correction will be recorded by the system, and a complete repair case will be generated through diagnosis. The repair case includes detailed fault cause analysis, measures taken, tools and technologies used, repair time-consuming and other information; all generated repair cases are integrated by using the case self-learning method and automatically flow into the company's preset fault knowledge base to update the knowledge base as a reference for dealing with similar problems in the future; at the same time, the online learning method is used to update the parameters of the collaborative filtering algorithm model (ItemCF-IUF), update the current training data set, and then regularly fine-tune and train the AI diagnosis model until the diagnosis accuracy rate reaches the preset threshold to obtain the updated AI diagnosis model. The AI diagnosis model realizes update and iteration through closed-loop learning. Among them, the processed fault repair data can be automatically stored into the graph database using the Python language through the API (Application Programming Interface) provided by the Neo4j database, and model iteration is carried out to complete the generation and update of the diagnosis content, effectively improving the immediacy and comprehensiveness of the fault troubleshooting process, and realizing an efficient automation closed-loop from the active discovery of faults to rapid positioning, troubleshooting and case generation.

[0055] In some embodiments, obtaining a preset knowledge graph includes: determining the fault code of the vehicle according to the vehicle identification code; determining the preset fault data in the preset fault knowledge base according to the fault code; extracting the semantic parameters of the preset fault data, and the semantic parameters include entity parameters, relationship parameters and attribute parameters; determining a knowledge triple according to the entity parameters, relationship parameters and attribute parameters; determining the preset knowledge graph according to the knowledge triple.

[0056] In an embodiment, the diagnostic device can automatically identify the specific model and its configuration information of a vehicle through the vehicle's VIN (Vehicle Identification Number), so as to accurately read and identify the vehicle's fault codes.

[0057] Analyze the preset fault data in the preset fault knowledge base, such as data like fault diagnosis processes, maintenance materials, maintenance data, etc., extract the semantic parameters in the above-mentioned preset fault data, the semantic parameters include entity parameters, relationship parameters and attribute parameters, corresponding to the ontology, entities, attributes and the relationships between entities. Among them, the entity parameters include each failed part, diagnostic fault codes, such as SPN (Suspect Parameter Number), FMI (Failure Mode Identifier), P-CODE intermediate code, fault name, fault phenomenon, fault cause, test steps, test results, maintenance solutions, etc., and the ontology is the abstract concept of these entities; the attribute parameters include fault level, fault frequency, etc.; the relationship parameters are the mutual relationships between each defined ontology. Construct the schema layer of the preset knowledge graph according to the entity parameters, relationship parameters and attribute parameters.

[0058] First, perform knowledge extraction, including entity extraction, relationship extraction and attribute extraction. Among them, the entity extraction process is based on preset fault data such as fault diagnosis processes, maintenance materials, maintenance data, etc. For unstructured data, entity and relationship extraction are performed, and the extraction result is a knowledge triple in the form of <entity, relationship, entity>. This triple constitutes the basic unit of the preset knowledge graph. Fill the data layer of the preset knowledge graph with the extracted knowledge triples, and combine the data layer and the schema layer to construct the preset knowledge graph from top to bottom. Finally, use the preset knowledge graph to display the internal connections of faults.

[0059] As Figure 3 shown, the vehicle fault troubleshooting case recommendation analysis based on the knowledge graph mainly constructs a preset fault knowledge graph by using the relationships between the structural parts of different fault-occurring devices, the relationships of cause entities, etc., provides support for fault cause ranking and maintenance step recommendation analysis, and uses the ItemCF-IUF algorithm to calculate the similarity of causes between different fault phenomena, so as to realize cause recommendation for a certain fault phenomenon.

[0060] The knowledge graph construction layer first constructs the concept ontology in the case, that is, the standardized term description. At the same time, it constructs a preliminary knowledge graph based on the entities, entity relationships and entity attributes extracted from the fault repair text, combined with the latest repair cases and expert experience, to provide the collaborative filtering analysis layer with standardized terms and the relationship between fault entities and cause entities. The timely update of the latest repair cases solves the problem that the latest vehicle fault data is not fully utilized in the current existing vehicle intelligent diagnosis technology.

[0061] The collaborative filtering analysis layer mainly uses the preliminary knowledge graph constructed by the knowledge graph construction layer to realize the standardization of fault phenomena and cause names, and perform unified coding. According to the ItemCF-IUF algorithm model, the fault cause similarity matrix is constructed, and the relationship between the vehicle fault entity and the cause entity in the preliminary knowledge graph is integrated. New relationships between fault entities and cause entities are mined to realize fault phenomenon cause recommendation analysis, and the new knowledge learned is supplemented to the knowledge graph.

[0062] The application layer mainly uses the analysis results obtained by the collaborative filtering analysis layer. When a new fault phenomenon case occurs, it recommends and ranks solutions, recommends the cause of the fault phenomenon and repair steps, and calculates the new cause probability and re-recommends troubleshooting cases in real time based on user troubleshooting interactive feedback information, re-ranks solutions, and realizes rapid handling and proactive prevention of faults.

[0063] The preset fault knowledge base stores massive records of fault phenomena and causes. The ItemCF-IUF collaborative filtering analysis model is applied to find the most similar fault causes by calculating the similarity of the causes corresponding to different fault phenomena, thereby realizing intelligent recommendation of fault causes.

[0064] The preset knowledge graph is semi-automatically constructed and graphically maintained. After the component-level and system-level preset knowledge graphs are constructed, the preset knowledge graphs are manually reviewed and maintained using graphical methods.

[0065] The automated case organization system can intelligently collect, classify and archive historical fault cases, effectively avoiding the tediousness and omissions in the manual organization process. Through intelligent learning and matching algorithms, the automated case organization can quickly retrieve solutions similar to the current fault from the case library, providing maintenance personnel with accurate references and significantly improving work efficiency and diagnostic accuracy.

[0066] Reference below Figure 4 The vehicle fault diagnosis method according to the embodiment of the present invention is described with examples.

[0067] like Figure 4 As shown, the vehicle fault diagnosis method according to the embodiment of the present invention at least includes steps S11 to S24.

[0068] Step S11, obtain the abnormal fault data of the vehicle.

[0069] Step S12, determine the fault code of the vehicle according to the vehicle identification code.

[0070] Step S13, determine the preset fault data in the preset fault knowledge base according to the fault code.

[0071] Step S14, extract the semantic parameters of the preset fault data, where the semantic parameters include entity parameters, relationship parameters, and attribute parameters.

[0072] Step S15, determine the knowledge triple according to the entity parameter, relationship parameter, and attribute parameter.

[0073] Step S16, determine the preset knowledge graph according to the knowledge triple.

[0074] Step S17, obtain the fault cause number in the abnormal status data and the common fault phenomena in the preset knowledge graph.

[0075] Step S18, determine the similarity of the fault cause according to the fault cause number and the common fault phenomena.

[0076] Step S19, determine the interest value of the fault phenomenon in the fault cause according to the similarity of the fault cause.

[0077] Step S20, determine the priority ranking of the preset guided troubleshooting solutions according to the interest value.

[0078] Step S21, sequentially execute the preset guided troubleshooting solutions for the vehicle according to the priority ranking.

[0079] Step S22, determine whether the guided troubleshooting solution diagnoses the fault of the vehicle. If so, execute Step S23; otherwise, execute Step S24.

[0080] Step S23, determine the repair measures corresponding to the guided troubleshooting solution.

[0081] Step S24, determine the actual repair measures according to the user's fault information.

[0082] The fault diagnosis method of a vehicle according to an embodiment of the present invention obtains a preset knowledge graph and captures abnormal fault data in the real-time operation state of the vehicle, reads the fault features corresponding to the abnormal fault data in the preset knowledge graph, performs intelligent convergence and focusing according to the interaction between the fault features, deeply mines the correlation between the fault phenomenon and the fault cause, breaks through the limitation of a single fault point, determines the similarity of the fault cause according to the multi-fault coupling effect, so as to comprehensively cover the fault causes with different probabilities related to the abnormal fault data, and sequentially executes a plurality of guiding troubleshooting schemes according to the similarity of the fault cause to diagnose the fault of the vehicle, improving the pertinence and efficiency of the vehicle fault diagnosis.

[0083] The following refers to Figure 5 describe the vehicle fault diagnosis device 2 according to an embodiment of the present invention. The vehicle fault diagnosis device 2 is arranged on the vehicle.

[0084] As Figure 5 shown, the vehicle fault diagnosis device 2 according to an embodiment of the present invention includes: an acquisition module 21, a determination module 22 and a diagnosis module 23, wherein, The acquisition module 21 is used to acquire the abnormal fault data and the preset knowledge graph of the vehicle; the determination module 22 is used to determine the similarity of the fault cause according to the abnormal fault data and the preset knowledge graph; the diagnosis module 23 is used to diagnose the fault of the vehicle according to the similarity of the fault cause.

[0085] In the embodiment, the acquisition module 21 monitors the running state of the vehicle. When the vehicle is in a normal driving or using state, once the vehicle encounters problems during operation, the sensors installed on the vehicle will monitor the state of the vehicle in real time to acquire the abnormal fault data of the vehicle, such as instrument fault light monitoring, DTC fault code monitoring, etc., and transmit the abnormal fault data such as CAN signals and fault codes to the built-in fault model system to be able to identify potential fault modes; once the fault model system detects the abnormal fault data, it starts the fault prediction function and records these prediction information to provide basic data for subsequent fault analysis, realizing the active discovery of faults. At the same time, the preset knowledge graph corresponding to the vehicle model is acquired, which includes the latest maintenance cases, expert experience and other latest data.

[0086] The determination module 22 uses the read abnormal fault data to read the fault features corresponding to the abnormal fault data in the preset knowledge graph. By introducing a collaborative filtering algorithm such as the ItemCF-IUF algorithm, it performs convergence and focusing according to the interaction between the fault features, and deeply mines the correlation between the fault phenomenon and the fault cause to determine the similarity of the fault cause, thus solving the problem of lack of correlation analysis and intelligent convergence and focusing in the current intelligent diagnosis technology of existing vehicles.

[0087] After the diagnosis module 23 determines the similarity of the fault causes, it sorts multiple guiding troubleshooting solutions according to the numerical value of the similarity of the fault causes. The guiding troubleshooting solutions include possible fault causes and repair suggestions. When the user performs an interactive operation and obtains the fault feedback result of the vehicle to be diagnosed, the guiding troubleshooting solutions are re-recommended, that is, the fault causes and repair steps are re-sorted until the problem is solved for the user, improving the pertinence and efficiency of vehicle fault diagnosis.

[0088] According to the vehicle fault diagnosis device 2 of the embodiment of the present invention, by obtaining a preset knowledge graph and real-time capturing abnormal fault data in the vehicle running state, reading the fault features corresponding to the abnormal fault data in the preset knowledge graph, performing intelligent convergence and focusing according to the interaction between the fault features, deeply mining the correlation between the fault phenomenon and the fault cause, breaking through the limitation of a single fault point, determining the similarity of the fault causes according to the multi-fault coupling effect, so as to comprehensively cover different probability fault causes related to the abnormal fault data, and sequentially executing multiple guiding troubleshooting solutions according to the similarity of the fault causes to diagnose the vehicle fault, improving the pertinence and efficiency of vehicle fault diagnosis.

[0089] The computer-readable storage medium of the embodiment of the present invention is described below. <{

[0090] A vehicle fault diagnosis program is stored on the computer-readable storage medium of the embodiment of the present invention. When the vehicle fault diagnosis program is executed by a processor, it implements the vehicle fault diagnosis method as described in the above embodiment.

[0091] In the description of this specification, the descriptions with reference to the terms "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples", or "some examples", etc. mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example.

[0092] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and purposes of the present invention. The scope of the present invention is defined by the claims and their equivalents.

Claims

1. A fault diagnosis method for a vehicle, characterized in that, Including: Obtain the abnormal fault data and the preset knowledge graph of the vehicle; Determine the similarity of fault causes according to the abnormal fault data and the preset knowledge graph; Diagnose the faults of the vehicle according to the similarity of fault causes.

2. The fault diagnosis method of the vehicle according to claim 1, characterized in that Determining the similarity of fault causes according to the abnormal fault data and the preset knowledge graph includes: Obtain the fault cause numbers in the abnormal state data and the common fault phenomena in the preset knowledge graph; Determine the similarity of the fault causes according to the fault cause numbers and the common fault phenomena.

3. The fault diagnosis method of a vehicle according to claim 2, characterized in that, When determining the similarity of the fault causes according to the fault cause numbers and the common fault phenomena, substitute the fault cause numbers and the common fault phenomena into the following formula: ， Among them, is the similarity of the failure cause, v is the common failure phenomenon, is the number of failure phenomena caused by the failure cause m, is the number of failure phenomena caused by the failure cause n, is the number of failure phenomena of the common failure phenomenon.

4. The fault diagnosis method of a vehicle according to claim 1, characterized in that, Diagnosing the faults of the vehicle according to the similarity of fault causes includes: Determine the interest value of the fault phenomenon for the fault cause according to the similarity of the fault cause; Diagnose the faults of the vehicle according to the interest value of the fault phenomenon for the fault cause.

5. The method for diagnosing a fault of a vehicle according to claim 4, wherein Determine the interest value of the fault phenomenon for the fault cause according to the similarity of the fault cause, and substitute the similarity of the fault cause into the following formula: ， Among them, is the interest value of the common fault phenomenon for the fault cause n, is the set of the k fault causes most similar to the fault cause n, is the fault cause similarity, is the interest value of the common fault phenomenon for the fault cause m.

6. The vehicle fault diagnosis method according to claim 4, wherein, Diagnosing the faults of the vehicle according to the interest value of the fault phenomenon for the fault cause includes: Determine the priority ranking of the preset guided troubleshooting plan according to the interest value; Execute the preset guided troubleshooting plan on the vehicle in sequence according to the priority ranking to diagnose the faults of the vehicle.

7. The method for diagnosing faults of a vehicle according to claim 6, characterized in that, Executing the guided troubleshooting plan on the vehicle in sequence according to the priority ranking includes: If the guided troubleshooting plan diagnoses the faults of the vehicle, determine the maintenance measures corresponding to the guided troubleshooting plan; If the guided troubleshooting plan does not diagnose the faults of the vehicle, determine the actual maintenance measures according to the user's fault information.

8. The fault diagnosis method of a vehicle according to claim 1, characterized in that, Obtaining the preset knowledge graph includes: Determine the fault code of the vehicle according to the identification code of the vehicle; Determine the preset fault data in the preset fault knowledge base according to the fault code; Extract the semantic parameters of the preset fault data, and the semantic parameters include entity parameters, relationship parameters and attribute parameters; Determine the knowledge triple according to the entity parameter, the relationship parameter and the attribute parameter; Determine the preset knowledge graph according to the knowledge triple.

9. A fault diagnosis device for a vehicle, characterized in that, Including: An acquisition module for obtaining the abnormal fault data and the preset knowledge graph of the vehicle; A determination module for determining the similarity of fault causes according to the abnormal fault data and the preset knowledge graph; A diagnosis module for diagnosing the faults of the vehicle according to the similarity of fault causes.

10. A computer-readable storage medium, characterized in that, A vehicle fault diagnosis program is stored on the computer-readable storage medium, and when the vehicle fault diagnosis program is executed by a processor, the vehicle fault diagnosis method according to any one of claims 1-8 is implemented.

Images