Multi-mode fusion vehicle fault diagnosis method and system based on big data

Abstract

The present invention discloses a multi-mode fusion vehicle fault diagnosis method and system based on big data, the method comprising the following steps: constructing a whole vehicle electronic and electrical fault tree, and outputting the whole vehicle electronic and electrical fault tree to the cloud; building a fault knowledge graph, and outputting the fault knowledge graph to the cloud; on the basis of the whole vehicle electronic and electrical fault tree and the fault knowledge graph, constructing an intelligent diagnosis kernel with a fault identification rule model, a fault identification algorithm model, and a fault intelligent positioning model in the cloud, and after judging the vehicle fault information received in real time through the intelligent diagnosis kernel, outputting the diagnosis information to the vehicle end or the client end or the after-sales system.

Description

Technical Field

[0001] The present invention belongs to the field of automobile fault diagnosis, and specifically relates to a multi-mode fusion vehicle fault diagnosis method and system based on big data. Background Art

[0002] Vehicle failure has always been an area that vehicle manufacturers have invested a lot of energy to solve in the R&D and after-sales stages. At the very least, it will lead to a decline in user experience and user reputation. At worst, it will threaten user driving safety. Severe batch failures will also trigger large-scale vehicle recalls.

[0003] Traditional vehicle fault repair has the following problems and challenges: (1) The diagnostic instrument detects too little data, the diagnostic information is limited, and 4S stores are difficult to repair; (2) Occasional faults are difficult to reproduce, the troubleshooting time is long and it is difficult to find the real cause of the problem; (3) The one-time fault resolution rate is low and parts are repeatedly replaced. Ultimately, users complain that fault repair takes a long time, the speed is slow, and the repair is not transparent, with multiple replacements and wrong replacements. The root cause is that the vehicle manufacturer obtains too little vehicle fault information, making it difficult to quickly locate the problem and form a fault problem knowledge base for subsequent use. Therefore, how to obtain and associate more information to determine vehicle faults is an urgent problem to be solved.

[0004] At present, the Chinese patent with the publication number of 105955233 discloses a vehicle fault diagnosis method and system based on data mining; by acquiring a large amount of vehicle fault data and mining the hidden relationship between each data parameter value and between the parameter attribute value and the fault class, the faulty parts of the vehicle are promptly and effectively determined, the fault type and fault cause are determined, and the fault deterioration is prevented from causing a major accident. At the same time, the combined classification method is used to diagnose the vehicle fault data, and the vehicle fault is diagnosed by constructing multiple fault diagnosis models, which significantly improves the fault diagnosis accuracy compared with the use of a single fault diagnosis model.

[0005] This diagnostic method only mines the relationship between each data parameter value and parameter attribute value and the fault from the fault data. The parameters obtained are limited, which will affect the accuracy of the subsequent fault diagnosis result judgment. Summary of the invention

[0006] In view of the above-mentioned deficiencies in the prior art, the technical problem to be solved by the present invention is: how to provide a multi-mode fusion vehicle fault diagnosis method and system with high accuracy of vehicle fault diagnosis results.

[0007] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0008] A multi-mode fusion vehicle fault diagnosis method based on big data includes the following steps: step 1, based on the whole vehicle electronic and electrical architecture, controller boundary map, controller diagnostic code list, controller development technical requirements, vehicle maintenance manual, controller fault diagnosis list information and domain controllers of various vehicle components, construct a whole vehicle electronic and electrical fault tree, and output the whole vehicle electronic and electrical fault tree to the cloud; step 2, based on the vehicle fault code, fault reminder data, vehicle condition data, and historical maintenance data provided by the cloud, build a fault knowledge graph, and output the fault knowledge graph to the cloud; step 3, based on the whole vehicle electronic and electrical fault tree and the fault knowledge graph, construct an intelligent diagnosis kernel with a fault identification rule model, a fault identification algorithm model, and a fault intelligent positioning model in the cloud, and after judging the vehicle fault information received in real time through the intelligent diagnosis kernel, output the diagnosis information to the vehicle end or client end or after-sales system.

[0009] Furthermore, when a vehicle failure occurs, the intelligent diagnosis kernel in the cloud receives real-time fault information sent by the vehicle computer, and automatically constructs a rule set through the fault knowledge graph received by the fault identification rule model and / or the known fault information in the vehicle electronic and electrical fault tree. After extracting and identifying the real-time received fault information, the fault type is judged by the rule logic in the rule set, and the decision result is output to the vehicle side, client side or after-sales system; the fault identification algorithm model generates a fault diagnosis model based on the fault knowledge graph modeling in the cloud, and then outputs the learning result to the fault intelligent positioning model through machine deep learning. At the same time, after extracting the real-time received fault information, the fault diagnosis is performed according to the judgment rules obtained after deep learning, and the decision result is output to the vehicle side, client side or after-sales system after the fault diagnosis; the fault intelligent positioning model performs diagnostic positioning calculation on the real-time received fault information based on the deep learning result of the fault identification algorithm model, finds the location of the faulty parts and outputs a maintenance plan or fault interpretation.

[0010] Furthermore, after receiving the online data from the vehicle side, the fault identification algorithm model can also fuse and analyze the online data, and automatically sort out the correlation models between the parameters in the received online data to obtain the optimal prediction function.

[0011] Furthermore, the online data includes vehicle sensor data, owner driving data and vehicle health multi-dimensional dynamic scanning data.

[0012] Furthermore, the intelligent diagnosis kernel also includes a life / fault prediction model, which performs data fusion based on the fault diagnosis results and the optimal prediction function, evaluates the life of vehicle components, generates maintenance recommendations or fault predictions for the vehicle, and then outputs the prediction results.

[0013] Furthermore, constructing the fault knowledge graph includes knowledge acquisition, knowledge modeling, knowledge representation, knowledge fusion, knowledge storage and knowledge computing.

[0014] A multi-mode fusion vehicle fault diagnosis system based on big data, including a fault knowledge input module and an intelligent diagnosis core module, wherein the fault knowledge input module is communicatively connected with the intelligent diagnosis core module, including a vehicle electronic and electrical fault tree constructed based on the vehicle electronic and electrical architecture, a controller boundary diagram, a controller diagnostic code list, controller development technical requirements, a vehicle maintenance manual, a controller fault diagnosis list information, and domain controllers of various vehicle components, and a fault knowledge graph constructed based on vehicle fault codes, fault reminder data, vehicle condition data, and historical maintenance data provided by the cloud; the intelligent diagnosis core module is located in the cloud, including a fault identification rule model, a fault identification algorithm model, and a fault intelligent positioning model, wherein the fault identification rule model and the fault identification algorithm model are both used to receive the vehicle electronic and electrical fault tree, the fault knowledge graph, and real-time fault information issued when a vehicle fault occurs, and respectively fuse the received real-time fault information with the vehicle electronic and electrical fault tree and / or the fault knowledge graph to obtain a fault diagnosis result, and output the obtained fault diagnosis result to a vehicle end or a client end or an after-sales system.

[0015] Furthermore, the fault identification algorithm model is also used to model the fault knowledge graph to generate a fault diagnosis model, and after machine deep learning, the learning results are output to the fault intelligent location model; the fault intelligent location model is used to perform diagnostic location calculations on the fault information received in real time based on the deep learning results of the fault identification algorithm model, find the location of the faulty components and output a maintenance plan or fault interpretation.

[0016] Compared with the prior art, the multi-mode fusion vehicle fault diagnosis method and system based on big data obtained by the present invention has the following advantages: 1) Real-time automatic fault identification and fault location. For the data uploaded to the cloud by the vehicle in real time, the diagnosis kernel automatically identifies which module of the fault tree the fault belongs to through the rules and algorithms obtained by knowledge graph training, and then locates the fault according to the fault set of the corresponding module in the fault tree; 2) Actively provide fault handling opinions. After the vehicle-side fault is identified and located in real time, the user is provided with fault handling opinions based on the fault handling opinions maintained by the knowledge graph; 3) The identified faults and maintenance suggestions are pushed to the user in categories to remind the user to repair it independently, and at the same time pushed to the dealer, so as to provide personalized fault maintenance services for the user. Through this system, user faults can be automatically identified in real time and maintenance suggestions can be provided. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a multi-mode fusion vehicle fault diagnosis flow chart in the embodiment;

[0018] Figure 2 The construction content of the fault tree in the embodiment;

[0019] Figure 3 The construction content of the fault knowledge graph in the embodiment;

[0020] Figure 4 4 is a system diagram of the intelligent diagnosis kernel in the embodiment. DETAILED DESCRIPTION

[0021] The present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0022] Example:

[0023] like Figure 1-Figure 4 As shown, the multi-mode fusion vehicle fault diagnosis method based on big data provided in this embodiment includes the following steps: Step 1, based on the whole vehicle electronic and electrical architecture, controller boundary map, controller diagnostic code list, controller development technical requirements, vehicle maintenance manual, controller fault diagnosis list information and domain controllers of each component of the vehicle, build a whole vehicle electronic and electrical fault tree, and output the whole vehicle electronic and electrical fault tree to the cloud; Step 2, based on the vehicle fault code, fault reminder data, vehicle condition data, and historical maintenance data provided by the cloud, build a fault knowledge graph, and output the fault knowledge graph to the cloud; Step 3, based on the whole vehicle electronic and electrical fault tree and the fault knowledge graph, build an intelligent diagnosis kernel with a fault identification rule model, a fault identification algorithm model, and a fault intelligent positioning model in the cloud, and after judging the vehicle fault information received in real time through the intelligent diagnosis kernel, output the diagnosis information to the vehicle end or the client end or the after-sales system. Specifically, when a vehicle failure occurs, the intelligent diagnosis kernel in the cloud receives the real-time fault information sent by the vehicle computer, and automatically constructs a rule set through the fault knowledge graph received by the fault identification rule model and / or the known fault information in the vehicle electronic and electrical fault tree. After extracting and identifying the real-time received fault information, the fault type is judged by the rule logic in the rule set, and the decision result is output to the vehicle side, client side or after-sales system; the fault identification algorithm model generates a fault diagnosis model based on the fault knowledge graph modeling in the cloud, and then outputs the learning result to the fault intelligent positioning model through machine deep learning. At the same time, after extracting the real-time received fault information, the fault diagnosis is performed according to the judgment rules obtained after deep learning, and the decision result is output to the vehicle side, client side or after-sales system after the fault diagnosis; the fault intelligent positioning model performs diagnostic positioning calculation on the real-time received fault information based on the deep learning result of the fault identification algorithm model, finds the location of the faulty parts and outputs a maintenance plan or fault interpretation.

[0024] Specifically, in this embodiment, the vehicle intelligent diagnosis technology is divided into two categories: 1) For hidden faults, multi-component cross-linking faults, new faults, etc., the node, edge, and associated parameter information of all domain controllers under the vehicle electronic and electrical architecture are combined with historical data to train the model by constructing a fault tree. 2) For vehicle faults with obvious fault phenomena, clear fault codes, and existing analysis and solution experience, they can be searched and queried in the constructed fault knowledge base.

[0025] When a vehicle fault occurs, the fault information will enter the intelligent diagnosis kernel, and the fault information feature value will be extracted by the fault identification rule model, and then the fault identification rule model will judge the fault type (knowledge graph / fault tree) based on the feature value; after the data enters the algorithm model, based on the fault information library and intelligent troubleshooting logic algorithm of the big data platform, the key information analysis and judgment will be completed, and finally the judgment result will be transmitted to the fault intelligent location model to find the root cause of the problem and output the maintenance plan or fault interpretation. In this process, the final feature information and judgment conditions of the fault will be recorded by the prediction model and the diagnosis model, so that the entire diagnosis model will be continuously enriched.

[0026] like Figure 2 As shown in the figure, the fault tree is based on the vehicle network topology, controller boundaries, related controller development technical requirements, controller fault diagnosis list and other information. The status of the vehicle's electronic appliances is split from the system level to the component level and then to the signal level. It is then uniformly defined, graded and classified. Finally, all system-level and component-level fault information is compiled into a fault tree, which can support the adjustment of the fault tree relationship and structure at any time.

[0027] The above fault identification algorithm model can also fuse and analyze the online data after receiving the online data from the vehicle side, and automatically sort out the correlation model between the parameters in the received online data to obtain the optimal prediction function.

[0028] Furthermore, the online data includes vehicle sensor data, owner driving data and vehicle health multi-dimensional dynamic scanning data.

[0029] Furthermore, the intelligent diagnosis kernel also includes a life / fault prediction model, which performs data fusion based on the fault diagnosis results and the optimal prediction function, evaluates the life of vehicle components, generates maintenance recommendations or fault predictions for the vehicle, and then outputs the prediction results.

[0030] like Figure 3As shown in the figure, building a fault knowledge graph includes knowledge acquisition, knowledge modeling, knowledge representation, knowledge fusion, knowledge storage and knowledge calculation. Knowledge modeling is carried out in the form of ontology modeling (the tool is Protege). At the same time, different ontology models are built according to different knowledge types, and different ontology models have certain uniformity (classes, attributes, relationships, etc.), which is also helpful for the subsequent realization of cross-graph knowledge reasoning.

[0031] A multi-mode fusion vehicle fault diagnosis system based on big data, including a fault knowledge input module and an intelligent diagnosis core module, wherein the fault knowledge input module is communicatively connected with the intelligent diagnosis core module, including a vehicle electronic and electrical fault tree constructed based on the vehicle electronic and electrical architecture, a controller boundary diagram, a controller diagnostic code list, controller development technical requirements, a vehicle maintenance manual, a controller fault diagnosis list information, and domain controllers of various vehicle components, and a fault knowledge graph constructed based on vehicle fault codes, fault reminder data, vehicle condition data, and historical maintenance data provided by the cloud; the intelligent diagnosis core module is located in the cloud, including a fault identification rule model, a fault identification algorithm model, and a fault intelligent positioning model, wherein the fault identification rule model and the fault identification algorithm model are both used to receive the vehicle electronic and electrical fault tree, the fault knowledge graph, and real-time fault information issued when a vehicle fault occurs, and respectively fuse the received real-time fault information with the vehicle electronic and electrical fault tree and / or the fault knowledge graph to obtain a fault diagnosis result, and output the obtained fault diagnosis result to a vehicle end or a client end or an after-sales system.

[0032] Furthermore, the fault identification algorithm model is also used to model the fault knowledge graph to generate a fault diagnosis model, and after machine deep learning, the learning results are output to the fault intelligent location model; the fault intelligent location model is used to perform diagnostic location calculations on the fault information received in real time based on the deep learning results of the fault identification algorithm model, find the location of the faulty components and output a maintenance plan or fault interpretation.

[0033] The intelligent diagnosis kernel module is developed and runs based on the cloud platform. It includes a fault information knowledge repository (Docker container), training middleware (K8S, AI JOB, heterogeneous enhanced scheduler, GPU Device), and an intelligent diagnosis kernel (fault identification rule model, fault identification algorithm model, fault intelligent location model, life / fault prediction model). It can analyze the fault information library, systematically reconstruct existing data, and complete the extraction, analysis, and learning of key information. It combines real-time vehicle fault diagnosis, user historical maintenance and other feedback information, and uses diagnostic algorithms to output the best fault predictions in different dimensions to complete intelligent fault diagnosis.

[0034] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit the technical solution. Although the applicant has described the present invention in detail with reference to the preferred embodiments, ordinary technicians in this field should understand that those modifications or equivalent substitutions of the technical solution of the present invention without departing from the purpose and scope of the technical solution should be included in the scope of the claims of the present invention.

Claims

1. A multi-mode fusion vehicle fault diagnosis method based on big data, characterized by: The method comprises the following steps: Step 1, based on the whole vehicle electronic and electrical architecture, controller boundary diagram, controller diagnostic code list, controller development technical requirements, vehicle maintenance manual, controller fault diagnosis list information and domain controllers of each vehicle component, construct a whole vehicle electronic and electrical fault tree, and output the whole vehicle electronic and electrical fault tree to the cloud; Step 2, based on the vehicle fault code, fault reminder data, vehicle condition data and historical maintenance data provided by the cloud, construct a fault knowledge graph, and output the fault knowledge graph to the cloud; Step 3, based on the whole vehicle electronic and electrical fault tree and the fault knowledge graph, construct an intelligent diagnosis kernel with a fault identification rule model, a fault identification algorithm model and a fault intelligent positioning model in the cloud, and after judging the vehicle fault information received in real time through the intelligent diagnosis kernel, output the diagnosis information to the vehicle end or the client end or the after-sales system; When a vehicle fault occurs, the intelligent diagnosis kernel in the cloud receives the real-time fault information from the vehicle computer, and automatically builds a rule set through the fault knowledge graph received by the fault identification rule model and / or the known fault information in the vehicle electronic and electrical fault tree. After extracting and identifying the real-time received fault information, the fault type is determined through the rule logic in the rule set, and the decision result is output to the vehicle, client or after-sales system. The fault identification algorithm model generates a fault diagnosis model based on the fault knowledge graph modeling in the cloud, and then outputs the learning results to the fault intelligent location model through machine deep learning. At the same time, after extracting the fault information received in real time, the fault diagnosis is performed according to the judgment rules obtained after deep learning, and the decision results are output to the vehicle end, the client end or the after-sales system after the fault diagnosis; The intelligent fault location model performs diagnostic location calculations on the fault information received in real time based on the deep learning results of the fault identification algorithm model, finds the location of the faulty component and outputs a maintenance plan or fault interpretation.

2. The multi-mode fusion vehicle fault diagnosis method based on big data according to claim 1 is characterized in that: The fault identification algorithm model can also fuse and analyze the online data after receiving the online data from the vehicle end, and automatically sort out the correlation model between the parameters in the received online data to obtain the optimal prediction function.

3. The multi-mode fusion vehicle fault diagnosis method based on big data according to claim 2 is characterized in that: The online data includes vehicle sensor data, vehicle owner driving data and vehicle health multi-dimensional dynamic scanning data.

4. The multi-mode fusion vehicle fault diagnosis method based on big data according to claim 1, 2 or 3, characterized in that: The intelligent diagnosis kernel also includes a life / fault prediction model, which performs data fusion based on the fault diagnosis results and the optimal prediction function, evaluates the life of vehicle components, generates maintenance suggestions or fault predictions for the vehicle, and then outputs the prediction results.

5. The multi-mode fusion vehicle fault diagnosis method based on big data according to claim 1 is characterized in that: Constructing fault knowledge graph includes knowledge acquisition, knowledge modeling, knowledge representation, knowledge fusion, knowledge storage and knowledge computing.

6. A multi-mode fusion vehicle fault diagnosis system based on big data, characterized in that: It includes a fault knowledge input module and an intelligent diagnosis kernel module, wherein the fault knowledge input module is in communication connection with the intelligent diagnosis kernel module, and includes a vehicle electronic and electrical fault tree constructed based on the vehicle electronic and electrical architecture, controller boundary diagram, controller diagnostic code list, controller development technical requirements, vehicle maintenance manual, controller fault diagnosis list information, and domain controllers of various vehicle components, as well as a fault knowledge graph constructed based on vehicle fault codes, fault reminder data, vehicle condition data, and historical maintenance data provided by the cloud; The intelligent diagnosis core module is located in the cloud, and includes a fault identification rule model, a fault identification algorithm model and a fault intelligent location model. The fault identification rule model and the fault identification algorithm model are used to receive the vehicle electronic and electrical fault tree, the fault knowledge graph and the real-time fault information issued when the vehicle fault occurs, and respectively fuse the received real-time fault information with the vehicle electronic and electrical fault tree and / or the fault knowledge graph to obtain the fault diagnosis result, and output the obtained fault diagnosis result to the vehicle end or the client end or the after-sales system; When a vehicle fault occurs, the intelligent diagnosis kernel in the cloud receives the real-time fault information from the vehicle computer, and automatically builds a rule set through the fault knowledge graph received by the fault identification rule model and / or the known fault information in the vehicle electronic and electrical fault tree. After extracting and identifying the real-time received fault information, the fault type is determined through the rule logic in the rule set, and the decision result is output to the vehicle, client or after-sales system. The fault identification algorithm model generates a fault diagnosis model based on the fault knowledge graph modeling in the cloud, and then outputs the learning results to the fault intelligent location model through machine deep learning. At the same time, after extracting the fault information received in real time, the fault diagnosis is performed according to the judgment rules obtained after deep learning, and the decision results are output to the vehicle end, the client end or the after-sales system after the fault diagnosis; The intelligent fault location model performs diagnostic location calculations on the fault information received in real time based on the deep learning results of the fault identification algorithm model, finds the location of the faulty component and outputs a maintenance plan or fault interpretation.

Images