Vehicle fault detection method, device, equipment and storage medium

Abstract

The application provides a vehicle fault detection method, which comprises the following steps: analyzing key information related to a fault according to vehicle information of a vehicle to be detected; obtaining at least one fault cause according to the key information and a fault knowledge graph, for user confirmation; performing emotional analysis on the confirmation information of the user; and diagnosing the fault cause of the vehicle to be detected from the at least one fault cause according to the result of the emotional analysis. Thus, the over-reliance on maintenance personnel is reduced, the fault cause is intelligently found, the accuracy of fault detection is improved, and the maintenance efficiency is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of fault detection, and in particular to a vehicle fault detection method, device, equipment and storage medium. BACKGROUND

[0002] With the popularity of automobiles, 4S and other repair sites consume a certain amount of time and manpower in troubleshooting, and the detection process combines subjective detection of vehicle faults, and due to the dependence on maintenance personnel, misdiagnosis or misdiagnosis often occurs during troubleshooting. SUMMARY

[0003] The embodiments of the present application provide a vehicle fault detection method, device, equipment and storage medium to solve the problems of related technologies, and the technical solutions are as follows:

[0004] In a first aspect, the embodiments of the present application provide a vehicle fault detection method, comprising:

[0005] According to the vehicle information of the vehicle to be detected, the key information related to the fault is parsed;

[0006] According to the key information and the fault knowledge graph, at least one fault reason is obtained for the user to confirm;

[0007] The confirmation information of the user is analyzed;

[0008] According to the result of the sentiment analysis, the fault reason of the vehicle to be detected is diagnosed among the at least one fault reason.

[0009] In an embodiment, the key information related to the fault is parsed according to the vehicle information of the vehicle to be detected, comprising:

[0010] The vehicle information of the vehicle to be detected is segmented and word vector structured to obtain the key information related to the fault.

[0011] In an embodiment, the above method further comprises:

[0012] According to the basic physical part information of multiple vehicle models, a basic physical part knowledge graph is constructed;

[0013] According to the historical fault interaction data, a triple containing fault phenomenon, sentiment analysis result and fault keyword is constructed; and

[0014] According to the triple, the fault reason in the historical fault interaction data and the basic physical part knowledge graph are associated to obtain a fault knowledge graph.

[0015] In an embodiment, the triple containing fault phenomenon, sentiment analysis result and fault keyword is constructed according to the historical fault interaction data, comprising:

[0016] The inquiry data in the historical fault interaction data is parsed by a named entity recognition model to obtain fault keywords;

[0017] The reply data in the historical fault interaction data is parsed by a sentiment analysis recurrent neural network to obtain a sentiment analysis result;

[0018] The fault keywords, the sentiment analysis result, and the fault phenomenon in the historical fault interaction data are used to construct a triple.

[0019] In an embodiment, the fault cause in the historical fault interaction data is associated with the basic physical component knowledge graph according to the triple to obtain a fault knowledge graph, including:

[0020] According to the fault keyword in the triple, matched basic physical component information is found in the basic physical component knowledge graph, and the fault phenomenon in the triple to which the fault keyword belongs is associated with the matched basic physical component information.

[0021] The fault cause corresponding to the phenomenon is obtained from the historical fault interaction data, and the obtained fault cause is associated with the fault phenomenon to obtain the fault knowledge graph.

[0022] In an embodiment, the confirmation information of the user is subjected to sentiment analysis, including:

[0023] The confirmation information of the user is subjected to sentiment analysis using a sentiment analysis recurrent neural network to determine the sentiment type of the confirmation information.

[0024] In an embodiment, according to the result of the sentiment analysis, the fault cause of the vehicle to be detected is diagnosed from at least one fault cause, including:

[0025] When the user has a positive sentiment for an item in the at least one fault cause according to the sentiment analysis, the fault cause is diagnosed as the fault cause of the vehicle to be detected.

[0026] In an embodiment, the key information related to the fault is parsed from the vehicle information of the vehicle to be detected, including:

[0027] It is determined whether the vehicle information of the vehicle to be detected contains a fault code, and in the case where it is determined that the vehicle information does not contain a fault code, the key information related to the fault is parsed from the vehicle information.

[0028] In an embodiment, the above method further includes:

[0029] In the case where it is determined that none of the at least one fault cause is the fault cause of the vehicle to be detected according to the sentiment analysis, manual diagnosis is prompted, and the result of the manual diagnosis is updated to the fault knowledge graph.

[0030] In an implementation, the method further comprises:

[0031] According to the diagnosed fault cause of the to-be-detected vehicle, a corresponding fault solution is determined and provided to the user.

[0032] In a second aspect, the embodiments of the present application provide a vehicle fault detection device, comprising:

[0033] The analysis module is configured to analyze the confirmation information of the user.

[0034] The processing module is configured to derive at least one fault cause according to the key information and the fault knowledge graph, for the user to confirm.

[0035] The analysis module is configured to analyze the confirmation information of the user.

[0036] The fault detection module is configured to diagnose the fault cause of the to-be-detected vehicle from the at least one fault cause according to the result of the sentiment analysis.

[0037] In an implementation, the analysis module comprises:

[0038] The first analysis unit is configured to perform word segmentation and word vector structural processing on the vehicle information of the to-be-detected vehicle to obtain the key information related to the fault.

[0039] In an implementation, the device further comprises:

[0040] The base physical part knowledge graph construction module is configured to construct a base physical part knowledge graph according to base physical part information of a plurality of vehicle models.

[0041] The triple construction module is configured to construct triples containing fault phenomena, sentiment analysis results, and fault keywords according to historical fault interaction data.

[0042] The fault knowledge graph construction module is configured to associate fault causes in the historical fault interaction data with the base physical part knowledge graph according to the triples to obtain a fault knowledge graph.

[0043] In an implementation, the triple construction module comprises:

[0044] The second analysis unit is configured to analyze inquiry data in the historical fault interaction data through a named entity recognition model to obtain fault keywords.

[0045] The third analysis unit is configured to analyze reply data in the historical fault interaction data through a sentiment analysis recurrent neural network to obtain sentiment analysis results.

[0046] The constructing unit is configured to construct a triple using the fault keyword, the sentiment analysis result, and a fault phenomenon in the historical fault interaction data.

[0047] In an embodiment, the fault knowledge graph constructing module comprises:

[0048] The first matching and associating unit is configured to find matched basic physical part information in the basic physical part knowledge graph according to the fault keyword in the triple, and associate the fault phenomenon in the triple to which the fault keyword belongs with the matched basic physical part information.

[0049] The second matching and associating unit is configured to obtain a fault cause corresponding to the fault phenomenon from the historical fault interaction data, and associate the obtained fault cause with the fault phenomenon to obtain the fault knowledge graph.

[0050] In an embodiment, the analyzing module comprises:

[0051] The sentiment analysis unit is configured to perform sentiment analysis on the confirmation information of the user using a sentiment analysis recurrent neural network to determine a sentiment type of the confirmation information.

[0052] In an embodiment, the fault detecting module comprises:

[0053] The sub-processing unit is configured to diagnose the fault cause as a fault cause of the vehicle to be detected when the user has a positive sentiment for one of the at least one fault cause according to the sentiment analysis.

[0054] In an embodiment, the analyzing module comprises:

[0055] The judging unit is configured to determine whether the vehicle information of the vehicle to be detected contains a fault code, and parse out key information related to the fault from the vehicle information when it is determined that the vehicle information does not contain the fault code.

[0056] In an embodiment, the device further comprises:

[0057] The artificial diagnosis module is configured to prompt for artificial diagnosis when it is determined that none of the at least one fault cause is a fault cause of the vehicle to be detected according to the sentiment analysis, and update a result of the artificial diagnosis to the fault knowledge graph.

[0058] In an embodiment, the device further comprises:

[0059] The scheme determining module is configured to determine a corresponding fault solution according to the diagnosed fault cause of the vehicle to be detected and provide the fault solution to the user.

[0060] In a third aspect, an embodiment of the present application provides a vehicle fault detection system, the device comprising: a memory and a processor. Wherein the memory and the processor communicate with each other through an internal connection path, the memory is used to store instructions, the processor is used to execute the instructions stored in the memory, and when the processor executes the instructions stored in the memory, the processor executes the method in any of the embodiments of the above aspects.

[0061] In a fourth aspect, an embodiment of the present application provides a computer readable storage medium, which stores a computer program, when the computer program runs on a computer, the method in any of the embodiments of the above aspects is executed.

[0062] The advantages or beneficial effects of the above technical solutions at least include: by analyzing the key information in the vehicle information, combining the fault knowledge graph to obtain at least one fault cause for the user to confirm, and then performing sentiment analysis on the confirmation information of the user, the fault cause of the vehicle to be detected is diagnosed in at least one fault cause according to the analysis result, thereby reducing the over-reliance on maintenance personnel, intelligently finding the fault cause, improving the accuracy of fault detection, and improving the maintenance efficiency.

[0063] The above summary is only for the purpose of the description and is not intended to limit in any way. In addition to the above described illustrative aspects, embodiments and features, further aspects, embodiments and features of the present application will be readily apparent to those skilled in the art by reference to the drawings and the following detailed description. BRIEF DESCRIPTION OF DRAWINGS

[0064] In the drawings, like reference numerals refer to same or similar components throughout the several views. These drawings are not necessarily to scale. It should be understood that these drawings only depict some embodiments in accordance with the disclosure and should not be considered to be limiting of the scope of the disclosure.

[0065] Figure 1 is a flowchart of a vehicle fault detection method according to an embodiment of the present disclosure;

[0066] Figure 2 is a schematic diagram of a fault knowledge graph according to an embodiment of the present disclosure;

[0067] Figure 3 is a flowchart of a fault knowledge graph construction method according to an embodiment of the present disclosure;

[0068] Figure 4 is a flowchart of a vehicle fault detection method according to another embodiment of the present disclosure;

[0069] Figure 5is a flowchart of a vehicle fault detection method according to an embodiment of the present disclosure;

[0070] Figure 6 is a schematic block diagram of a vehicle fault detection device according to an embodiment of the present disclosure;

[0071] Figure 7 is a schematic block diagram of constructing a fault knowledge graph in a vehicle fault detection device according to an embodiment of the present disclosure;

[0072] Figure 8 is a schematic block diagram of a vehicle fault detection device according to an embodiment of the present disclosure;

[0073] Figure 9 is a block diagram of a vehicle fault detection system for implementing an embodiment of the present disclosure. DETAILED DESCRIPTION

[0074] Hereinafter, only some exemplary embodiments are simply described. As those skilled in the art can recognize, the described embodiments can be modified in various different ways without departing from the spirit or scope of the present application. Therefore, the drawings and the description are considered to be exemplary in nature rather than limiting.

[0075] Figure 1 is a flowchart of a vehicle fault detection method according to an embodiment of the present application. As shown in Figure 1 , the vehicle fault detection method can include:

[0076] S11, according to the vehicle information of the vehicle to be detected, the key information related to the fault is parsed.

[0077] It should be noted that the control end of the vehicle fault detection is preferably a vehicle fault detection system (or even a cloud server, etc.). It can be understood that the vehicle fault detection system can be integrated in the vehicle control system (i.e. ECU), or can be independent of the vehicle control system. The detection system can be realized by software and / or hardware, and the detection system can be composed of two or more physical entities, or can be composed of one physical entity.

[0078] The vehicle information can be text information uploaded by the user to the detection system, and can also be voice information uploaded by the user to the detection system, and then the voice information is converted into text information, which is not limited herein. Exemplarily, the voice of the user can be collected by a sound collector on the vehicle, and then the voice is recognized into corresponding text information. For another example, the user collects the voice of the user through the microphone of the mobile phone and uploads it to the detection system. It can be understood that the above are examples given by the present application, and the present application does not limit the way of sound collection and analysis, and all ways of sound collection and analysis are within the protection scope of the present application.

[0079] Exemplarily, the vehicle information is a current vehicle fault phenomenon found by the user, for example, the user says "the car cannot start", then the voice of the user is converted into text for further analysis to obtain the key information "cannot start".

[0080] S12, at least one fault reason is obtained according to the key information and the fault knowledge graph for the user to confirm.

[0081] It should be noted that the fault is obtained according to the key information and the fault knowledge graph, and at least one fault reason is provided for the user to confirm. The knowledge graph includes the system, component and part corresponding to each vehicle model, and each part has a corresponding fault phenomenon, and the fault phenomenon corresponds to at least one fault reason. In the embodiment, the key information can be indexed and searched for the component and part that fails by using Match language, and the component and part that fails can also be indexed and searched by using a decision tree or a binary tree algorithm. It can be understood that the vehicle model number and other information can also be added for indexing, which can quickly find the fault reason.

[0082] Exemplarily, when the key information is "cannot start", since "cannot start" is a fault phenomenon, at least one fault reason corresponding to the fault phenomenon is searched in the fault knowledge graph and sent to the user for confirmation. It can be understood that the voice question and answer can be performed between the user and the vehicle voice interaction system, and the information can also be sent to the user in the form of a short message or an email, which is not limited herein.

[0083] S13, the confirmation information of the user is analyzed.

[0084] It should be noted that the confirmation information of the user can be in the form of text or in the form of voice, which is not limited herein.

[0085] It should be noted that the sentiment analysis is an analysis of the content of the confirmation information, which can be positive emotion, negative emotion or other emotion. The sentiment analysis can be analyzed by a neural network model, a self-attention mechanism model or the like, which is not limited herein.

[0086] S14, diagnosing the fault cause of the vehicle to be detected from the at least one fault cause according to the result of the sentiment analysis.

[0087] Further, when the user has a positive sentiment for one of the at least one fault cause according to the sentiment analysis, the fault cause is diagnosed as the fault cause of the vehicle to be detected.

[0088] It can be understood that when the sentiment analysis structure is negative sentiment or other sentiment, it means that the current fault cause is not the fault cause of the vehicle to be detected.

[0089] For example, when the user answers "yes" to the question "Is the oil tank found to be leaking?", it means that the oil tank is the fault cause of the vehicle to be detected; when the user answers "no", it means that the oil tank is not the fault cause of the vehicle to be detected.

[0090] In summary, by analyzing the key information in the vehicle information, combining the fault knowledge graph to obtain at least one fault cause for the user to confirm, and then performing sentiment analysis on the user's confirmation information, the fault cause of the vehicle to be detected is diagnosed from the at least one fault cause according to the analysis result, thereby reducing the over-reliance on maintenance personnel, intelligently finding the fault cause, improving the accuracy of fault detection, and improving the maintenance efficiency.

[0091] In one embodiment, step S11, the key information related to the fault is analyzed according to the vehicle information of the vehicle to be detected, comprising:

[0092] The vehicle information of the vehicle to be detected is cut and word vector structured to obtain the key information related to the fault.

[0093] Specifically, the vehicle information is cut by Jieba to obtain the corresponding word vector, so as to realize the cutting of natural language and propose some stop words (such as mood words). For example, "the engine of the car is damaged", the cutting result is [‘car’, ‘engine’, ‘damage’].

[0094] Since the natural language cannot be recognized in the calculation process, the key information is obtained by structured processing by Word2vec, the non-structured word vector is converted into a structured vector, and the natural language information can be preserved after Word2vec conversion, so as to facilitate subsequent calculation of the key information, and also without losing the original information. For example, the word vector conversion preferentially one-hot encodes the word, Word2vec randomly initializes the W weight matrix, and performs inner product operation on the one-hot encoded conversion, and obtains θ by weighted average calculation on the inner product result. The word2vec adds a full connection network W at the end, and the word vector = θW.

[0095] In one embodiment, as shown in Figures 2-3 The method for detecting vehicle failure further comprises:

[0096] S31, constructing a basic physical part knowledge graph according to the basic physical part information of multiple vehicle models.

[0097] In the present embodiment, the attributes and numbers of each system, component and part under the same vehicle model are extracted. The nodes of each system, component and part are constructed in the gallery using the Match creation statement, and the relationships between the nodes are created according to the Match. For example, the basic physical part knowledge graph is constructed according to the composition of the vehicle. The vehicle model and the systems under the vehicle model are one kind of relationship. Each system and the components under the system are another kind of relationship. Each component and the parts under the component are another kind of relationship. The basic physical part knowledge graph is constructed through the above-mentioned several kinds of relationships. It should be noted that the Match language can perform the addition, deletion, query and modification operations of the graph database. The gallery can be a Neo4J gallery, and can also be other galleries, which are not limited herein.

[0098] S32, constructing a triple containing a failure phenomenon, an emotional analysis result and a failure keyword according to historical failure interaction data.

[0099] In the present embodiment, the historical failure interaction data refers to the question and answer content between the maintenance personnel and the user. It can be understood that the failure cause of the vehicle can be determined through the question and answer content between the maintenance personnel and the user. Therefore, the failure phenomenon, the failure keyword and the emotional analysis result are extracted from the historical failure interaction data, and the relationship among the three is established, which helps to quickly establish the relationship between the failure cause and the original failure knowledge graph.

[0100] Exemplarily, the failure keyword is, for example, “oil tank leakage”, “battery power loss” and the like. The failure phenomenon is “not starting”, and the emotional analysis result refers to “yes”, “no” and the like. Therefore, the triple content can be not starting-yes-battery power loss, or not starting-no-oil tank leakage.

[0101] S33, associating the failure cause in the historical failure interaction data with the basic physical part knowledge graph according to the triple, to obtain a failure knowledge graph.

[0102] It should be noted that since the triple includes the failure keyword and the failure phenomenon, the failure cause and the basic physical part knowledge graph can be associated through the failure keyword, and the failure cause and the basic physical part knowledge graph can also be associated through the failure phenomenon, so as to obtain the failure knowledge graph. Therefore, the user can determine the failure cause according to the failure knowledge graph, and reduce the dependence on the maintenance personnel.

[0103] In an embodiment, step S20, constructing a triple including a fault phenomenon, a sentiment analysis result and a fault keyword according to historical fault interaction data, includes:

[0104] The inquiry data in the historical fault interaction data is parsed by a named entity recognition model to obtain the fault keyword.

[0105] The reply data in the historical fault interaction data is parsed by a sentiment analysis recurrent neural network to obtain the sentiment analysis result.

[0106] The fault keyword, the sentiment analysis result and the fault phenomenon in the historical fault interaction data are used to construct a triple.

[0107] It should be noted that the named entity recognition model can be a bidirectional RNN structure combined with a CRF model, or a Bi-LSMT combined with a CRF model. In this embodiment, a Bi-LSMT combined with a CRF model is used. The training process of the Bi-LSTM+CRF model is as follows. The inquiry data of the maintenance personnel is converted into a word vector, and the fault keyword is manually labeled. Then, the word vector and the fault keyword in the inquiry data are converted into numbers, and the converted inquiry data and the converted fault keyword are trained by the Bi-LSTM+CRF model, so that the Bi-LSTM+CRF model can recognize the inquiry data of the maintenance personnel.

[0108] The training of the word vector by the Bi-LSMT considers the situation before and after the current time, so the effect of natural language processing is better, thereby reducing the recognition error. The Bi-LSMT combined with the CRF model is to find the probability of the most likely sequence labeling in a sequence under the condition that the probability distribution of each word is known. The hidden layer (N-dimensional vector) calculated by the Bi-LSTM is transmitted to the CRF model, so that the CRF model is used to process the local normalization problem existing in the Bi-LSTM.

[0109] For example, the following formula can be specifically referred to:

[0110]

[0111] In this embodiment, the sentiment analysis recurrent neural network is an LSTM neural network. The LSTM neural network is trained so that it can judge the sentiment of an input sentence, such as positive sentiment, negative sentiment and other sentiment.

[0112] It should be noted that the training process of the sentiment analysis recurrent neural network is to convert the reply data (i.e. the reply data of the user) into a word vector and manually identify the meaning of the reply data, 0 represents yes, 1 represents no, and 2 represents other. The converted word vector and the meaning are trained by the LSTM recurrent neural network, so that the LSTM recurrent neural network can perform sentiment analysis.

[0113] In an embodiment, step S33, the fault reason in the historical fault interaction data is associated with the basic physical knowledge graph according to the triple, and the fault knowledge graph is obtained, including:

[0114] According to the fault keyword in the triple, the matching basic physical information is found in the basic physical knowledge graph, and the fault phenomenon in the triple to which the fault keyword belongs is associated with the matching basic physical information;

[0115] The fault reason corresponding to the fault phenomenon is obtained from the historical fault interaction data, and the obtained fault reason is associated with the fault phenomenon to obtain the fault knowledge graph.

[0116] It should be noted that in the historical fault interaction data, the fault reason corresponding to the fault phenomenon is determined according to the inquiry data and the reply data. In this embodiment, the corresponding component or part is matched through the fault keyword, and then the fault phenomenon node is constructed using the Match creation statement, the fault phenomenon node is spliced to the matched component or part, and then the fault reason node causing the fault phenomenon is established using the Match creation statement and spliced to the fault phenomenon.

[0117] In an embodiment, step S13, the confirmation information of the user is subjected to sentiment analysis, including:

[0118] The confirmation information of the user is subjected to sentiment analysis using a sentiment analysis recurrent neural network to determine the sentiment type of the confirmation information.

[0119] In this embodiment, the confirmation information of the user is input into the LSTM neural network, and the LSTM neural network is analyzed to determine the corresponding sentiment type, and whether to continue to let the user confirm is determined according to positive sentiment, negative sentiment or other sentiment.

[0120] In an embodiment, step S14, according to the result of the sentiment analysis, the fault reason of the vehicle to be detected is diagnosed from at least one fault reason, including:

[0121] When the user has a positive sentiment for one of the at least one fault reason according to the sentiment analysis, the fault reason is diagnosed as the fault reason of the vehicle to be detected.

[0122] In the embodiment, when multiple fault causes are determined according to the key information, the user is sequentially asked about the fault cause through voice questioning, for example, the fault cause is converted into voice by the vehicle-mounted voice interaction system to ask the user. According to the voice answer of the user, the emotional type is judged, and the inquiry is ended until the emotional type is positive emotion, and the fault cause of the vehicle to be detected is determined.

[0123] In an embodiment, step S11, the key information related to the fault is parsed from the vehicle information of the vehicle to be detected, including:

[0124] It is judged whether the fault code is contained in the vehicle information of the vehicle to be detected, and in the case where it is determined that the fault code is not contained in the vehicle information, the key information related to the fault is parsed from the vehicle information.

[0125] Specifically, the automobile fault code is the fault code reflected by the ECU after the automobile fault occurs. In the embodiment, if the fault code is not contained in the vehicle information, it means that the fault cause cannot be directly determined, and therefore, the vehicle information needs to be further parsed to determine the fault cause.

[0126] Further, it is judged that the fault code is contained in the vehicle information of the vehicle to be detected, and the corresponding fault cause is directly found according to the fault code.

[0127] In the embodiment, the corresponding fault code is generally displayed on the instrument panel or console to prompt the user to the corresponding fault cause. For example, the user sees P0123 displayed on the console, and the user can input “P0123” by voice or text to the detection system, the detection system finds the corresponding fault cause of P0123, and then sends the fault cause to the customer. For example, the fault cause corresponding to the fault code is found in the OBD diagnosis system. For another example, the original data of the host factory system is extracted in advance, and the corresponding relationship between the fault code and the fault cause is created, so that the fault code can find the corresponding fault cause through the above-mentioned corresponding relationship. When the fault code is contained in the vehicle information, it means that the fault cause can be determined by querying, and therefore, the vehicle information does not need to be further parsed, so that the fault detection process is more intelligent.

[0128] Figure 4 is a flowchart of a fault detection method of a vehicle according to another embodiment of the disclosure. As shown in Figures 4-5 the method can include:

[0129] S41, the key information related to the fault is parsed from the vehicle information of the vehicle to be detected.

[0130] S42, at least one fault cause is derived according to the key information and the fault knowledge graph for the user to confirm.

[0131] S43, performing sentiment analysis on the confirmation information of the user.

[0132] S44, in the case that it is determined according to the sentiment analysis that none of the at least one fault cause is the fault cause of the vehicle to be detected, prompting manual diagnosis, and updating the result of the manual diagnosis to the fault knowledge graph.

[0133] In the embodiment, when the vehicle information does not contain fault codes, the vehicle information is parsed to obtain key information, at least one corresponding fault cause is found in the fault knowledge graph according to the key information, the fault cause is sequentially inquired from the user, and the confirmation information of the user is received. When the confirmation information of the user is all negative sentiment (i.e. negative sentiment or other sentiment), it means that the fault cause found is not the correct fault cause. Therefore, manual diagnosis is needed. It should be noted that since the fault cause of the vehicle to be detected cannot be determined, it means that there is no such type of fault in the fault knowledge graph. Therefore, the interactive data (i.e. the inquiry data of the maintenance personnel and the reply data of the user) in the process of manual diagnosis is obtained, the inquiry data of the maintenance personnel is parsed to obtain fault keywords through a named entity recognition model, and the reply data is parsed through a sentiment analysis recurrent neural network to obtain a sentiment analysis result. According to the key information (i.e. the fault phenomenon) determined by the user input vehicle information, the above three are combined into a triple and the fault knowledge graph is updated. As can be seen, the fault interactive data obtained through manual diagnosis constantly updates the knowledge graph, so that the vehicle fault detection is more accurate, greatly reduces the dependence on maintenance personnel, and improves the accuracy of fault detection.

[0134] In one embodiment, the vehicle fault detection method further comprises:

[0135] According to the diagnosed fault cause of the vehicle to be detected, a corresponding fault solution is determined and provided to the user.

[0136] Specifically, the corresponding relationship table of the fault cause and the fault solution can be pre-stored in the fault detection system, or the node of the fault solution can be pre-created in the fault knowledge graph and connected after the corresponding fault keyword node. Then, when the fault cause of the vehicle to be detected is determined, the corresponding fault solution is extracted and sent to the user through short message, email or voice broadcast, so that the user can quickly solve the fault according to the solution, thereby improving the user experience.

[0137] Figure 6 A structural block diagram of a vehicle fault detection device according to an embodiment of the present application is shown. As shown in Figure 6 The device can include:

[0138] The parsing module 61 is configured to parse key information related to the fault according to the vehicle information of the vehicle to be detected.

[0139] The processing module 62 is configured to derive at least one fault cause according to the key information and the fault knowledge graph, for user confirmation.

[0140] The analysis module 63 is configured to perform sentiment analysis on the confirmation information of the user.

[0141] The fault detection module 64 is configured to diagnose the fault cause of the vehicle to be detected from the at least one fault cause according to the result of the sentiment analysis.

[0142] In an embodiment, the parsing module 61 comprises:

[0143] The first parsing unit is configured to perform word segmentation and word vector structural processing on the vehicle information of the vehicle to be detected, to obtain fault-related key information.

[0144] In an embodiment, as shown in Figure 7 The device further comprises:

[0145] The basic physical part knowledge graph construction module 71 is configured to construct a basic physical part knowledge graph according to basic physical part information of a plurality of vehicle models.

[0146] The triple construction module 72 is configured to construct triples containing fault phenomena, sentiment analysis results and fault keywords according to historical fault interaction data.

[0147] The fault knowledge graph construction module 73 is configured to associate fault causes in the historical fault interaction data with the basic physical part knowledge graph according to the triples, to obtain a fault knowledge graph.

[0148] In an embodiment, the triple construction module 72 comprises:

[0149] The second parsing unit is configured to parse inquiry data in the historical fault interaction data by a named entity recognition model, to obtain fault keywords.

[0150] The third parsing unit is configured to parse reply data in the historical fault interaction data by a sentiment analysis recurrent neural network, to obtain sentiment analysis results.

[0151] The construction unit is configured to construct triples using the fault keywords, the sentiment analysis results and fault phenomena in the historical fault interaction data.

[0152] In an embodiment, the fault knowledge graph construction module 73 comprises:

[0153] The first matching and associating unit is configured to find matched basic physical part information in the basic physical part knowledge graph according to the fault keyword in the triple, and associate the fault phenomenon in the triple to which the fault keyword belongs with the matched basic physical part information.

[0154] The second matching and associating unit is configured to obtain a fault cause corresponding to the fault phenomenon from the historical fault interaction data, and associate the obtained fault cause with the fault phenomenon to obtain the fault knowledge graph.

[0155] In an embodiment, the analysis module 63 comprises:

[0156] The sentiment analysis unit is configured to perform sentiment analysis on the confirmation information of the user by using a sentiment analysis recurrent neural network, and determine a sentiment type of the confirmation information.

[0157] In an embodiment, the fault detection module 64 comprises:

[0158] The sub-processing unit is configured to, when the user has a positive sentiment for one of the at least one fault cause according to the sentiment analysis, diagnose the fault cause as a fault cause of the vehicle to be detected.

[0159] In an embodiment, the parsing module 61 comprises:

[0160] The judging unit is configured to determine whether the vehicle information of the vehicle to be detected contains a fault code, and parse the key information related to the fault from the vehicle information in a case where it is determined that the vehicle information does not contain the fault code.

[0161] In an embodiment, as shown in Figure 8 The device further comprises:

[0162] The artificial diagnosis module 65 is configured to, in a case where it is determined that none of the at least one fault cause is a fault cause of the vehicle to be detected according to the sentiment analysis, prompt to perform artificial diagnosis, and update a result of the artificial diagnosis to the fault knowledge graph.

[0163] In an embodiment, the device further comprises:

[0164] The scheme determination module is configured to determine a corresponding fault solution according to the diagnosed fault cause of the vehicle to be detected, and provide the fault solution to the user.

[0165] In this way, the device of the embodiment of the application can determine the fault cause of the vehicle to be detected by combining the key information and the fault knowledge graph with the sentiment analysis result of the confirmation information of the user.

[0166] The functions of each module in each device of the embodiments of the application can be referred to the corresponding description in the above method, which will not be described herein again.

[0167] Figure 9 A structural block diagram of a vehicle fault detection system according to an embodiment of the present application is shown. As shown in the figure, the vehicle fault detection system comprises a memory 910 and a processor 920, and the memory 910 stores a computer program executable on the processor 920. The processor 920 implements the vehicle fault detection method in the above embodiment when executing the computer program. The number of the memory 910 and the processor 920 can be one or more. Figure 9

[0168] The vehicle fault detection system further comprises:

[0169] A communication interface 930 for communicating with external devices and transmitting data.

[0170] If the memory 910, the processor 920 and the communication interface 930 are independently implemented, the memory 910, the processor 920 and the communication interface 930 can be connected to each other through a bus and complete communication therebetween. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 9 only one thick line is used in the figure, but it does not mean that there is only one bus or only one type of bus.

[0171] Optionally, in specific implementation, if the memory 910, the processor 920 and the communication interface 930 are integrated on a chip, the memory 910, the processor 920 and the communication interface 930 can complete communication therebetween through an internal interface.

[0172] The embodiment of the present application provides a computer readable storage medium, which stores a computer program, and the program is executed by a processor to implement the method provided in the embodiment of the present application.

[0173] The embodiment of the present application further provides a chip, which comprises a processor, is used for calling and running instructions stored in a memory to enable a communication device installed with the chip to execute the method provided in the embodiment of the present application.

[0174] ​The embodiment of the present application further provides a chip, comprising: an input interface, an output interface, a processor and a memory, the input interface, the output interface, the processor and the memory are connected through internal connection paths, the processor is used for executing the code in the memory, and when the code is executed, the processor is used for executing the method provided by the embodiment of the present application.

[0175] It should be understood that the processor described above can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc. It should be noted that the processor can be a processor supporting an advanced RISC machine (ARM) architecture.

[0176] Further, the aforementioned memory can include a read-only memory, and a random access memory, and can further include a nonvolatile random access memory. The memory can be a volatile memory or a nonvolatile memory, or can include both volatile and nonvolatile memory. The nonvolatile memory can include a read-only memory (ROM), a programmable ROM (PROM), an erasable PROM (EPROM), an electrically EPROM (EEPROM), or a flash memory. The volatile memory can include a random access memory (RAM), which functions as an external cache. Many forms of RAM are available. For example, a static RAM (SRAM), a dynamic RAM (DRAM), a synchronous DRAM (SDRAM), a double data rate SDRAM (DDR SDRAM), an enhanced SDRAM (ESDRAM), a synchlink DRAM (SLDRAM), and a direct rambus RAM (DR RAM) are available.

[0177] In the above-described embodiments, all or a part can be implemented by software, hardware, firmware, or any combination thereof. When implemented as software, it can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed by a computer, all or a part of the procedures or functions according to the present disclosure are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable apparatus. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another computer-readable storage medium.

[0178] In the description of the application, the description of the terms "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" etc. means 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 application. Moreover, the specific features, structures, materials or characteristics described can be combined in any appropriate manner in one or more embodiments or examples. In addition, different embodiments or examples described in the specification and characteristics of different embodiments or examples can be combined and combined by those skilled in the art without contradiction.

[0179] In addition, the terms "first", "second" are only for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include at least one of the features. In the description of the application, the meaning of "multiple" is two or more, unless otherwise specifically limited.

[0180] Any process or method descriptions in flow charts or described elsewhere herein can be understood as representing code modules, segments, or portions of code that include one or more executable instructions for implementing specific logic functions or steps in the process. And the scope of preferred embodiments of the present application includes additional implementation in which the functions are performed in different orders, in substantially simultaneous fashion, or in reverse order.

[0181] The logic and / or steps represented in flow charts or otherwise described herein, for example, can be considered as a sequence of executable instructions, which can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor- based system, or other system that can fetch instructions from a instruction execution system, apparatus, or device and execute the instructions, or in conjunction with which the instructions may be executed.

[0182] It should be understood that parts of the present application can be implemented in hardware, software, firmware or a combination thereof. In the above-described embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. All or part of the steps of the above-described embodiment method can be instructed by a program to complete the relevant hardware, which can be stored in a computer readable storage medium, and the program includes one or a combination of steps of the method embodiments when executed.

[0183] In addition, each of the function units in each embodiment of the present application can be integrated in one processing module, or each unit can be physically present separately, or two or more units can be integrated in one module. The integrated module can be realized in the form of hardware or in the form of a software function module. When the integrated module is realized in the form of a software function module and sold or used as an independent product, it can also be stored in a computer readable storage medium. The storage medium can be a read-only memory, a magnetic disk or an optical disk, etc.

[0184] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited to this. Any person skilled in the art can easily think of various changes or replacements within the technical scope disclosed in the present application, and these should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for detecting vehicle faults, characterized in that, include: Based on the vehicle information uploaded by the user, key information related to the fault is extracted. Based on the key information and the fault knowledge graph, at least one cause of the fault is determined for user confirmation; Perform sentiment analysis on the user's confirmation information; Based on the results of the sentiment analysis, the cause of the malfunction of the vehicle under test is diagnosed from the at least one cause of malfunction. The method further includes: constructing a basic physical component knowledge graph based on the basic physical component information of multiple vehicle models; constructing a triplet containing fault phenomena, sentiment analysis results, and fault keywords based on historical fault interaction data; and associating the fault causes in the historical fault interaction data with the basic physical component knowledge graph based on the triplet to obtain the fault knowledge graph.

2. The method according to claim 1, characterized in that, The process of parsing key fault-related information from the vehicle information of the vehicle to be tested includes: The vehicle information of the vehicle to be detected is processed by word segmentation and word vector structuring to obtain key information related to the fault.

3. The method according to claim 1, characterized in that, The construction of triples containing fault phenomena, sentiment analysis results, and fault keywords based on historical fault interaction data includes: The query data in historical fault interaction data is parsed using a named entity recognition model to obtain fault keywords; The response data in the historical fault interaction data is analyzed by a sentiment analysis recurrent neural network to obtain the sentiment analysis results; Triples are constructed using the fault keywords, the sentiment analysis results, and the fault phenomena in the historical fault interaction data.

4. The method according to claim 1, characterized in that, The step of associating the fault causes in the historical fault interaction data with the basic physical component knowledge graph based on the triples to obtain the fault knowledge graph includes: Based on the fault keywords in the triplet, matching basic physical component information is found in the basic physical component knowledge graph, and the fault phenomenon in the triplet to which the fault keyword belongs is associated with the matching basic physical component information. The fault causes corresponding to the fault phenomena are obtained from the historical fault interaction data, and the obtained fault causes are associated with the fault phenomena to obtain the fault knowledge graph.

5. The method according to claim 1, characterized in that, The sentiment analysis of the user's confirmation information includes: A recurrent neural network for sentiment analysis is used to perform sentiment analysis on the user's confirmation information to determine the sentiment type of the confirmation information.

6. The method according to claim 1, characterized in that, The step of diagnosing the cause of the malfunction of the vehicle under test from the at least one cause of malfunction based on the result of the sentiment analysis includes: When the sentiment analysis determines that the user has a positive sentiment towards one of the at least one causes of failure, that cause of failure is diagnosed as the cause of failure of the vehicle under test.

7. The method according to claim 1, characterized in that, The process of parsing key fault-related information from the vehicle information of the vehicle to be tested includes: Determine whether the vehicle information of the vehicle to be tested contains fault codes. If it is determined that the vehicle information does not contain fault codes, extract the key information related to the fault from the vehicle information.

8. The method according to claim 1, characterized in that, The method further includes: If, based on the sentiment analysis, it is determined that none of the at least one cause of the fault is the cause of the fault in the vehicle under test, a manual diagnosis is prompted, and the result of the manual diagnosis is updated in the fault knowledge graph.

9. The method according to claim 1, characterized in that, The method further includes: Based on the diagnosed cause of the vehicle's malfunction, a corresponding solution is determined and provided to the user.

10. A vehicle fault detection device, characterized in that, include: The parsing module is used to extract key information related to the fault from the vehicle information of the vehicle to be tested uploaded by the user. The processing module is used to determine at least one cause of the fault based on the key information and the fault knowledge graph, for user confirmation; The analysis module is used to perform sentiment analysis on the user's confirmation information; The fault detection module is used to diagnose the cause of the fault in the vehicle under test from among the at least one fault cause, based on the result of the sentiment analysis. The device further includes: a basic physical component knowledge graph construction module, used to construct a basic physical component knowledge graph based on basic physical component information of multiple vehicle models; a triplet construction module, used to construct triplets containing fault phenomena, sentiment analysis results, and fault keywords based on historical fault interaction data; and a fault knowledge graph construction module, used to associate the fault causes in the historical fault interaction data with the basic physical component knowledge graph based on the triplets to obtain the fault knowledge graph.

11. The apparatus according to claim 10, characterized in that, The parsing module includes: The first parsing unit is used to perform word segmentation and word vector structuring on the vehicle information of the vehicle to be detected to obtain key information related to the fault.

12. The apparatus according to claim 10, characterized in that, The triplet construction module includes: The second parsing unit is used to parse the query data in the historical fault interaction data through the named entity recognition model to obtain fault keywords; The third parsing unit is used to parse the response data in the historical fault interaction data through a sentiment analysis recurrent neural network to obtain the sentiment analysis results; The construction unit is used to construct triples using the fault keywords, the sentiment analysis results, and the fault phenomena in the historical fault interaction data.

13. The apparatus according to claim 10, characterized in that, The fault knowledge graph construction module includes: The first matching and association unit is used to associate the fault phenomenon in the triple to which the fault keyword belongs with the basic physical component information of the matching; The second matching and association unit is used to obtain the fault cause corresponding to the fault phenomenon from the historical fault interaction data, and associate the obtained fault cause with the fault phenomenon to obtain the fault knowledge graph.

14. The apparatus according to claim 10, characterized in that, The analysis module includes: The sentiment analysis unit is used to perform sentiment analysis on the user's confirmation information using a sentiment analysis recurrent neural network to determine the sentiment type of the confirmation information.

15. The apparatus according to claim 10, characterized in that, The fault detection module includes: The subprocessing unit is configured to diagnose the fault cause as the fault cause of the vehicle under test when the user has a positive sentiment towards one of the at least one fault causes based on the sentiment analysis.

16. The apparatus according to claim 10, characterized in that, The parsing module includes: The judgment unit is used to determine whether the vehicle information of the vehicle to be tested contains a fault code. If it is determined that the vehicle information does not contain a fault code, the unit parses out the key information related to the fault from the vehicle information.

17. The apparatus according to claim 10, characterized in that, The device further includes: The manual diagnosis module is used to prompt for manual diagnosis when the sentiment analysis determines that none of the at least one fault cause is the fault cause of the vehicle under test, and to update the result of the manual diagnosis to the fault knowledge graph.

18. The apparatus according to claim 10, characterized in that, The device further includes: The solution determination module is used to determine the corresponding fault solution based on the diagnosed fault cause of the vehicle under test and provide it to the user.

19. A vehicle fault detection system, characterized in that, include: A processor and a memory, wherein instructions are stored in the memory and loaded and executed by the processor to implement the method as claimed in any one of claims 1 to 9.

20. A computer-readable storage medium storing a computer program that, when executed by a processor, implements the method as described in any one of claims 1-9.

Images