Vehicle fault analysis method, device and system

Abstract

The invention discloses a vehicle fault analysis method, device and system, and the method comprises the following steps: extracting a candidate document set from a preset expert knowledge base according to a fault query text; according to the fault query text, extracting a final document set from the candidate document set; and outputting the final document set as a vehicle fault analysis result. According to the method, the retrieval of the expert knowledge base is divided into a primary screening stage and a fine screening stage, and the efficiency and the accuracy of the retrieval are considered through staged screening.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle fault analysis, and in particular to a vehicle fault analysis method, device and system. Background Art

[0002] With the continuous expansion of modern automobile functions and the increasing complexity of electronic control units, the diversity and complexity of vehicle faults are also increasing. In this context, how to efficiently use fault phenomenon description information and quickly and accurately retrieve relevant fault causes and maintenance strategies from the fault database is crucial to assist maintenance technicians in quickly analyzing vehicle fault problems and improving their work efficiency. The industry has been working hard to explore and improve this field in order to achieve more efficient fault diagnosis.

[0003] The existing technology uses a single retrieval algorithm to retrieve fault-related information in a knowledge base, which often has the problem of low retrieval efficiency or inaccurate retrieval results. Summary of the invention

[0004] The object of the present invention is to provide a vehicle fault analysis method, device and system to improve the efficiency and accuracy of retrieval of fault information.

[0005] In order to solve the above technical problems, the present invention provides a vehicle fault analysis method, comprising:

[0006] According to the fault query text, a candidate document set is extracted from the preset expert knowledge base;

[0007] According to the fault query text, extract the final document set from the candidate document set;

[0008] The final document set is output as the vehicle fault analysis result.

[0009] According to the above scheme, the expert knowledge base includes a number of documents, including interrelated vehicle model description texts, fault assembly part description texts, fault phenomenon description texts, fault cause description texts, and repair method case description texts;

[0010] The method for extracting a candidate document set from a preset expert knowledge base according to the fault query text includes:

[0011] Preprocess the fault query text and each fault phenomenon description text in the expert knowledge base to generate a fault query text word segmentation list and multiple fault phenomenon description text word segmentation lists respectively;

[0012] Calculate the TF value of each term in the fault query text for each fault phenomenon description text segmentation list;

[0013] Calculate the IDF value of each term in the fault query text for the entire fault phenomenon description text;

[0014] According to the TF value and the IDF value, determine the first relevance score of each term in the fault query text and each fault phenomenon description text;

[0015] Determine the similarity score between the fault query text and each fault phenomenon description text according to the correlation between each term in the fault query text and each fault phenomenon description text;

[0016] The documents corresponding to several fault phenomenon description texts with the highest similarity scores are taken as candidate document sets.

[0017] According to the above scheme, the method for extracting the final document set from the candidate document set according to the fault query text includes:

[0018] The fault query text is concatenated with each fault phenomenon description text in the candidate document set to obtain a plurality of concatenated texts;

[0019] Input each concatenated text into a preset model to obtain a second relevance score of each concatenated text;

[0020] The documents corresponding to the several concatenated texts with the second highest relevance scores are taken as the final document set.

[0021] The present invention also provides a vehicle fault analysis device, comprising:

[0022] The first screening module is used to extract a candidate document set from a preset expert knowledge base according to the fault query text;

[0023] The second screening module is used to extract a final document set from the candidate document set according to the fault query text;

[0024] The output module is used to output the final document set as the vehicle fault analysis result.

[0025] According to the above scheme, the expert knowledge base includes a number of documents, including interrelated vehicle model description texts, fault assembly part description texts, fault phenomenon description texts, fault cause description texts, and repair method case description texts;

[0026] The method for extracting a candidate document set from a preset expert knowledge base according to the fault query text includes:

[0027] Preprocess the fault query text and each fault phenomenon description text in the expert knowledge base to generate a fault query text word segmentation list and multiple fault phenomenon description text word segmentation lists respectively;

[0028] Calculate the TF value of each term in the fault query text for each fault phenomenon description text segmentation list;

[0029] Calculate the IDF value of each term in the fault query text for the entire fault phenomenon description text;

[0030] According to the TF value and the IDF value, determine the first relevance score of each term in the fault query text and each fault phenomenon description text;

[0031] Determine the similarity score between the fault query text and each fault phenomenon description text according to the correlation between each term in the fault query text and each fault phenomenon description text;

[0032] The documents corresponding to several fault phenomenon description texts with the highest similarity scores are taken as candidate document sets.

[0033] According to the above scheme, the method for extracting the final document set from the candidate document set according to the fault query text includes:

[0034] The fault query text is concatenated with each fault phenomenon description text in the candidate document set to obtain a plurality of concatenated texts;

[0035] Input each concatenated text into a preset model to obtain a second relevance score of each concatenated text;

[0036] The documents corresponding to the several concatenated texts with the second highest relevance scores are taken as the final document set.

[0037] The present invention also provides a vehicle fault analysis system, comprising:

[0038] Vehicle fault analysis device;

[0039] Input module, used to input fault query text;

[0040] A display module, used for displaying the final document set output by the vehicle fault analysis device;

[0041] The vehicle fault analysis device includes:

[0042] The first screening module is used to extract a candidate document set from a preset expert knowledge base according to the fault query text;

[0043] The second screening module is used to extract a final document set from the candidate document set according to the fault query text;

[0044] The output module is used to output the final document set as the vehicle fault analysis result.

[0045] According to the above scheme, the expert knowledge base includes a number of documents, including interrelated vehicle model description texts, fault assembly part description texts, fault phenomenon description texts, fault cause description texts, and repair method case description texts;

[0046] The method for extracting a candidate document set from a preset expert knowledge base according to the fault query text includes:

[0047] Preprocess the fault query text and each fault phenomenon description text in the expert knowledge base to generate a fault query text word segmentation list and multiple fault phenomenon description text word segmentation lists respectively;

[0048] Calculate the TF value of each term in the fault query text for each fault phenomenon description text segmentation list;

[0049] Calculate the IDF value of each term in the fault query text for the entire fault phenomenon description text;

[0050] According to the TF value and the IDF value, determine the first relevance score of each term in the fault query text and each fault phenomenon description text;

[0051] Determine the similarity score between the fault query text and each fault phenomenon description text according to the correlation between each term in the fault query text and each fault phenomenon description text;

[0052] The documents corresponding to several fault phenomenon description texts with the highest similarity scores are taken as candidate document sets.

[0053] According to the above scheme, the method for extracting the final document set from the candidate document set according to the fault query text includes:

[0054] The fault query text is concatenated with each fault phenomenon description text in the candidate document set to obtain a plurality of concatenated texts;

[0055] Input each concatenated text into a preset model to obtain a second relevance score of each concatenated text;

[0056] The documents corresponding to the several concatenated texts with the second highest relevance scores are taken as the final document set.

[0057] The present invention also provides a car provided with the vehicle fault analysis system as described above.

[0058] The beneficial effects of the present invention are as follows: the present method divides the overall retrieval process into a preliminary screening stage for extracting a candidate document set and a fine screening stage for extracting a final document set, taking into account both the efficiency and accuracy of the retrieval process. The preliminary screening stage can quickly narrow the search scope and improve the retrieval efficiency, and the fine screening stage can further refine the screening and improve the retrieval accuracy. The method helps testers to quickly locate the cause of a fault and solve the fault during the vehicle production test stage, and can also help car owners quickly troubleshoot the fault during the actual use of the vehicle, thereby improving the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 , Figure 3 is a flow chart of a vehicle fault analysis method according to a first embodiment of the present invention;

[0060] Figure 2 It is a schematic diagram of the preset model structure of the first embodiment of the present invention. DETAILED DESCRIPTION

[0061] In order to make the purpose, technical solution and advantages of the embodiments of the present disclosure clearer, the technical solution of the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings of the embodiments of the present disclosure. Obviously, the described embodiments are part of the embodiments of the present disclosure, not all of the embodiments. Based on the described embodiments of the present disclosure, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present disclosure.

[0062] Embodiment 1:

[0063] This embodiment discloses a vehicle fault analysis method, including:

[0064] S1. Extract candidate document sets from a preset expert knowledge base based on the fault query text;

[0065] S2. Extracting a final document set from the candidate document set according to the fault query text;

[0066] S3. Output the final document set as the vehicle fault analysis result.

[0067] In step S1, the expert knowledge base includes a number of documents, each of which includes data in the following five dimensions: vehicle model description text, faulty assembly part description text, fault phenomenon description text, fault cause description text, and repair method case description text.

[0068] Step S1 is the initial screening stage in this method. In this stage, the query text q is matched with the fault phenomenon description text in the expert knowledge base using the BM25 similarity retrieval model. This stage specifically includes the following steps:

[0069] The method of extracting a candidate document set from a preset expert knowledge base according to the fault query text in S101 includes:

[0070] S102, pre-processing the fault query text and each fault phenomenon description text in the expert knowledge base, and generating a fault query text word segmentation list and multiple fault phenomenon description text word segmentation lists respectively;

[0071] Specifically, preprocessing includes text segmentation and removal of stop words;

[0072] S103, calculating the TF value of each term in the fault query text for each fault phenomenon description text segmentation list;

[0073] The TF value indicates the number of times the term in the selected fault query text appears in the selected fault phenomenon description text segmentation list;

[0074] S104, calculating the IDF value of each term in the fault query text with respect to the entire fault phenomenon description text;

[0075] The IDF value is expressed as IDF(qi), and its calculation formula is as follows:

[0076]

[0077] In the above formula, qi represents the selected term in the fault query text, N represents the total number of fault phenomenon description texts in the expert knowledge base, and w represents the number of fault phenomenon description texts in which qi appears in the expert knowledge base;

[0078] S105, determining a first relevance score between each term of the fault query text and each fault phenomenon description text according to the TF value and the IDF value;

[0079] The first relevance score is denoted as R(qi,f), and its calculation formula is as follows:

[0080]

[0081] In the above formula, f represents the selected fault phenomenon description text, k1 and b are adjustment factors. In this embodiment, k1=2, b=0.75, l is the length of f, and avg_l represents the average length of all fault phenomenon description texts in the expert knowledge base;

[0082] S106, determining a similarity score between the fault query text and each fault phenomenon description text according to the correlation between each term in the fault query text and each fault phenomenon description text; the similarity score may be expressed as Q(q,f);

[0083] Specifically, for a fault query text and a selected fault phenomenon description text, the similarity score between the two is calculated as follows: the first correlation score of each term in the fault query text and the fault phenomenon description text is added up, and the obtained value is the similarity score between the fault query text and the fault phenomenon description text;

[0084] S107, taking the documents corresponding to the several fault phenomenon description texts with the highest similarity scores as a candidate document set;

[0085] Specifically, all fault phenomenon description texts are sorted according to the similarity scores, the first K fault phenomenon description texts with the highest similarity scores are selected, and then the documents corresponding to these K fault phenomenon description texts are collected as the candidate document collection in the initial screening stage.

[0086] Step S2 specifically includes the following steps:

[0087] S201, respectively concatenating the fault query text with each fault phenomenon description text in the candidate document set to obtain a plurality of concatenated texts;

[0088] Specifically, the [sep] character is used for concatenation, and the concatenation process is as follows:

[0089] q + =concat(q,[sep],d i )

[0090] In the above formula, q + Indicates concatenated text, d i Indicates the fault phenomenon description text selected from the candidate document set;

[0091] In this embodiment, a cross encoder is used for splicing;

[0092] S202, inputting each concatenated text into a preset model to obtain a second relevance score of each concatenated text;

[0093] Specifically, the preset model includes a trained BERT model and a classifier. After the concatenated text is input into the BERT model, the [cls] vector output by BERT is first obtained. Then, the classifier obtains the second relevance score according to the [cls] vector. The second relevance score is expressed as score i , and its calculation process is expressed as follows:

[0094] E i =bert(q + )

[0095] score i =classifier(E i[cls] )

[0096] Among them, E i Represents the last hidden layer vector after the concatenated text is input into the BERT model, E i [cls] Represents the last hidden layer vector of the [cls] character, and classifier represents the classifier.

[0097] S202: Take the documents corresponding to the concatenated texts with the highest second relevance scores as the final document set.

[0098] In step S3, each document in the final document set is output in the order of the second relevance score.

[0099] Embodiment 2:

[0100] This embodiment provides a vehicle fault analysis device, including:

[0101] The first screening module is used to extract a candidate document set from a preset expert knowledge base according to the fault query text;

[0102] The second screening module is used to extract a final document set from the candidate document set according to the fault query text;

[0103] The output module is used to output the final document set as the vehicle fault analysis result.

[0104] The first screening module specifically performs the following steps:

[0105] Preprocess the fault query text and each fault phenomenon description text in the expert knowledge base to generate a fault query text word segmentation list and multiple fault phenomenon description text word segmentation lists respectively;

[0106] Calculate the TF value of each term in the fault query text for each fault phenomenon description text segmentation list;

[0107] Calculate the IDF value of each term in the fault query text for the entire fault phenomenon description text;

[0108] According to the TF value and the IDF value, determine the first relevance score of each term in the fault query text and each fault phenomenon description text;

[0109] Determine the similarity score between the fault query text and each fault phenomenon description text according to the correlation between each term in the fault query text and each fault phenomenon description text;

[0110] The documents corresponding to several fault phenomenon description texts with the highest similarity scores are taken as candidate document sets.

[0111] The second screening module specifically performs the following steps:

[0112] The fault query text is concatenated with each fault phenomenon description text in the candidate document set to obtain a plurality of concatenated texts;

[0113] Input each concatenated text into a preset model to obtain a second relevance score of each concatenated text;

[0114] The documents corresponding to the several concatenated texts with the second highest relevance scores are taken as the final document set.

[0115] Embodiment three:

[0116] This embodiment provides a vehicle fault analysis system, including:

[0117] Vehicle fault analysis device;

[0118] Input module, used to input fault query text;

[0119] The display module is used to display the final document set output by the vehicle fault analysis device.

[0120] The vehicle fault analysis device includes:

[0121] The first screening module is used to extract a candidate document set from a preset expert knowledge base according to the fault query text;

[0122] The second screening module is used to extract a final document set from the candidate document set according to the fault query text;

[0123] The output module is used to output the final document set as the vehicle fault analysis result.

[0124] The first screening module specifically performs the following steps:

[0125] Preprocess the fault query text and each fault phenomenon description text in the expert knowledge base to generate a fault query text word segmentation list and multiple fault phenomenon description text word segmentation lists respectively;

[0126] Calculate the TF value of each term in the fault query text for each fault phenomenon description text segmentation list;

[0127] Calculate the IDF value of each term in the fault query text for the entire fault phenomenon description text;

[0128] According to the TF value and the IDF value, determine the first relevance score of each term in the fault query text and each fault phenomenon description text;

[0129] Determine the similarity score between the fault query text and each fault phenomenon description text according to the correlation between each term in the fault query text and each fault phenomenon description text;

[0130] The documents corresponding to several fault phenomenon description texts with the highest similarity scores are taken as candidate document sets.

[0131] The second screening module specifically performs the following steps:

[0132] The fault query text is concatenated with each fault phenomenon description text in the candidate document set to obtain a plurality of concatenated texts;

[0133] Input each concatenated text into a preset model to obtain a second relevance score of each concatenated text;

[0134] The documents corresponding to the several concatenated texts with the second highest relevance scores are taken as the final document set.

[0135] Embodiment 4:

[0136] This embodiment provides a car, which is equipped with the vehicle fault analysis system described in the third embodiment.

[0137] Embodiment five:

[0138] This embodiment provides a schematic diagram of the structure of a computer device, such as a smart phone, tablet computer, laptop computer, desktop computer, rack server, blade server, tower server or cabinet server (including an independent server or a server cluster composed of multiple servers) that can execute programs. The computer device of this embodiment includes at least but is not limited to: a memory and a processor that can communicate with each other through a system bus.

[0139] In this embodiment, the memory (i.e., readable storage medium) includes flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM). The memory may also be an external storage device of a computer device, such as a plug-in hard disk equipped on the computer device, a smart memory card (SmartMedia Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. Of course, the memory may also include both the internal storage unit of the computer device and its external storage device. In this embodiment, the memory is generally used to store the operating system and various application software installed on the computer device, such as the program code of the vehicle fault analysis method in the first embodiment. In addition, the memory may also be used to temporarily store various types of data that have been output or are to be output.

[0140] The processor may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip in some embodiments. The processor is generally used to control the overall operation of a computer device. In this embodiment, the processor is used to run the program code stored in the memory or process data, such as running the program code of the vehicle fault analysis method, to implement the vehicle fault analysis method in Embodiment 1.

[0141] Embodiment six:

[0142] The present application also provides a computer-readable storage medium, such as a flash memory, a hard disk, a multimedia card, a card-type memory (e.g., an SD or DX memory, etc.), a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a disk, an optical disk, a server, an App application store, etc., on which a computer program is stored, and the program realizes the corresponding function when executed by the processor. The computer-readable storage medium of this embodiment is used to store the program code of the vehicle fault analysis method in Embodiment 1, and realizes the vehicle fault analysis method in Embodiment 1 when executed by the processor.

[0143] It should be pointed out that, according to the needs of implementation, the various steps / components described in this application can be split into more steps / components, and two or more steps / components or partial operations of steps / components can be combined into new steps / components to achieve the purpose of the present invention.

[0144] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0145] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A vehicle fault analysis method, characterized in that: include: According to the fault query text, a candidate document set is extracted from the preset expert knowledge base; According to the fault query text, a final document set is extracted from the candidate document set; The final document set is output as the vehicle fault analysis result.

2. The vehicle fault analysis method according to claim 1, characterized in that: The expert knowledge base includes a number of documents, including interrelated vehicle model description texts, fault assembly part description texts, fault phenomenon description texts, fault cause description texts, and repair method case description texts; The method for extracting a candidate document set from a preset expert knowledge base according to the fault query text includes: Preprocess the fault query text and each fault phenomenon description text in the expert knowledge base to generate a fault query text word segmentation list and multiple fault phenomenon description text word segmentation lists respectively; Calculate the TF value of each term in the fault query text for each fault phenomenon description text segmentation list; Calculate the IDF value of each term in the fault query text for the entire fault phenomenon description text; According to the TF value and the IDF value, determine the first relevance score of each term in the fault query text and each fault phenomenon description text; Determine the similarity score between the fault query text and each fault phenomenon description text according to the correlation between each term in the fault query text and each fault phenomenon description text; The documents corresponding to several fault phenomenon description texts with the highest similarity scores are taken as candidate document sets.

3. The vehicle fault analysis method according to claim 2, characterized in that: The method for extracting a final document set from a candidate document set according to the fault query text comprises: The fault query text is concatenated with each fault phenomenon description text in the candidate document set to obtain a plurality of concatenated texts; Input each concatenated text into a preset model to obtain a second relevance score for each concatenated text; The documents corresponding to the several concatenated texts with the second highest relevance scores are taken as the final document set.

4. A vehicle fault analysis device, characterized in that: include: The first screening module is used to extract a candidate document set from a preset expert knowledge base according to the fault query text; The second screening module is used to extract a final document set from the candidate document set according to the fault query text; The output module is used to output the final document set as the vehicle fault analysis result.

5. The vehicle failure analysis device according to claim 4, characterized in that: The expert knowledge base includes a number of documents, including interrelated vehicle model description texts, fault assembly part description texts, fault phenomenon description texts, fault cause description texts, and repair method case description texts; The method for extracting a candidate document set from a preset expert knowledge base according to the fault query text includes: Preprocess the fault query text and each fault phenomenon description text in the expert knowledge base to generate a fault query text word segmentation list and multiple fault phenomenon description text word segmentation lists respectively; Calculate the TF value of each term in the fault query text for each fault phenomenon description text segmentation list; Calculate the IDF value of each term in the fault query text for the entire fault phenomenon description text; According to the TF value and the IDF value, determine the first relevance score of each term in the fault query text and each fault phenomenon description text; Determine the similarity score between the fault query text and each fault phenomenon description text according to the correlation between each term in the fault query text and each fault phenomenon description text; The documents corresponding to several fault phenomenon description texts with the highest similarity scores are taken as candidate document sets.

6. The vehicle failure analysis device according to claim 5, characterized in that: The method for extracting a final document set from a candidate document set according to the fault query text comprises: The fault query text is concatenated with each fault phenomenon description text in the candidate document set to obtain a plurality of concatenated texts; Input each concatenated text into a preset model to obtain a second relevance score for each concatenated text; The documents corresponding to the several concatenated texts with the second highest relevance scores are taken as the final document set.

7. A vehicle fault analysis system, characterized in that: include: Vehicle fault analysis device; Input module, used to input fault query text; A display module, used for displaying the final document set output by the vehicle fault analysis device; The vehicle fault analysis device includes: The first screening module is used to extract a candidate document set from a preset expert knowledge base according to the fault query text; The second screening module is used to extract a final document set from the candidate document set according to the fault query text; The output module is used to output the final document set as the vehicle fault analysis result.

8. The vehicle fault analysis system according to claim 7, characterized in that: The expert knowledge base includes a number of documents, including interrelated vehicle model description texts, fault assembly part description texts, fault phenomenon description texts, fault cause description texts, and repair method case description texts; The method for extracting a candidate document set from a preset expert knowledge base according to the fault query text includes: Preprocess the fault query text and each fault phenomenon description text in the expert knowledge base to generate a fault query text word segmentation list and multiple fault phenomenon description text word segmentation lists respectively; Calculate the TF value of each term in the fault query text for each fault phenomenon description text segmentation list; Calculate the IDF value of each term in the fault query text for the entire fault phenomenon description text; According to the TF value and the IDF value, determine the first relevance score of each term in the fault query text and each fault phenomenon description text; Determine the similarity score between the fault query text and each fault phenomenon description text according to the correlation between each term in the fault query text and each fault phenomenon description text; The documents corresponding to several fault phenomenon description texts with the highest similarity scores are taken as candidate document sets.

9. The vehicle fault analysis system according to claim 8, characterized in that: The method for extracting a final document set from a candidate document set according to the fault query text comprises: The fault query text is concatenated with each fault phenomenon description text in the candidate document set to obtain a plurality of concatenated texts; Input each concatenated text into a preset model to obtain a second relevance score for each concatenated text; The documents corresponding to the several concatenated texts with the second highest relevance scores are taken as the final document set.

10. An automobile, characterized in that: A vehicle fault analysis system as described in any one of claims 7 to 9 is provided.

Images