Fault diagnosis method

Abstract

The invention relates to the technical field of automobiles, in particular to a fault diagnosis method. Receiving target vehicle data sent by the target vehicle and fault code information corresponding to the target vehicle data; the fault code information is generated after the target vehicle detects target vehicle data; determining a standby fault type corresponding to the target vehicle based on the fault code information; identifying the target vehicle data, and determining a target fault type corresponding to the target vehicle based on a preset fault diagnosis model; detecting whether the target fault type is consistent with the standby fault type; generating a control instruction based on the detection result and the target fault type, and sending the control instruction to the target vehicle; wherein the control instruction comprises a preset operation strategy corresponding to the target fault type. Therefore, double fault diagnosis of the target vehicle is realized, the accuracy of fault diagnosis of the target vehicle is ensured, and the accuracy of the generated control instruction is ensured.

Description

Technical Field

[0001] The invention relates to the technical field of automobiles, and in particular to a fault diagnosis method. Background Art

[0002] Remote diagnosis and maintenance of new energy vehicles is an intelligent service system based on modern communication technology and the Internet of Things. It uses on-board sensors, data transmission and cloud platform analysis to achieve real-time monitoring of the operating status of new energy vehicles, fault warning and remote maintenance, greatly improving the safety of vehicles and user experience. The system usually includes multiple links such as data acquisition, transmission, storage, analysis and remote control. The data acquisition module is responsible for collecting key information such as the battery status, motor performance, charging status of the vehicle, and sending these data to the cloud server through wireless transmission technology. The server stores and analyzes the data. Once an abnormality is found, the system will immediately issue an early warning and can perform fault troubleshooting and some maintenance operations through remote control. The advantage of remote diagnosis and maintenance of new energy vehicles is that it can detect and deal with potential problems in a timely manner, reduce vehicle failure rates, and reduce maintenance costs. At the same time, it provides convenient online services for car owners, such as appointment maintenance and online consultation, which effectively improves the quality of after-sales service.

[0003] However, existing fault diagnosis methods often rely on a single algorithm, which is difficult to deal with complex and changeable fault conditions, and the diagnostic accuracy needs to be improved. At the same time, the existing remote diagnosis and maintenance system has deficiencies in module integration, user interaction and system management, which affects the overall performance of the system and user experience.

[0004] Therefore, how to diagnose faults of new energy vehicles has become an urgent problem to be solved. Summary of the invention

[0005] In view of this, the present invention provides a fault diagnosis method to solve the problem of how to perform fault diagnosis on new energy vehicles.

[0006] In a first aspect, the present invention provides a fault diagnosis method, applied to a server, the method comprising:

[0007] Receiving target vehicle data and fault code information corresponding to the target vehicle data sent by the target vehicle; the fault code information is generated after the target vehicle detects the target vehicle data;

[0008] Based on the fault code information, determine the alternate fault type corresponding to the target vehicle;

[0009] Identify the target vehicle data and determine the target fault type corresponding to the target vehicle based on a preset fault diagnosis model;

[0010] Check whether the target fault type is consistent with the backup fault type;

[0011] Based on the detection result and the target fault type, a control instruction is generated and sent to the target vehicle; wherein the control instruction includes a preset operation strategy corresponding to the target fault type.

[0012] The fault diagnosis method provided by the embodiment of the present application receives the target vehicle data sent by the target vehicle and the fault code information corresponding to the target vehicle data. Then, based on the fault code information, the standby fault type corresponding to the target vehicle is determined, and the accuracy of the determined standby fault type is ensured. Then, the target vehicle data is identified, and the target fault type corresponding to the target vehicle is determined based on the preset fault diagnosis model, and the accuracy of the target fault type corresponding to the determined target vehicle is ensured. Then, whether the target fault type is consistent with the standby fault type is detected; based on the detection result and the target fault type, a control instruction is generated, and the control instruction is sent to the target vehicle, thereby ensuring the accuracy of the generated control instruction, so that the target vehicle can receive the control instruction and then execute the preset operation strategy. In addition, the above method first receives the fault code information generated by the target vehicle after detecting the target vehicle data, and identifies the target vehicle data, and determines the target fault type corresponding to the target vehicle based on the preset fault diagnosis model, thereby realizing dual fault diagnosis of the target vehicle, ensuring the accuracy of fault diagnosis of the target vehicle, and then ensuring the accuracy of the generated control instruction.

[0013] In an optional implementation, the preset fault diagnosis model includes a fault determination model and a fault type identification model; identifying the target vehicle data, and determining the target fault type corresponding to the target vehicle based on the preset fault diagnosis model, including:

[0014] Inputting the target vehicle data into a fault determination model to determine whether the target vehicle has a fault;

[0015] If a target vehicle breaks down, the target vehicle data is input into a fault type identification model to determine a target fault type corresponding to the target vehicle.

[0016] The fault diagnosis method provided in the embodiment of the present application inputs the target vehicle data into the fault determination model to determine whether the target vehicle has a fault, thereby ensuring the accuracy of the result of determining whether the target vehicle has a fault. If the target vehicle has a fault, the target vehicle data is input into the fault type identification model to determine the target fault type corresponding to the target vehicle, thereby ensuring the accuracy of the target fault type corresponding to the determined target vehicle.

[0017] In an optional implementation, the fault type identification model includes a target fault tree network structure. If a target vehicle fails, the target vehicle data is input into the fault type identification model to determine the target fault type corresponding to the target vehicle, including:

[0018] Traverse each top event in the target fault tree network structure;

[0019] Detect the correlation between the intermediate events and bottom events corresponding to each top event and the target vehicle data;

[0020] Based on the correlation, determining a target intermediate event and / or a target bottom event;

[0021] Obtaining the logical relationship between each target intermediate event and / or each target bottom event;

[0022] According to the logical relationship, determine the preliminary fault type corresponding to the target vehicle;

[0023] Based on the preliminary fault type, extracting fault data related to the preliminary fault type from the target vehicle data;

[0024] Extract features from fault data and generate target fault features;

[0025] matching the target fault signature with predetermined fault signatures in a predefined fault mode library;

[0026] According to the matching results, the target fault type corresponding to the target vehicle is determined.

[0027] The fault diagnosis method provided by the embodiment of the present application traverses each top event in the target fault tree network structure; detects the correlation between the intermediate events and bottom events corresponding to each top event and the target vehicle data, and ensures the accuracy of the obtained correlation. Based on the correlation, the target intermediate event and / or the target bottom event are determined, and the accuracy of the determined target intermediate event and / or the target bottom event is ensured. Then, the logical relationship between each target intermediate event and / or each target bottom event is obtained; according to the logical relationship, the preliminary fault type corresponding to the target vehicle is determined, and the accuracy of the preliminary fault type corresponding to the determined target vehicle is ensured. Then, based on the preliminary fault type, fault data related to the preliminary fault type is extracted from the target vehicle data; feature extraction is performed on the fault data to generate target fault features; the target fault features are matched with the predetermined fault features in the predefined fault mode library, and the target fault type corresponding to the target vehicle is determined according to the matching result. Further refinement of fault diagnosis is achieved, and the accuracy of diagnosis is improved by comparison at the feature level, and the predefined fault mode library can be constructed based on a large amount of historical data and experience, which enhances the reliability of diagnosis, thereby ensuring the accuracy of the target fault type corresponding to the determined target vehicle.

[0028] In an optional implementation, the preset fault diagnosis model includes a fault determination model and a fault type identification model, and the training process of the preset fault diagnosis model includes:

[0029] Obtain historical fault data and historical normal data;

[0030] Generate fault training data based on fault case data in historical fault data;

[0031] Generate normal training data based on historical normal data;

[0032] Generate a first training data set based on the fault training data and the normal training data;

[0033] Based on the first training data set, training an initial fault determination network to generate a fault determination model;

[0034] Based on historical fault data, determine the top event, the middle event and the bottom event, and use logic gates to connect the top event, the middle event and the bottom event to form a fault tree;

[0035] Based on the historical fault data, a second training data set is generated, where the second training data set includes the fault type corresponding to each historical fault data;

[0036] Based on the fault tree and the second training data set, an initial fault type identification network is trained to generate a fault type identification model.

[0037] The fault diagnosis method provided in the embodiment of the present application obtains historical fault data and historical normal data; generates fault training data based on the fault case data in the historical fault data; generates normal training data based on the historical normal data; generates a first training data set based on the fault training data and the normal training data, thereby ensuring the accuracy of the generated first training data set. Based on the first training data set, the initial fault determination network is trained to generate a fault determination model, thereby ensuring the accuracy of the generated fault determination model. Then. Based on the historical fault data, the top event, the middle event, and the bottom event are determined, and the top event, the middle event, and the bottom event are connected using logic gates to form a fault tree, thereby ensuring the accuracy of the generated fault tree. Based on the historical fault data, a second training data set is generated, and based on the fault tree and the second training data set, the initial fault type recognition network is trained to generate a fault type recognition model, thereby ensuring the accuracy of the fault type recognition model obtained through training.

[0038] In an optional implementation, training an initial fault determination network based on the first training data set to generate a fault determination model includes:

[0039] Select the radial basis kernel function;

[0040] Constructing optimization problems;

[0041] Using the Lagrange multiplier method, the optimization problem is transformed to obtain the Lagrange function;

[0042] Solve the Lagrangian function through KKT conditions and obtain the solution;

[0043] Based on the solution results, an initial fault determination network is generated;

[0044] Based on the first training data set, cross-validate the initial fault determination network to determine the model performance corresponding to the initial fault determination network;

[0045] The parameters of the initial fault determination network are adjusted according to the model performance to obtain a fault determination model.

[0046] The fault diagnosis method provided in the embodiment of the present application selects a radial basis kernel function; constructs an optimization problem; uses the Lagrangian multiplier method to deform the optimization problem to obtain a Lagrangian function, thereby ensuring the accuracy of the obtained Lagrangian function. Then, the Lagrangian function is solved by the KKT condition to obtain a solution result, thereby ensuring the accuracy of the obtained solution result. Based on the solution result, an initial fault determination network is generated, thereby ensuring the accuracy of the generated initial fault determination network. Then, based on the first training data set, the initial fault determination network is cross-validated to determine the model performance corresponding to the initial fault determination network, thereby ensuring the accuracy of the model performance corresponding to the determined initial fault determination network. The parameters of the initial fault determination network are adjusted according to the model performance to obtain a fault determination model, thereby ensuring the accuracy of the obtained fault determination model.

[0047] In an optional implementation, based on the fault tree and the second training data set, training an initial fault type identification network to generate a fault type identification model includes:

[0048] According to the hierarchical structure and logical relationship of the fault tree, an initial fault type identification network adapted to the fault tree is generated; the initial fault type identification network is an initial fault tree network structure;

[0049] According to the logic gates in the fault tree, a logic gate simulation unit is generated in the initial fault type identification network;

[0050] The second training data is input into the initial fault type identification network, and the initial fault type identification network is trained based on a preset training algorithm to generate a fault type identification model; the fault type identification model includes a target fault tree network structure.

[0051] The fault diagnosis method provided in the embodiment of the present application generates an initial fault type identification network adapted to the fault tree according to the hierarchical structure and logical relationship of the fault tree, thereby ensuring the accuracy of the generated initial fault type identification network. Then, according to the logic gates in the fault tree, a logic gate simulation unit is generated in the initial fault type identification network; the second training data is input into the initial fault type identification network, and the initial fault type identification network is trained based on a preset training algorithm to generate a fault type identification model, thereby ensuring the accuracy of the generated fault type identification model.

[0052] In an optional implementation, receiving target vehicle data sent by the target vehicle and fault code information corresponding to the target vehicle data includes:

[0053] Receive the encrypted data packet sent by the target vehicle;

[0054] Decrypt the encrypted data packet to obtain target vehicle data and fault code information corresponding to the target vehicle data;

[0055] Verify the target vehicle data and fault code information;

[0056] If the target vehicle data and the fault code information are successfully verified, the various sub-target vehicle data included in the target vehicle data are identified;

[0057] According to the recognition results corresponding to the various sub-target vehicle data, the target database corresponding to the various sub-target vehicle data is determined;

[0058] The data of each sub-target vehicle is stored in the corresponding target database, and indexes are established for the data of each sub-target vehicle in the target database.

[0059] The fault diagnosis method provided in the embodiment of the present application receives an encrypted data packet sent by the target vehicle; decrypts the encrypted data packet to obtain the target vehicle data and the fault code information corresponding to the target vehicle data, thereby ensuring the accuracy and completeness of the acquired target vehicle data and the fault code information corresponding to the target vehicle data. The target vehicle data and the fault code information are verified; if the target vehicle data and the fault code information are successfully verified, the various sub-target vehicle data included in the target vehicle data are identified; based on the identification results corresponding to the various sub-target vehicle data, the target database corresponding to the various sub-target vehicle data is determined, thereby ensuring the accuracy of the target database corresponding to the various sub-target vehicle data. The various sub-target vehicle data are stored in the corresponding target database, and the various sub-target vehicle data are indexed in the target database, thereby ensuring the accuracy of the indexing of the various sub-target vehicle data in the target database, thereby optimizing the query efficiency.

[0060] In an optional implementation, generating a control instruction based on the detection result and the target fault type includes:

[0061] If the detection result is that the target fault type is consistent with the backup fault type, the target fault type is displayed to the user in a graphical interface, and a preset operation strategy corresponding to the target fault type is generated, and a control instruction is generated based on the result that the target fault type is consistent with the backup fault type based on the operation strategy;

[0062] If the detection result is that the target fault type is inconsistent with the backup fault type, the target fault type is displayed to the user in a graphical interface, and a preset operation strategy corresponding to the target fault type is generated; based on the operation strategy and the result of the inconsistency between the target fault type and the backup fault type, a control instruction is generated.

[0063] The fault diagnosis method provided in the embodiment of the present application, if the detection result is that the target fault type is consistent with the backup fault type, the target fault type is displayed to the user in a graphical interface, and a preset operation strategy corresponding to the target fault type is generated, and a control instruction is generated based on the operation strategy based on the result that the target fault type is consistent with the backup fault type; if the detection result is that the target fault type is inconsistent with the backup fault type, the target fault type is displayed to the user in a graphical interface, and a preset operation strategy corresponding to the target fault type is generated; based on the operation strategy based on the result that the target fault type is inconsistent with the backup fault type, a control instruction is generated, thereby ensuring the accuracy of the generated control instruction.

[0064] In an optional embodiment, the method further includes:

[0065] Monitoring the execution result of the target vehicle executing the operation strategy corresponding to the target fault type;

[0066] Record the target fault type, target fault type occurrence time and execution result corresponding to the target vehicle.

[0067] The fault diagnosis method provided in the embodiment of the present application monitors the execution results of the operation strategy corresponding to the target fault type executed by the target vehicle, records the target fault type corresponding to the target vehicle, the time when the target fault type occurs, and the execution result, thereby tracing the fault of the target vehicle based on the recorded target fault type corresponding to the target vehicle, the time when the target fault type occurs, and the execution result.

[0068] In a second aspect, the present invention provides a fault diagnosis method, which is applied to a target vehicle, and the method comprises:

[0069] Collecting initial vehicle data corresponding to the target vehicle;

[0070] Perform data fusion and cleaning on the initial vehicle data to obtain processed target vehicle data;

[0071] Comparing various sub-target vehicle data in the target vehicle data with data thresholds corresponding to the sub-target vehicle data;

[0072] If there is at least one type of sub-target vehicle data that is greater than the corresponding data threshold, generating fault code information based on the sub-target vehicle data that is greater than the corresponding data threshold;

[0073] The target vehicle data and fault code information are encrypted to generate an encrypted data packet; and the encrypted data packet is transmitted to a server; the server is a server of the first aspect or any corresponding implementation manner thereof.

[0074] The fault diagnosis method provided in the embodiment of the present application collects the initial vehicle data corresponding to the target vehicle; performs data fusion and cleaning processing on the initial vehicle data to obtain the processed target vehicle data, thereby ensuring the accuracy of the obtained target vehicle data. The various sub-target vehicle data in the target vehicle data are compared with the data threshold corresponding to the sub-target vehicle data, so that the various sub-target vehicle data can be detected. If there is at least one type of sub-target vehicle data greater than the corresponding data threshold, fault code information is generated based on the sub-target vehicle data greater than the corresponding data threshold, thereby ensuring the accuracy of the generated fault code information. The target vehicle data and the fault code information are encrypted to generate an encrypted data packet; and the encrypted data packet is transmitted to the server, so that the server can obtain the target vehicle data and the fault code information corresponding to the target vehicle data, and then determine the target fault type corresponding to the target vehicle based on the target vehicle data and the fault code information corresponding to the target vehicle data. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0076] Figure 1 is a flow chart of a fault diagnosis method according to an embodiment of the present invention;

[0077] Figure 2 is a flow chart of another fault diagnosis method according to an embodiment of the present invention;

[0078] Figure 3 is a flow chart of another fault diagnosis method according to an embodiment of the present invention;

[0079] Figure 4 is a structural block diagram of a fault diagnosis device according to an embodiment of the present invention;

[0080] Figure 5 is a structural block diagram of another fault diagnosis device according to an embodiment of the present invention;

[0081] Figure 6 It is a schematic diagram of the hardware structure of the server according to an embodiment of the present invention. DETAILED DESCRIPTION

[0082] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0083] Remote diagnosis and maintenance of new energy vehicles is an intelligent service system based on modern communication technology and the Internet of Things. It uses on-board sensors, data transmission and cloud platform analysis to achieve real-time monitoring of the operating status of new energy vehicles, fault warning and remote maintenance, greatly improving the safety of vehicles and user experience. The system usually includes multiple links such as data acquisition, transmission, storage, analysis and remote control. The data acquisition module is responsible for collecting key information such as the battery status, motor performance, charging status of the vehicle, and sending these data to the cloud server through wireless transmission technology. The server stores and analyzes the data. Once an abnormality is found, the system will immediately issue an early warning and can perform fault troubleshooting and some maintenance operations through remote control. The advantage of remote diagnosis and maintenance of new energy vehicles is that it can detect and deal with potential problems in a timely manner, reduce vehicle failure rates, and reduce maintenance costs. At the same time, it provides convenient online services for car owners, such as appointment maintenance and online consultation, which effectively improves the quality of after-sales service.

[0084] However, existing fault diagnosis methods often rely on a single algorithm, which is difficult to deal with complex and changeable fault conditions, and the diagnostic accuracy needs to be improved. At the same time, the existing remote diagnosis and maintenance system has deficiencies in module integration, user interaction and system management, which affects the overall performance of the system and user experience.

[0085] Therefore, how to diagnose faults of new energy vehicles has become an urgent problem to be solved.

[0086] According to an embodiment of the present invention, a fault diagnosis method embodiment is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0087] In this embodiment, a fault diagnosis method is provided, which can be applied to a server. Figure 1 is a flow chart of a fault diagnosis method according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:

[0088] Step S101, receiving target vehicle data sent by a target vehicle and fault code information corresponding to the target vehicle data.

[0089] Among them, the fault code information is generated after the target vehicle detects the target vehicle data.

[0090] Specifically, the server may receive the target vehicle data and the fault code information corresponding to the target vehicle data sent by the target vehicle based on the communication connection with the target vehicle.

[0091] Among them, the target vehicle data may include battery cell data, motor data, transmission system data and battery environment data corresponding to the target vehicle. Among them, the battery cell data includes battery voltage data, battery current data, battery temperature data, battery internal resistance data, battery remaining capacity (SOC) data, battery health status (SOH) data, etc. The motor data includes motor speed data, motor torque data, motor temperature data, motor current data, motor vibration data and motor efficiency data, etc. The transmission system data includes transmission system speed data, transmission system torque data, transmission system oil temperature data, transmission system oil pressure data, transmission system vibration data, transmission system clearance data, etc. The battery environment data includes humidity data of the battery surrounding environment, temperature data of the battery surrounding environment, etc. The embodiment of the present application does not specifically limit the target vehicle data.

[0092] This step will be described in detail below.

[0093] Step S102: determining the backup fault type corresponding to the target vehicle based on the fault code information.

[0094] Specifically, the server may determine the alternate fault type corresponding to the target vehicle according to the correspondence between the fault code information and the alternate fault category.

[0095] Exemplarily, the server may use the fault code mapping formula: DTC_description=Map(DTC_code) to determine the alternate fault type corresponding to the target vehicle.

[0096] Where DTC_code is the fault code information, and DTC_description is the backup fault type.

[0097] Step S103, identifying the target vehicle data, and determining the target fault type corresponding to the target vehicle based on a preset fault diagnosis model.

[0098] Specifically, the server can first perform data cleaning operations on the target vehicle data, including removing null values, outliers, and duplicate data. Then, the server performs data conversion on the cleaned target vehicle data, including converting the voltage value from an analog signal to a digital signal, or unifying the date and time format. In addition, the server will also perform data normalization or standardization processing so that data of different dimensions can be compared and analyzed; using the minimum-maximum normalization formula:

[0099] X'=(X-X_min) / (X_max-X_min);

[0100] Where X is the original data, X_min and X_max are the minimum and maximum values ​​of the data respectively, and X' is the standardized data.

[0101] Then, the server inputs the standardized target vehicle data into a preset fault diagnosis model, the preset fault diagnosis model extracts features from the standardized target vehicle data, and then determines the target fault type corresponding to the target vehicle based on the extracted features.

[0102] This step will be described in detail below.

[0103] Step S104, detecting whether the target fault type is consistent with the backup fault type.

[0104] Specifically, the server may compare the target fault type corresponding to the determined target vehicle to detect whether the target fault type is consistent with the backup fault type.

[0105] Step S105: generating a control instruction based on the detection result and the target fault type, and sending the control instruction to the target vehicle.

[0106] The control instruction includes a preset operation strategy corresponding to the target fault type.

[0107] Specifically, the server may generate a control instruction based on the detection result and the target fault type, and send the control instruction to the target vehicle.

[0108] This step will be described in detail below.

[0109] The fault diagnosis method provided in the embodiment of the present application receives the target vehicle data sent by the target vehicle and the fault code information corresponding to the target vehicle data. Then, based on the fault code information, the alternative fault type corresponding to the target vehicle is determined, thereby ensuring the accuracy of the determined alternative fault type. Then, the target vehicle data is identified, and the target fault type corresponding to the target vehicle is determined based on a preset fault diagnosis model, thereby ensuring the accuracy of the target fault type corresponding to the determined target vehicle. Then, it is detected whether the target fault type is consistent with the alternative fault type; based on the detection result and the target fault type, a control instruction is generated, and the control instruction is sent to the target vehicle, thereby ensuring the accuracy of the generated control instruction, so that the target vehicle can receive the control instruction and then execute the preset operation strategy. In addition, the above method first receives the fault code information generated by the target vehicle after detecting the target vehicle data, and identifies the target vehicle data, and determines the target fault type corresponding to the target vehicle based on the preset fault diagnosis model, thereby realizing dual fault diagnosis of the target vehicle, ensuring the accuracy of fault diagnosis of the target vehicle, and then ensuring the accuracy of the generated control instruction.

[0110] In this embodiment, a fault diagnosis method is provided, which can be applied to a server. Figure 2 is a flow chart of a fault diagnosis method according to an embodiment of the present invention. Figure 2 As shown, the process includes the following steps:

[0111] Step S201, receiving target vehicle data sent by a target vehicle and fault code information corresponding to the target vehicle data.

[0112] Among them, the fault code information is generated after the target vehicle detects the target vehicle data.

[0113] Specifically, the above step S201 may include the following steps:

[0114] Step S2011, receiving the encrypted data packet sent by the target vehicle.

[0115] Specifically, the server may receive an encrypted data packet sent by the target vehicle based on the communication connection with the target vehicle, wherein the encrypted data packet may include a preset encryption algorithm and a key corresponding to the encrypted data packet.

[0116] Step S2012, decrypt the encrypted data packet to obtain the target vehicle data and the fault code information corresponding to the target vehicle data.

[0117] Specifically, the server may decrypt the encrypted data packet based on a decryption algorithm and a key that match a preset encryption algorithm, and obtain the target vehicle data and the fault code information corresponding to the target vehicle data.

[0118] For example, if the target vehicle is encrypted using the AES (Advanced Encryption Standard) preset encryption algorithm, the server uses the same AES algorithm and the corresponding key to decrypt the encrypted data packet to obtain the target vehicle data and the fault code information corresponding to the target vehicle data.

[0119] Optionally, when the server decrypts the encrypted data packet based on the decryption algorithm and key that match the preset encryption algorithm, a hardware security module (HSM) can be used to store and manage the decryption key to ensure the security and confidentiality of the key.

[0120] Step S2013, verifying the target vehicle data and fault code information.

[0121] Specifically, after decrypting the target vehicle data and the fault code information corresponding to the target vehicle data, the server can use a hash function (such as SHA-256) to calculate the hash value of the decrypted target vehicle data and fault code information, and then compare it with the original hash value sent by the target vehicle. If the two hash values ​​are consistent, it means that the data has not been tampered with during the transmission process, ensuring the integrity of the data.

[0122] At the same time, the server can also perform format verification and semantic verification on the target vehicle data and fault code information. Format verification checks whether the data conforms to the predetermined data format specifications, such as whether the length and type of the data field are correct. Semantic verification checks according to the business logic of the data, such as whether the battery voltage value is within a reasonable range, whether the vehicle speed conforms to the laws of physics, etc. If data format or semantic errors are found, the system will automatically mark the data and try to repair it or request retransmission.

[0123] Step S2014: if the target vehicle data and the fault code information are successfully verified, then various sub-target vehicle data included in the target vehicle data are identified.

[0124] Specifically, if the target vehicle data and the fault code information are successfully verified, the server identifies various sub-target vehicle data included in the target vehicle data.

[0125] Step S2015, determining the target database corresponding to each type of sub-target vehicle data according to the recognition results corresponding to each type of sub-target vehicle data.

[0126] Specifically, the server determines the target database corresponding to each type of sub-target vehicle data according to the recognition results corresponding to each type of sub-target vehicle data.

[0127] For example, the storage database is selected according to the characteristics and query requirements of various sub-target vehicle data. For data with strong structure and complex query requirements (such as multi-table association query, transaction processing, etc.), the relational database MySQL is selected. For example, there are many associations between the basic information of the vehicle (frame number, model, production date, etc.), user information (owner name, contact information, etc.) and fault diagnosis history records (fault code, diagnosis time, maintenance measures, etc.), which are suitable for storage in MySQL to facilitate association query and data analysis.

[0128] For semi-structured or unstructured data, such as real-time data streams collected by vehicle sensors (data sequences that change over time, such as temperature and voltage), vehicle driving trajectory data (GPS coordinate point sequences), etc., choose the NoSQL database MongoDB. MongoDB's document-based data model can easily store and process such flexible data structures, and can provide efficient write performance, which is suitable for processing large amounts of real-time data.

[0129] Step S2016, storing each type of sub-target vehicle data in a corresponding target database, and indexing each type of sub-target vehicle data in the target database.

[0130] Specifically, the server stores various sub-target vehicle data in the corresponding target database. Then, the electronic device can index various sub-target vehicle data in the target database based on a preset method.

[0131] For example, in MySQL, a B-tree index is created for the fault code field in the vehicle fault diagnosis history record table, so that when querying the history record of a specific fault code, the relevant records can be quickly located, greatly improving the query speed.

[0132] In MongoDB, appropriate indexes are created based on the query patterns of the data. If sensor data is frequently queried based on vehicle ID and time range, a composite index (vehicle_id, timestamp) can be created to optimize such query operations. At the same time, index automatic optimization technology is used to automatically adjust the index structure based on the execution of query statements and changes in data distribution to ensure that the index always remains efficient.

[0133] In an optional implementation of the present application, the server can automatically hierarchically store various sub-target vehicle data in storage media with different performances according to the access frequency and importance of various sub-target vehicle data. For example, various sub-target vehicle data that are frequently accessed recently are stored in a high-speed solid-state drive (SSD) to provide fast data reading and writing performance; while various sub-target vehicle data with low access frequency are stored in a large-capacity mechanical hard disk (HDD) or an archive storage layer of cloud storage, which can not only meet data storage needs but also reduce storage costs.

[0134] In addition, the server can also establish a data classification management strategy to automatically migrate various sub-target vehicle data between different storage levels as the frequency of data access changes. For example, when some historical data is suddenly frequently accessed (such as for fault backtracking analysis), the system automatically migrates it from the archive storage layer to the high-speed storage layer to increase data access speed, and then migrates it back to the original storage layer after the access frequency decreases.

[0135] Step S202: Determine the backup fault type corresponding to the target vehicle based on the fault code information.

[0136] For details about this step, please refer to the above description of step S102, which will not be elaborated here.

[0137] Step S203, identifying the target vehicle data, and determining the target fault type corresponding to the target vehicle based on a preset fault diagnosis model.

[0138] Specifically, the preset fault diagnosis model includes a fault determination model and a fault type identification model. The above step S203 may include the following steps:

[0139] Step S2031, inputting the target vehicle data into a fault determination model to determine whether the target vehicle has a fault.

[0140] Specifically, the server may pre-process the target vehicle data, wherein the pre-processing process may include operations such as data cleaning, data normalization or standardization.

[0141] Then, the server inputs the preprocessed target vehicle data into the fault determination model, and the fault determination model extracts spare features related to the fault from the preprocessed target vehicle data.

[0142] For example, for battery cell data, in addition to the basic voltage, current, and temperature, features such as the battery charge and discharge rate (by calculating the change in power per unit time), the rate of change of the battery internal resistance (reflecting the change in the internal structure of the battery) and other features can also be extracted. For motor data, in addition to speed, torque, and temperature, the motor efficiency (the ratio of output power to input power), power factor (reflecting the efficiency of motor power utilization), and the spectrum characteristics of motor vibration (obtained by Fourier transforming the vibration signal, different faults may correspond to different spectrum characteristics) are extracted. For transmission system data, transmission efficiency (the ratio of output torque to input torque), oil temperature change rate, oil pressure fluctuation characteristics, etc. are extracted. These extracted features can more comprehensively reflect the operating status of the vehicle system and provide more valuable information for fault diagnosis.

[0143] Next, the fault determination model uses a preset feature selection method to select the target feature subset that is most helpful for fault diagnosis from the numerous extracted alternative features.

[0144] For example, the correlation analysis method can be used to calculate the correlation coefficient between each feature and the fault label (fault or normal), remove the features with low correlation, and retain the key features with high correlation. Another method is feature selection based on information gain, which calculates the information gain of each feature for the classification result (fault or normal) and selects the features with large information gain. Through feature selection, the data dimension is reduced, the amount of calculation is reduced, and the "dimensionality disaster" is avoided, thereby improving the diagnostic accuracy of the fault determination model.

[0145] Then, based on the target feature subset and the decision function Calculate the function value corresponding to the target feature subset.

[0146] Among them, if the calculated function value is greater than 0, the fault determination model predicts that the target vehicle has a fault (the category label is +1); if the function value is less than or equal to 0, the fault determination model predicts that the target vehicle is in a normal state (the category label is -1).

[0147] Step S2032: If the target vehicle fails, the target vehicle data is input into a fault type identification model to determine a target fault type corresponding to the target vehicle.

[0148] Specifically, the fault type identification model includes a target fault tree network structure, and the above step S2032 may include the following steps:

[0149] Step a1, traverse each top event in the target fault tree network structure.

[0150] Specifically, when the target vehicle is determined to have a fault, the server inputs the pre-processed target vehicle data and the historical data fragments related to the time of the fault (such as data within a period of time before the fault) as a data sample into the fault type identification model. These data contain parameter information of each key system of the vehicle (such as battery system, motor system, control system, etc.), such as battery voltage, current, temperature, motor speed, torque, control signal, etc.

[0151] Then, the fault type identification model first analyzes the target fault tree network structure based on the pre-built target fault tree network structure. Starting from the top event of the target fault tree network structure (such as "the vehicle cannot start" and "the vehicle lacks power", etc.), it traverses down along the branches of the tree to check the correlation between each intermediate event and bottom event and the input data.

[0152] For example, for the top event "the vehicle cannot start", check the associated intermediate events "battery failure", "motor failure", "starting system failure", etc. By analyzing the battery-related parameters in the target vehicle data (such as whether the battery voltage is too low, whether the battery current is abnormal, etc.), determine whether there are signs of "battery failure"; evaluate whether it is a "motor failure" based on parameters such as the motor speed, torque data, and motor temperature; determine whether it is a "starting system failure" based on the control signal of the starting system, the status of the starting relay, and other data.

[0153] Step a2: detecting the correlation between the intermediate events and bottom events corresponding to each top event and the target vehicle data.

[0154] Specifically, the fault type identification model may calculate the correlation coefficient between the vehicle data features related to each intermediate event and the bottom event and the target vehicle data based on the Pearson correlation coefficient.

[0155] Step a3, determining the target intermediate event and / or the target bottom event based on the correlation.

[0156] Specifically, the fault type identification model compares the calculated correlation with the correlation threshold, and determines the intermediate events and / or bottom events whose correlation is greater than the correlation threshold as target intermediate events and / or target bottom events.

[0157] Step a4, obtaining the logical relationship between each target intermediate event and / or each target bottom event.

[0158] Specifically, the fault type identification model obtains the logical relationship between each target intermediate event and / or each target bottom event according to the relationship between the logic gates in the target fault tree network structure.

[0159] Step a5, determining the preliminary fault type corresponding to the target vehicle according to the logical relationship.

[0160] Specifically, the fault type identification model determines the preliminary fault type corresponding to the target vehicle according to the logical relationship.

[0161] For example, if it is an AND gate connection, such as some complex vehicle control system failures that may require multiple sensors to be abnormal at the same time to be triggered, the model will check whether all relevant conditions are met. Only when all subordinate events related to the AND gate have corresponding abnormal performance in the data, it is determined to belong to the preliminary fault type.

[0162] Step a6: based on the preliminary fault type, extracting fault data related to the preliminary fault type from the target vehicle data.

[0163] Specifically, the server may extract fault data related to the preliminary fault type from the target vehicle data based on the preliminary fault type.

[0164] For example, taking "battery failure" as an example, in addition to the basic battery parameters mentioned above, the server may also calculate the battery's internal resistance change rate, battery capacity attenuation and other characteristics. For motor failure, analyze the motor's vibration spectrum characteristics, electromagnetic torque fluctuation characteristics, etc. These characteristics can more deeply reflect the specific nature of the fault.

[0165] Step a7, extracting features from the fault data to generate target fault features.

[0166] Specifically, the server can use statistical analysis methods and signal processing technology to extract features from fault data and generate target fault features.

[0167] For example, calculate the mean, variance, maximum, minimum and other statistics of the features, and perform spectrum analysis and wavelet transform on the signal to obtain information about the features in different domains, so as to better identify the fault mode.

[0168] Step a8, matching the target fault signature with a predetermined fault signature in a predefined fault mode library.

[0169] Specifically, the server matches the target fault feature with the predetermined fault feature in the predefined fault mode library. The fault mode library stores the predetermined fault features corresponding to various known fault types, and these patterns are obtained by analyzing and summarizing a large amount of historical fault data. For example, for a battery cell short circuit fault, it has a specific pattern in terms of battery voltage, current curve, and internal resistance change; a motor winding short circuit fault has a typical manifestation in the motor vibration spectrum and electromagnetic torque characteristics.

[0170] Then, the server may calculate the similarity (such as Euclidean distance, cosine similarity, etc.) between the target fault feature and the predetermined fault feature.

[0171] Step a9: Determine the target fault type corresponding to the target vehicle according to the matching result.

[0172] Specifically, the server may determine, based on the similarity between the target fault feature and the predetermined fault feature, that the fault type corresponding to the predetermined fault feature with the greatest similarity is the target fault type corresponding to the target vehicle.

[0173] In an optional implementation manner of the present application, the preset fault diagnosis model includes a fault determination model and a fault type identification model, and the training process of the preset fault diagnosis model includes:

[0174] Step b1, obtaining historical fault data and historical normal data.

[0175] Specifically, the server can receive historical fault data input by users, or historical fault data sent by multiple vehicles. Historical fault data can include the after-sales maintenance record database of the vehicle manufacturer, the maintenance case records of the automobile maintenance service station, the historical fault data uploaded by the vehicle remote diagnosis system, and the data provided by professional automobile fault research institutions, etc. These data cover the fault conditions of vehicles of different models, different usage environments, and different years of use, ensuring the diversity and comprehensiveness of the data.

[0176] In addition, the server can receive historical normal data input by the user, and can also receive historical normal data sent by multiple vehicles.

[0177] Step b2: generating fault training data based on the fault case data in the historical fault data.

[0178] Specifically, the server may label the historical fault data based on the fault case data corresponding to each historical fault data, that is, label each historical fault data with a fault label, thereby generating fault training data.

[0179] Step b3, generating normal training data based on historical normal data.

[0180] Specifically, the server labels historical normal data and generates normal training data.

[0181] Step b4: generating a first training data set based on the fault training data and the normal training data.

[0182] Specifically, the server mixes the fault training data and the normal training data to generate a first training data set.

[0183] Step b5: Based on the first training data set, train the initial fault determination network to generate a fault determination model.

[0184] Specifically, the above step b5 may include the following steps:

[0185] Step b51, selecting a radial basis kernel function.

[0186] Specifically, the server may select the radial basis kernel function K(x, x′).

[0187] Where K(x,x′)=exp(-γ‖xx′‖ 2 )

[0188] Among them, K(x,x′) represents the kernel function, which is used to calculate the inner product of two sample points x and x′ in the original feature space in the mapped feature space, that is, K(x,x′)=φ(x)Tφ(x′); the radial basis function is used as the kernel function here; γ represents the kernel function parameter, which is used to control the width of the RBF kernel function; a larger γ value will result in a narrower kernel function shape, while a smaller γ value will result in the opposite;

[0189] Step b52, constructing the optimization problem.

[0190] Specifically, the objective function is usually to minimize the bi-norm of the weight vector ||w|| 2 , its purpose is to find an optimal hyperplane so that different categories of data (faulty data and normal data) are separated as much as possible in the feature space, and the interval is the largest. This hyperplane is determined by the weight vector and the bias term, which determines the direction of the hyperplane and the distance between the hyperplane and the origin.

[0191] The constraints are based on each sample point x in the training data set. i Mark y i (y i ∈{-1, +1}, representing normal and fault categories respectively) and the feature mapping function φ(x i ). The constraint condition requires that for each sample point x i , must satisfy y i (w T φ(x i )+b)≥1, which means that the distance from the sample point to the hyperplane must be greater than or equal to a minimum interval (here 1 is taken for the convenience of calculation and standardization). This constraint ensures the accuracy of classification, that is, the fault data and normal data are correctly divided on both sides of the hyperplane and have a certain interval to improve the robustness of the model.

[0192] Therefore, the optimization problem is constructed as:

[0193]

[0194] Among them, w represents the weight vector, which represents the normal vector of the hyperplane defined in the feature space; in SVM, w is the key component of the decision function, which determines the direction of the hyperplane; α i Represents the Lagrange multiplier. Each sample point xi has a corresponding Lagrange multiplier α i , these multipliers are used to balance the model complexity and the penalty for misclassification during the optimization process; φ(x) represents the feature mapping function, which is used to map the original features to a higher-dimensional feature space for easy classification; N represents the number of training samples.

[0195] Step b53, using the Lagrange multiplier method, transforms the optimization problem to obtain the Lagrangian function.

[0196] Specifically, in order to solve the above optimization problem with constraints, the Lagrange multiplier α is introduced i (i=1,2,...N, N is the number of training samples). By using the Lagrange multiplier method, the constraints are combined with the objective function to construct the Lagrangian function)

[0197]

[0198] Among them, L(w,b,α) represents the Lagrangian function, which is the Lagrangian form of the original optimization problem and is used to minimize the objective function while satisfying the constraints.

[0199] The deformation of the Lagrangian function allows us to solve the optimization problem by taking partial derivatives of w, b, and α and setting them to 0. In the process of taking partial derivatives, the kernel function K(x, x′) = φ(x) is used. T The property of φ(x′) transforms the inner product operation in high-dimensional space into the calculation of kernel function in low-dimensional space, avoiding complex calculations directly in high-dimensional space. For example, when calculating the partial derivative of w with respect to the Lagrangian function, we can get This expression represents the weight vector w as a linear combination of training samples, which facilitates subsequent calculations and understanding of the model.

[0200] Step b54, solving the Lagrangian function through the KKT condition to obtain a solution result.

[0201] Specifically, the server solves the extreme value of the Lagrangian function through the KKT (Karush-Kuhn-Tucker) condition. The KKT condition includes complementary slackness, primal feasibility, and dual feasibility. Based on these conditions, we can get the i Solving this system of equations gives us α iThen, using the already obtained α i and The weight vector and the bias term can be determined by using the relationship between them, thus obtaining the initial vehicle diagnosis network model. This initial model is constructed based on the training data set and the selected radial basis kernel function by solving the optimization problem, laying the foundation for subsequent model optimization and verification.

[0202] Step b55: Generate an initial fault determination network based on the solution result.

[0203] Specifically, the initial fault determination network can be regarded as a classifier based on the principle of support vector machine, the core of which is the hyperplane determined by the weight vector w and the bias term b. After the input data is feature extracted (such as extracting various parameter features in vehicle operation data), it is mapped to a high-dimensional feature space through the radial basis kernel function, and then the decision rule of the hyperplane (such as f(x) = w T φ(x)+b, when f(x)>0, it is judged as a fault category, and when f(x)≤0, it is judged as a normal category) to classify and diagnose the new data samples.

[0204] Step b56: cross-validate the initial fault determination network based on the first training data set to determine the model performance corresponding to the initial fault determination network.

[0205] Specifically, the server may input the first training data set into the initial fault determination network, perform cross-validation on the initial fault determination network, and determine the model performance corresponding to the initial fault determination network.

[0206] Exemplarily, the server can use a k-fold cross-validation method to evaluate the performance of the initial fault determination network. Divide the training data set into k subsets of similar size (usually k=5 or k=10). Take one of the subsets as the validation set and the remaining k-1 subsets as the training set, and perform k training and validation. In each training process, the initial fault determination network is trained using the training set, and then the performance indicators of the model are calculated on the validation set, such as accuracy, recall, precision, F1 value, etc.

[0207] The accuracy calculation formula is: TP (True Positive) represents true positive examples, that is, the number of samples that are actually faulty and predicted as faulty by the model; TN (True Negative) represents true negative examples, that is, the number of samples that are actually normal and predicted as normal by the model; FP (False Positive) represents false positive examples, that is, the number of samples that are actually normal but predicted as faulty by the model; FN (False Negative) represents false negative examples, that is, the number of samples that are actually faulty but predicted as normal by the model. The accuracy reflects the overall classification accuracy of the model.

[0208] The recall rate calculation formula is: It measures the model's ability to identify fault samples, that is, the proportion of actual fault samples that are correctly predicted to be faults. The accuracy calculation formula is: Indicates the proportion of samples predicted by the model to be faulty that are actually faulty. It is a more comprehensive performance evaluation indicator that takes into account both precision and recall. Through the calculation and analysis of these indicators, we can fully understand the performance of the initial fault determination network on different data subsets and evaluate the stability and reliability of the model.

[0209] Step b57, adjusting the parameters of the initial fault determination network according to the model performance to obtain a fault determination model.

[0210] Specifically, the server can adjust the parameters of the initial fault determination network according to the model performance indicators obtained by cross-validation. The main parameters to be adjusted include γ in the radial basis kernel function, the penalty parameter C, etc. The penalty parameter is used to control the complexity of the model and the degree of penalty for misclassified samples. When C is large, the model will impose a large penalty on misclassified samples and will try to minimize the classification error of the training set, but may cause overfitting; when C is small, the model will impose a small penalty on misclassified samples, and the model will tend to be simplified, which may lead to underfitting. By adjusting these parameters, a balance is found between the complexity and generalization ability of the model to improve the performance of the model.

[0211] Use grid search or random search to find the optimal parameter combination. Grid search is to traverse all possible combinations in a pre-set parameter value grid, calculate the performance indicators of the model under each combination, and select the parameter combination with the best performance. Random search is to randomly select a certain number of parameter combinations in the parameter space for evaluation. Although it cannot guarantee the global optimal solution, it can improve the search efficiency to a certain extent, especially when the parameter space is large.

[0212] Finally, a fault determination model is generated based on the parameter combination with the best performance.

[0213] Step b6, based on the historical fault data, determine the top event, the middle event and the bottom event, and use logic gates to connect the top event, the middle event and the bottom event to form a fault tree.

[0214] Specifically, the server can identify historical fault data, determine the fault frequency statistics and the impact level of vehicle key system faults. Then, based on the fault frequency statistics and the impact level of vehicle key system faults, select representative faults with a severity greater than the preset severity level as top events. For example, in addition to the common "vehicle cannot start" and "vehicle lacks power", top events such as "vehicle suddenly loses power while driving" and "battery thermal runaway risk" can also be considered. These top events directly affect the safety and normal use of the vehicle, and are fault conditions that require special attention and analysis.

[0215] The definition of top events should be as clear and specific as possible to avoid ambiguity. For example, for the top event of "battery thermal runaway risk", it can be clearly defined as the battery temperature rising above a certain threshold in a short period of time and continuing to rise, accompanied by abnormal fluctuations in battery voltage and current, so as to accurately guide the construction and analysis of subsequent fault trees.

[0216] Then, the server can determine the intermediate events and bottom events that lead to the top event based on the analysis results of the historical fault data. Taking the top event "the vehicle cannot start" as an example, in addition to "battery failure", "motor failure", and "starting system failure", the intermediate events can also consider "vehicle anti-theft system abnormality" and "electronic control unit (ECU) failure". For the intermediate event "battery failure", the bottom events include "battery cell short circuit", "loose battery connector", "battery electrolyte leakage", "battery management system failure", etc.

[0217] When determining intermediate events and bottom events, the interaction between different systems should be fully considered. For example, a motor failure may cause excessive discharge of the battery, which in turn causes a battery failure. This interaction relationship should be accurately reflected in the fault tree, and it may be necessary to add associated events related to the motor failure under the battery failure branch.

[0218] Finally, the server uses logic gates to connect top events, middle events, and bottom events based on the causal and logical relationships of the failure. For "AND" connections, for example, some complex vehicle control system failures may require multiple sensors to be abnormal at the same time to be triggered, so these sensor failure events and control system failure events are connected with "AND" gates. For "OR" connections, for example, "battery failure" may be caused by multiple reasons such as "battery cell short circuit" or "battery connector loosening", they are connected with "battery failure" using "OR" gates.

[0219] When constructing logical relationships, the uncertainty factor is introduced. For example, for some events that are difficult to determine the cause and effect relationship, fuzzy logic gates can be used to connect them, and fuzzy membership can be used to represent the degree of association between events, so that the fault tree is more in line with the actual complex situation. At the same time, combined with expert experience and historical data, the parameters of the logic gate (such as the threshold of the AND gate, the probability distribution of the OR gate, etc.) are optimized to improve the accuracy of the fault tree.

[0220] Step b7: Generate a second training data set based on historical fault data.

[0221] The second training data set includes the fault type corresponding to each historical fault data.

[0222] Specifically, the server may identify the historical fault data, determine the fault type corresponding to the historical fault data, and mark the fault type of the historical fault data based on the fault tree, thereby generating a second training data set.

[0223] For example, if a data sample contains features such as "battery cell short circuit" and "battery connector loose", and according to the fault tree logic, these features will lead to "battery failure", which may lead to "vehicle cannot start" (top event), then the sample is labeled as "vehicle cannot start-battery failure" type. This labeling method can closely integrate the training data with the logic of the fault tree and provide accurate target output for network learning.

[0224] For some complex fault situations, there may be multiple possible fault paths and type combinations. When labeling, multi-label labeling or probability labeling is used. For example, for a data sample that involves partial anomalies of both the battery and motor systems, it may be labeled as "Vehicle insufficient power-battery failure (probability 0.6)-motor failure (probability 0.4)", indicating that the sample has a high probability of being caused by battery failure due to vehicle insufficient power, but there is also a certain probability of involving motor failure, so that the labeling can better reflect the uncertainty of the actual fault.

[0225] Step b8: Based on the fault tree and the second training data set, train an initial fault type identification network to generate a fault type identification model.

[0226] Specifically, the above step b8 may include the following steps:

[0227] Step b81, generating an initial fault type identification network adapted to the fault tree according to the hierarchical structure and logical relationship of the fault tree.

[0228] Among them, the initial fault type identification network is an initial fault tree network structure.

[0229] Specifically, the server can design an initial fault type identification network that is compatible with the fault tree according to the hierarchical structure and logical relationship of the fault tree.

[0230] For example, a hierarchical neural network structure can be used, in which the top-level network nodes correspond to the top events of the fault tree (such as "vehicle cannot start", "vehicle power system failure", etc.), the middle-level nodes correspond to the intermediate events (such as "battery failure", "motor failure", "starting system failure", etc.), and the bottom-level nodes correspond to the bottom events (such as "battery monomer short circuit", "battery connector loose", "motor winding short circuit", etc.). Such an architecture can intuitively reflect the cause-effect relationship in the fault tree, which is convenient for model learning and reasoning.

[0231] Step b82, generating logic gate simulation units in the initial fault type identification network according to the logic gates in the fault tree.

[0232] Specifically, according to the logic gates in the fault tree, a logic gate simulation unit is generated in the initial fault type identification network.

[0233] For example, for the AND gate relationship, a multiplication unit can be designed to output a higher value only when all input conditions (corresponding to the lower-level events connected to the AND gate) are met. For the OR gate relationship, an addition unit is used, and a higher value is output as long as one input condition is met. Through these logic gate simulation units, the network can accurately simulate the logical relationship in the fault tree and improve the accuracy of fault type identification.

[0234] Step b83, inputting the second training data into the initial fault type recognition network, training the initial fault type recognition network based on a preset training algorithm, and generating a fault type recognition model.

[0235] Among them, the fault type identification model includes the target fault tree network structure.

[0236] Specifically, the server may input the second training data into the initial fault type identification network, train the initial fault type identification network based on a preset training algorithm, adjust the parameters of the initial fault type identification network, and generate a fault type identification model.

[0237] The preset training algorithm can be stochastic gradient descent (SGD) and its variants (Adagrad, Adadelta, Adam, etc.). According to the characteristics of the training data and the network architecture, the parameters of the algorithm (such as learning rate, momentum, etc.) are adjusted. For example, in the early stage of training, a larger learning rate can be used to speed up the convergence speed. As the training progresses, the learning rate is gradually reduced to prevent the model from oscillating around the optimal solution.

[0238] Step S204, detecting whether the target fault type is consistent with the backup fault type.

[0239] For details about this step, please refer to the above description of step S104, which will not be elaborated here.

[0240] Step S205: Generate a control instruction based on the detection result and the target fault type, and send the control instruction to the target vehicle.

[0241] The control instruction includes a preset operation strategy corresponding to the target fault type.

[0242] Specifically, the above step S205 may include the following steps:

[0243] Step S2051, if the detection result is that the target fault type is consistent with the backup fault type, the target fault type is displayed to the user in a graphical interface, and a preset operation strategy corresponding to the target fault type is generated. Based on the operation strategy and the result that the target fault type is consistent with the backup fault type, a control instruction is generated.

[0244] Specifically, if the detection result is that the target fault type is consistent with the backup fault type, the target fault type is displayed to the user in a graphical interface, and a preset operation strategy corresponding to the target fault type is generated. Based on the operation strategy and the result that the target fault type is consistent with the backup fault type, a control instruction is generated.

[0245] Step S2052, if the detection result is that the target fault type is inconsistent with the backup fault type, the target fault type is displayed to the user in a graphical interface, and a preset operation strategy corresponding to the target fault type is generated; based on the operation strategy and the result that the target fault type is inconsistent with the backup fault type, a control instruction is generated.

[0246] Specifically, if the detection result is that the target fault type is inconsistent with the backup fault type, the target fault type is displayed to the user in a graphical interface, and a preset operation strategy corresponding to the target fault type is generated; based on the operation strategy and the result of the inconsistency between the target fault type and the backup fault type, a control instruction is generated.

[0247] For example, the control instructions may include preset operating strategies such as adjusting the working mode of the battery management system, starting an emergency shutdown procedure, and executing a software upgrade.

[0248] Optionally, users can enter query conditions through the interface to obtain historical data or specific information. In addition, the target vehicle can also receive control commands sent by the user, including remote start charging and scheduled maintenance services; all user operations will be recorded and fed back to the system management module for permission verification and operation auditing.

[0249] Step S206, monitoring the execution result of the target vehicle executing the operation strategy corresponding to the target fault type.

[0250] Specifically, when the target vehicle executes the operation strategy for the target fault type, the target vehicle can establish a real-time monitoring mechanism through the on-board sensors and monitoring system. For example, for the fault operation strategy involving the battery management system (such as adjusting the charging current, voltage, etc.), the sensors inside the battery management system are used to continuously monitor the changes in battery voltage, current, temperature and other parameters; for the fault strategy related to the motor control system (such as adjusting the motor speed, torque control parameters, etc.), the feedback information of the actual motor operating parameters (such as real-time speed, output torque, motor temperature, etc.) is obtained through the sensors in the motor control system; for the fault handling strategy of the vehicle transmission system (such as adjusting the shifting strategy, hydraulic system pressure, etc.), the various sensors in the transmission system (such as speed sensor, oil pressure sensor, oil temperature sensor, etc.) are relied on to collect the operating status data of the transmission system after executing the operation strategy.

[0251] Then, the target vehicle sends the monitored changes in the parameters of the battery unit, the motor, and the transmission system to the server.

[0252] Then, the server determines the parameter changes of the battery cells, the parameter changes of the motors and the parameter changes of the transmission system according to the parameter changes of the battery cells, the parameter changes of the motors and the parameter changes of the transmission system received from the target battery.

[0253] Optionally, the server may adjust a preset operation strategy according to the execution result of the operation strategy corresponding to the target fault type executed by the target vehicle.

[0254] Step S207, recording the target fault type, target fault type occurrence time and execution result corresponding to the target vehicle.

[0255] Specifically, the server accurately records the target fault type corresponding to the target vehicle, including the detailed fault name, fault code (if the vehicle diagnostic system can provide it) and a specific description of the fault. For example, the record is "battery cell short circuit fault (fault code: P0A3F), the No. 3 single cell in the battery pack is short-circuited, causing the battery voltage to drop abnormally and the vehicle power output to be limited." Accurate recording of fault types facilitates subsequent data analysis, fault statistics and case summaries, and provides detailed historical information for vehicle fault diagnosis and maintenance.

[0256] In addition, the server accurately records the occurrence time of the target fault type, and uses the vehicle system time or high-precision time synchronization technology (such as GPS time synchronization) to ensure the accuracy of the time record. The recorded time format should be unified and easy to query and analyze, such as the ISO 8601 standard format (YYYY-MM-DDTHH:MM:SS.sssZ). The time of fault occurrence is an important basis for subsequent analysis of the cause of the fault, assessment of the impact of the fault, and formulation of maintenance plans. Through statistical analysis of the occurrence time of a large number of faults, the time distribution pattern of the fault can be found, such as whether certain faults are more likely to occur in specific seasons, specific mileage, or specific time periods (such as after the vehicle is started or has been running for a long time).

[0257] Finally, the server records the results after the operation strategy is executed in detail, including changes in vehicle operating parameters, whether the fault has been repaired or alleviated, whether new abnormal conditions have occurred, etc. For the record of the execution results of the battery fault handling strategy, for example, "Operation strategy: reduce the charging current to 0.5C, and observe the battery status after continuous charging for 30 minutes. Execution result: the battery voltage gradually rises from the abnormally low value of 2.5V to 3.5V, the battery temperature stabilizes at 40℃ (normal range), the vehicle can start normally, the power output returns to normal, and no new abnormal conditions are found". For the record of the results after the motor fault handling, such as "Operation strategy: adjust the motor torque control parameters to increase the motor output torque by 10% in the low speed range. Execution result: the vehicle's power is significantly enhanced when starting and climbing, the motor speed is stable, the motor temperature rises slightly but is still within the normal range, and the fault phenomenon is effectively alleviated". The record of the execution results should be as detailed and objective as possible to provide a basis for evaluating the effectiveness of the operation strategy and further optimization.

[0258] The fault diagnosis method provided in the embodiment of the present application receives an encrypted data packet sent by the target vehicle; decrypts the encrypted data packet to obtain the target vehicle data and the fault code information corresponding to the target vehicle data, thereby ensuring the accuracy and completeness of the acquired target vehicle data and the fault code information corresponding to the target vehicle data. The target vehicle data and the fault code information are verified; if the target vehicle data and the fault code information are successfully verified, the various sub-target vehicle data included in the target vehicle data are identified; based on the identification results corresponding to the various sub-target vehicle data, the target database corresponding to the various sub-target vehicle data is determined, thereby ensuring the accuracy of the target database corresponding to the various sub-target vehicle data. The various sub-target vehicle data are stored in the corresponding target database, and the various sub-target vehicle data are indexed in the target database, thereby ensuring the accuracy of the indexing of the various sub-target vehicle data in the target database, thereby optimizing the query efficiency.

[0259] Then, the target vehicle data is input into the fault determination model to determine whether the target vehicle has a fault, thereby ensuring the accuracy of the result of determining whether the target vehicle has a fault. If the target vehicle has a fault, each top event in the target fault tree network structure is traversed; the correlation between the intermediate events and bottom events corresponding to each top event and the target vehicle data is detected, thereby ensuring the accuracy of the obtained correlation. Based on the correlation, the target intermediate event and / or the target bottom event are determined, thereby ensuring the accuracy of the determined target intermediate event and / or the target bottom event. Then, the logical relationship between each target intermediate event and / or each target bottom event is obtained; based on the logical relationship, the preliminary fault type corresponding to the target vehicle is determined, thereby ensuring the accuracy of the preliminary fault type corresponding to the determined target vehicle. Then, based on the preliminary fault type, fault data related to the preliminary fault type is extracted from the target vehicle data; feature extraction is performed on the fault data to generate target fault features; the target fault features are matched with predetermined fault features in a predefined fault mode library, and the target fault type corresponding to the target vehicle is determined based on the matching results. Further refinement of fault diagnosis is achieved, and the accuracy of diagnosis is improved through comparison at the feature level.

[0260] Among them, the training process of the preset fault diagnosis model includes: obtaining historical fault data and historical normal data; generating fault training data based on fault case data in the historical fault data; generating normal training data based on historical normal data; generating a first training data set based on the fault training data and the normal training data, thereby ensuring the accuracy of the generated first training data set. Selecting a radial basis kernel function; constructing an optimization problem; using the Lagrange multiplier method to deform the optimization problem to obtain a Lagrangian function, thereby ensuring the accuracy of the obtained Lagrangian function. Then, solving the Lagrangian function through the KKT condition to obtain a solution result, thereby ensuring the accuracy of the obtained solution result. Based on the solution result, generating an initial fault determination network, thereby ensuring the accuracy of the generated initial fault determination network. Then, based on the first training data set, cross-validating the initial fault determination network, determining the model performance corresponding to the initial fault determination network, thereby ensuring the accuracy of the model performance corresponding to the determined initial fault determination network. Adjusting the parameters of the initial fault determination network according to the model performance to obtain a fault determination model, thereby ensuring the accuracy of the obtained fault determination model.

[0261] Then. Based on the historical fault data, the top event, the middle event and the bottom event are determined, and the top event, the middle event and the bottom event are connected by logic gates to form a fault tree, thereby ensuring the accuracy of the generated fault tree. Based on the historical fault data, a second training data set is generated. A radial basis kernel function is selected; an optimization problem is constructed; the Lagrangian multiplier method is used to deform the optimization problem to obtain the Lagrangian function, thereby ensuring the accuracy of the obtained Lagrangian function. Then, the Lagrangian function is solved by the KKT condition to obtain the solution result, thereby ensuring the accuracy of the obtained solution result. Based on the solution result, an initial fault determination network is generated, thereby ensuring the accuracy of the generated initial fault determination network. Then, based on the first training data set, the initial fault determination network is cross-validated to determine the model performance corresponding to the initial fault determination network, thereby ensuring the accuracy of the model performance corresponding to the determined initial fault determination network. The parameters of the initial fault determination network are adjusted according to the model performance to obtain a fault determination model, thereby ensuring the accuracy of the obtained fault determination model.

[0262] In addition, if the detection result is that the target fault type is consistent with the backup fault type, the target fault type is displayed to the user in a graphical interface, and a preset operation strategy corresponding to the target fault type is generated, and a control instruction is generated based on the operation strategy based on the result that the target fault type is consistent with the backup fault type; if the detection result is that the target fault type is inconsistent with the backup fault type, the target fault type is displayed to the user in a graphical interface, and a preset operation strategy corresponding to the target fault type is generated; a control instruction is generated based on the operation strategy based on the result that the target fault type is inconsistent with the backup fault type, thereby ensuring the accuracy of the generated control instruction.

[0263] Monitor the execution results of the operation strategy corresponding to the target fault type executed by the target vehicle, record the target fault type corresponding to the target vehicle, the time when the target fault type occurs, and the execution result, so as to trace the fault of the target vehicle according to the recorded target fault type corresponding to the target vehicle, the time when the target fault type occurs, and the execution result.

[0264] In the embodiment of the present application, by combining fault tree analysis (FTA) with support vector machine (SVM) algorithm, significant advantages are brought to the remote diagnosis and maintenance system of new energy vehicles integrated with QNX platform. The FTA algorithm helps the system understand the propagation path and root cause of the fault by constructing a logical relationship diagram of the fault, while the SVM algorithm uses its powerful classification ability to efficiently process a large amount of high-dimensional data to accurately identify the fault mode. This fusion not only significantly improves the accuracy of fault diagnosis and enhances the fault prediction ability of the system, but also optimizes the maintenance decision-making process, thereby improving the reliability of new energy vehicles, reducing maintenance costs, and improving user experience. In addition, predictive maintenance reduces sudden failures and improves the efficiency of vehicle use. At the same time, in-depth data analysis also provides valuable data support for vehicle design and performance optimization, further promoting the development of intelligent and networked new energy vehicles.

[0265] In addition, efficient data management ensures the persistent storage and rapid retrieval of massive vehicle data. The remote control module allows technicians to perform maintenance operations remotely, greatly improving the service response speed and convenience; the user interface module uses a friendly interactive design to enable car owners and maintenance personnel to easily obtain vehicle status and diagnostic information, improving the user experience. The above method ensures the stability and reliability of the system, and at the same time improves the security and operation efficiency of the system through permission management and performance monitoring. It not only improves the intelligence level of new energy vehicles, but also reduces maintenance costs, enhances the safety performance of vehicles, and ultimately improves the market competitiveness of car companies and consumer satisfaction.

[0266] In this embodiment, a fault diagnosis method is provided, which can be applied to a target vehicle. Figure 3 is a flow chart of a fault diagnosis method according to an embodiment of the present invention. Figure 3 As shown, the process includes the following steps:

[0267] Step S301, collecting initial vehicle data corresponding to the target vehicle.

[0268] Specifically, the target vehicle collects battery cell data, motor data, transmission system data and battery environment data corresponding to the target vehicle based on various sensors installed on the battery cell, motor, transmission system and battery environment.

[0269] Among them, the battery cell data includes battery voltage data, battery current data, battery temperature data, battery internal resistance data, battery remaining capacity (SOC) data, battery health status (SOH) data, etc. The motor data includes motor speed data, motor torque data, motor temperature data, motor current data, motor vibration data, and motor efficiency data, etc. The transmission system data includes transmission system speed data, transmission system torque data, transmission system oil temperature data, transmission system oil pressure data, transmission system vibration data, transmission system clearance data, etc. The battery environment data includes humidity data of the battery surrounding environment, temperature data of the battery surrounding environment, etc. The embodiment of the present application does not specifically limit the initial vehicle data.

[0270] Step S302, performing data fusion and cleaning processing on the initial vehicle data to obtain processed target vehicle data.

[0271] Specifically, the target vehicle can perform time synchronization processing on the initial vehicle data collected by each sensor, and then perform calculations according to the preset data fusion method. If the weighted average method is used, the initial vehicle data after time synchronization is fused and calculated according to the predetermined weights; if the Kalman filter method is used, the filter parameters are initialized according to the dynamic model and measurement model of the system, and then the initial vehicle data after time synchronization is input in sequence for iterative fusion calculation; for the fuzzy logic reasoning method, it is necessary to first define fuzzy sets and fuzzy rules, fuzzify the sensor data, and then perform fusion calculations.

[0272] Then, the target vehicle processes the outliers and missing values ​​of the fused initial vehicle data to obtain the processed target vehicle data.

[0273] Step S303: comparing each type of sub-target vehicle data in the target vehicle data with the data threshold corresponding to the sub-target vehicle data.

[0274] Specifically, the target vehicle can identify the target vehicle data, and classify the target vehicle data according to the identification result to obtain various types of sub-target vehicle data.

[0275] Exemplarily, the target vehicle data may include battery cell data, motor data, transmission system data and battery environment data corresponding to the target vehicle. Among them, the battery cell data includes battery voltage data, battery current data, battery temperature data, battery internal resistance data, battery remaining capacity (SOC) data, battery health status (SOH) data, etc. The motor data includes motor speed data, motor torque data, motor temperature data, motor current data, motor vibration data and motor efficiency data, etc. The transmission system data includes transmission system speed data, transmission system torque data, transmission system oil temperature data, transmission system oil pressure data, transmission system vibration data, transmission system clearance data, etc. The battery environment data includes humidity data of the battery surrounding environment, temperature data of the battery surrounding environment, etc. The embodiment of the present application does not specifically limit the target vehicle data.

[0276] The target vehicle can receive data thresholds corresponding to various types of sub-target vehicle data input by the user, or set data thresholds corresponding to various types of sub-target vehicle data according to the actual situation of the target vehicle. The embodiment of the present application does not specifically limit the data thresholds corresponding to various types of sub-target vehicle data.

[0277] Then, the target vehicle compares the data of each type of sub-target vehicle with the data threshold corresponding to the sub-target vehicle data.

[0278] Step S304: if there is at least one type of sub-target vehicle data that is greater than the corresponding data threshold, fault code information is generated based on the sub-target vehicle data that is greater than the corresponding data threshold.

[0279] Specifically, if there is at least one type of sub-target vehicle data that is greater than a corresponding data threshold, fault code information is generated based on the sub-target vehicle data that is greater than the corresponding data threshold.

[0280] Step S305, encrypting the target vehicle data and the fault code information to generate an encrypted data packet; and transmitting the encrypted data packet to the server.

[0281] The server is any one of the servers in the above implementations.

[0282] Specifically, the target vehicle may use a preset encryption algorithm to encrypt the target vehicle data and fault code information to generate an encrypted data packet.

[0283] Among them, the preset encryption algorithm can be a symmetric encryption algorithm, an asymmetric encryption algorithm, etc., among which the symmetric encryption algorithm can be an AES algorithm, and the asymmetric encryption algorithm can be an RSA algorithm.

[0284] The target vehicle then transmits the encrypted data packet to the server.

[0285] The fault diagnosis method provided in the embodiment of the present application collects the initial vehicle data corresponding to the target vehicle; performs data fusion and cleaning processing on the initial vehicle data to obtain the processed target vehicle data, thereby ensuring the accuracy of the obtained target vehicle data. The various sub-target vehicle data in the target vehicle data are compared with the data threshold corresponding to the sub-target vehicle data, so that the various sub-target vehicle data can be detected. If there is at least one type of sub-target vehicle data greater than the corresponding data threshold, fault code information is generated based on the sub-target vehicle data greater than the corresponding data threshold, thereby ensuring the accuracy of the generated fault code information. The target vehicle data and the fault code information are encrypted to generate an encrypted data packet; and the encrypted data packet is transmitted to the server, so that the server can obtain the target vehicle data and the fault code information corresponding to the target vehicle data, and then determine the target fault type corresponding to the target vehicle based on the target vehicle data and the fault code information corresponding to the target vehicle data.

[0286] In the embodiment of the present application, the target vehicle ensures the timeliness and accuracy of the target vehicle data by real-time monitoring of the vehicle status, providing a reliable basis for subsequent analysis. Then, by encrypting and optimizing the data stream, the security and transmission efficiency of the data are guaranteed. The stability and reliability of the system are ensured, and the security and operation efficiency of the system are improved through authority management and performance monitoring. The integrated operation of these modules not only improves the intelligence level of new energy vehicles, but also reduces maintenance costs, enhances the safety performance of vehicles, and ultimately improves the market competitiveness of car companies and consumer satisfaction.

[0287] In the present embodiment, a fault diagnosis device is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware of a predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware is also possible and conceived.

[0288] This embodiment provides a fault diagnosis device, which is applied to a server, such as Figure 4 As shown, including:

[0289] The receiving module 401 is used to receive the target vehicle data and the fault code information corresponding to the target vehicle data sent by the target vehicle; the fault code information is generated after the target vehicle detects the target vehicle data;

[0290] A first determination module 402, configured to determine a backup fault type corresponding to the target vehicle based on the fault code information;

[0291] The second determination module 403 is used to identify the target vehicle data and determine the target fault type corresponding to the target vehicle based on a preset fault diagnosis model;

[0292] A detection module 404 is used to detect whether the target fault type is consistent with the backup fault type;

[0293] The generation module 405 is used to generate a control instruction based on the detection result and the target fault type, and send the control instruction to the target vehicle; wherein the control instruction includes a preset operation strategy corresponding to the target fault type.

[0294] This embodiment provides a fault diagnosis device, which is applied to a target vehicle, such as Figure 5 As shown, including:

[0295] The acquisition module 501 is used to acquire the initial vehicle data corresponding to the target vehicle;

[0296] The processing module 502 is used to perform data fusion and cleaning processing on the initial vehicle data to obtain processed target vehicle data;

[0297] A comparison module 503 is used to compare various sub-target vehicle data in the target vehicle data with data thresholds corresponding to the sub-target vehicle data;

[0298] A first generating module 504, configured to generate fault code information based on the sub-target vehicle data that is greater than the corresponding data threshold if there is at least one type of sub-target vehicle data that is greater than the corresponding data threshold;

[0299] The second generating module 505 is used to encrypt the target vehicle data and the fault code information to generate an encrypted data packet; and transmit the encrypted data packet to a server; the server is any server of the above-mentioned implementation mode.

[0300] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0301] The fault diagnosis device in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0302] The embodiment of the present invention also provides a server having the above Figure 4 The fault diagnosis device shown.

[0303] See also Figure 6 , Figure 6is a schematic diagram of the structure of a server provided by an optional embodiment of the present invention, such as Figure 6 As shown, the server includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. The various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process instructions executed in the server, including instructions stored in or on the memory to display graphical information of the GUI on an external input / output device (such as a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple servers can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 6 A processor 10 is taken as an example.

[0304] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0305] The memory 20 stores instructions executable by at least one processor 10, so that at least one processor 10 executes the method shown in the above embodiment.

[0306] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required by at least one function; the data storage area may store data created according to the use of the server, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the server via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0307] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0308] The server further comprises a communication interface 30 for the server to communicate with other devices or a communication network.

[0309] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0310] A part of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the existence of the computer program instruction in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc., and accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium accessible to the computer.

[0311] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A fault diagnosis method, characterized in that: Applied to a server, the method comprises: Receiving target vehicle data and fault code information corresponding to the target vehicle data sent by the target vehicle; the fault code information is generated after the target vehicle detects the target vehicle data; Based on the fault code information, determining a backup fault type corresponding to the target vehicle; Identifying the target vehicle data, and determining a target fault type corresponding to the target vehicle based on a preset fault diagnosis model; Detecting whether the target fault type is consistent with the backup fault type; Based on the detection result and the target fault type, a control instruction is generated and sent to the target vehicle; wherein the control instruction includes a preset operation strategy corresponding to the target fault type.

2. The method according to claim 1, characterized in that The preset fault diagnosis model includes a fault determination model and a fault type identification model; the identifying the target vehicle data and determining the target fault type corresponding to the target vehicle based on the preset fault diagnosis model includes: Inputting the target vehicle data into the fault determination model to determine whether the target vehicle has a fault; If the target vehicle fails, the target vehicle data is input into the fault type identification model to determine the target fault type corresponding to the target vehicle.

3. The method according to claim 2, characterized in that The fault type identification model includes a target fault tree network structure. If the target vehicle fails, the target vehicle data is input into the fault type identification model to determine the target fault type corresponding to the target vehicle, including: Traversing each top event in the target fault tree network structure; Detecting the correlation between the intermediate events and bottom events corresponding to each of the top events and the target vehicle data; Based on the correlation, determining a target intermediate event and / or a target bottom event; Acquire the logical relationship between each of the target intermediate events and / or each of the target bottom events; Determining a preliminary fault type corresponding to the target vehicle according to the logical relationship; Based on the preliminary fault type, extracting fault data related to the preliminary fault type from the target vehicle data; Extracting features from the fault data to generate target fault features; Matching the target fault signature with predetermined fault signatures in a predefined fault mode library; According to the matching result, the target fault type corresponding to the target vehicle is determined.

4. The method according to claim 1 or 2, characterized in that: The preset fault diagnosis model includes a fault determination model and a fault type identification model. The training process of the preset fault diagnosis model includes: Obtain historical fault data and historical normal data; Generate fault training data based on fault case data in historical fault data; Generate normal training data based on historical normal data; Generate a first training data set based on the fault training data and the normal training data; Based on the first training data set, training an initial fault determination network to generate the fault determination model; Based on the historical fault data, determine a top event, an intermediate event, and a bottom event, and use logic gates to connect the top event, the intermediate event, and the bottom event to form a fault tree; Based on the historical fault data, generating a second training data set, wherein the second training data set includes a fault type corresponding to each historical fault data; Based on the fault tree and the second training data set, an initial fault type identification network is trained to generate the fault type identification model.

5. The method according to claim 4, characterized in that The step of training an initial fault determination network based on the first training data set to generate the fault determination model includes: Select the radial basis kernel function; Constructing optimization problems; Using the Lagrange multiplier method, the optimization problem is deformed to obtain a Lagrange function; Solving the Lagrangian function by using the KKT condition to obtain a solution result; Based on the solution result, generating the initial fault determination network; Based on the first training data set, cross-validate the initial fault determination network to determine the model performance corresponding to the initial fault determination network; The parameters of the initial fault determination network are adjusted according to the model performance to obtain the fault determination model.

6. The method according to claim 4, characterized in that The step of training an initial fault type identification network based on the fault tree and the second training data set to generate the fault type identification model includes: According to the hierarchical structure and logical relationship of the fault tree, an initial fault type identification network adapted to the fault tree is generated; the initial fault type identification network is an initial fault tree network structure; generating a logic gate simulation unit in the initial fault type identification network according to the logic gates in the fault tree; The second training data is input into the initial fault type identification network, and the initial fault type identification network is trained based on a preset training algorithm to generate the fault type identification model; the fault type identification model includes a target fault tree network structure.

7. The method according to claim 1, characterized in that The receiving target vehicle data sent by the target vehicle and the fault code information corresponding to the target vehicle data includes: Receive the encrypted data packet sent by the target vehicle; Decrypting the encrypted data packet to obtain the target vehicle data and the fault code information corresponding to the target vehicle data; Verifying the target vehicle data and the fault code information; If the target vehicle data and the fault code information are successfully verified, identifying various sub-target vehicle data included in the target vehicle data; Determine the target database corresponding to each type of sub-target vehicle data according to the recognition results corresponding to each type of sub-target vehicle data; Each type of sub-target vehicle data is stored in the corresponding target database, and an index is established for each type of sub-target vehicle data in the target database.

8. The method according to claim 1, characterized in that The generating of a control instruction based on the detection result and the target fault type comprises: If the detection result is that the target fault type is consistent with the backup fault type, the target fault type is displayed to the user in a graphical interface, and a preset operation strategy corresponding to the target fault type is generated, and the control instruction is generated based on the operation strategy and the result that the target fault type is consistent with the backup fault type; If the detection result is that the target fault type is inconsistent with the backup fault type, the target fault type is displayed to the user in a graphical interface, and a preset operation strategy corresponding to the target fault type is generated; based on the operation strategy and the result that the target fault type is inconsistent with the backup fault type, the control instruction is generated.

9. The method according to claim 1, characterized in that: The method further comprises: Monitoring the execution result of the target vehicle executing the operation strategy corresponding to the target fault type; The target fault type corresponding to the target vehicle, the time when the target fault type occurs, and the execution result are recorded.

10. A fault diagnosis method, characterized in that: Applied to a target vehicle, the method comprises: Collecting initial vehicle data corresponding to the target vehicle; Performing data fusion and cleaning processing on the initial vehicle data to obtain processed target vehicle data; Comparing various sub-target vehicle data in the target vehicle data with data thresholds corresponding to the sub-target vehicle data; If there is at least one type of sub-target vehicle data that is greater than the corresponding data threshold, generating fault code information based on the sub-target vehicle data that is greater than the corresponding data threshold; The target vehicle data and the fault code information are encrypted to generate an encrypted data packet; and the encrypted data packet is transmitted to a server; the server is the server described in any one of claims 1-9.

Images