Vehicle fault identification method, device and equipment and storage medium

Abstract

The invention discloses a vehicle fault identification method, device and equipment and a storage medium, and the method comprises the steps: obtaining to-be-detected data which is vehicle operation data; converting the to-be-detected data into a to-be-detected image; inputting the to-be-detected image into a multi-feature information enhanced fault recognition model, extracting fault features from the to-be-detected image by the multi-feature information enhanced fault recognition model, and performing interactive fusion and feature classification recognition on the fault features to obtain a fault recognition result; wherein the fault features comprise local semantic relation features and global semantic relation features. According to the method, the running state of the vehicle is monitored in real time, and the running data of the vehicle is converted into the image, so that the fault identification process is quicker and more accurate. Meanwhile, various fault features including local semantic relation features and global semantic relation features are extracted from the image, the running state of the vehicle can be comprehensively understood, and therefore the accuracy and reliability of fault recognition are improved.

Description

Technical Field

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

[0002] Under the current global energy transformation and environmental protection background, the rapid development of new energy vehicles has become a key link in promoting green, environmentally friendly and sustainable development strategies in various countries. However, although new energy vehicles have achieved remarkable results in environmental protection and energy saving, their safety issues cannot be ignored. In particular, the complex electronic control system and battery management system inside new energy vehicles make fault monitoring and identification a technical problem that needs to be solved urgently.

[0003] At present, fault identification of new energy vehicles mainly relies on the manufacturer's after-sales personnel to use fault diagnostic instruments to check and repair the vehicle communication components. This traditional fault identification method has obvious limitations: on the one hand, it can only be repaired offline after the fault occurs, and it is impossible to achieve real-time monitoring of the vehicle's operating status; on the other hand, due to the high integration of the electronic control system and battery management system of new energy vehicles, the fault point is difficult to accurately locate, and the fault types are diverse, which seriously affects the efficiency and accuracy of fault identification.

[0004] Therefore, how to develop a method that can monitor the vehicle's operating status in real time and identify vehicle faults efficiently and accurately has become a major challenge facing the current field of fault diagnosis. Summary of the invention

[0005] In view of the above problems, the present invention is proposed to provide a vehicle fault identification method, device, equipment and storage medium that overcomes the above problems or at least partially solves the above problems, and has the remarkable characteristics of real-time monitoring, high efficiency and accuracy, comprehensiveness and reliability.

[0006] In a first aspect, a vehicle fault identification method is provided, the vehicle fault identification method comprising:

[0007] Acquiring data to be detected, wherein the data to be detected is vehicle operation data;

[0008] Converting the data to be detected into an image to be detected;

[0009] The image to be detected is input into a multi-feature information enhanced fault recognition model, the multi-feature information enhanced fault recognition model extracts fault features from the image to be detected, and interactively fuses and classifies the fault features to obtain a fault recognition result;

[0010] The fault features include local semantic relationship features and global semantic relationship features.

[0011] Optionally, before acquiring the data to be detected, the vehicle fault identification method further includes:

[0012] Obtain vehicle historical operation data;

[0013] Building a data set based on the historical operation data, the data set includes a training set and a test set;

[0014] Based on the training set, a pre-built multi-feature information enhanced fault recognition network is trained to obtain the multi-feature information enhanced fault recognition model;

[0015] Based on the test set, the trained multi-feature information enhanced fault recognition model is tested to obtain the multi-feature information enhanced fault recognition model that meets the preset loss function requirements.

[0016] Optionally, constructing a data set based on the historical fault data includes:

[0017] The historical operation data is annotated according to the fault type to obtain annotated data;

[0018] Preprocessing the labeled data, and converting the preprocessed labeled data into a labeled image;

[0019] A sample set is constructed based on the annotated image, and the sample set includes a training set and a test set divided according to a preset ratio.

[0020] Optionally, the multi-feature information enhanced fault identification network includes:

[0021] A local semantic relationship feature extraction module, used to extract local semantic relationship features from the annotated images of the training set, wherein the local semantic relationship feature extraction module adopts a convolutional neural network (CNN) architecture;

[0022] A global semantic relationship feature extraction module, used to extract global semantic relationship features from the annotated images of the training set, wherein the global semantic relationship feature extraction module adopts a Transformer architecture;

[0023] A feature interaction fusion module, used for performing a splicing operation on the local semantic relationship feature and the global semantic relationship feature along the channel dimension to obtain a fused feature;

[0024] The feature classification and identification module is used to classify and identify fault features of the fused features to obtain and output fault identification results.

[0025] Optionally, the image to be detected is a time-frequency spectrum diagram or a vibration signal diagram.

[0026] Optionally, before converting the data to be detected into an image to be detected, the vehicle fault identification method further includes:

[0027] The data to be detected is interpolated, normalized and denoised.

[0028] In a second aspect, a vehicle fault identification device is provided, the vehicle fault identification device comprising:

[0029] An acquisition module, used for acquiring data to be detected, wherein the data to be detected is vehicle operation data;

[0030] A conversion module, used for converting the data to be detected into an image to be detected;

[0031] A recognition module, used for inputting the image to be detected into a multi-feature information enhanced fault recognition model, extracting fault features from the image to be detected by the multi-feature information enhanced fault recognition model, and interactively fusing and classifying the fault features to obtain a fault classification recognition result;

[0032] The fault features include local semantic relationship features and global semantic relationship features.

[0033] Optionally, the vehicle fault identification device further includes a model building module, which is used to:

[0034] Obtain vehicle historical fault data;

[0035] Building a data set based on the historical fault data, the data set comprising a training set and a test set;

[0036] Based on the training set, a pre-built multi-feature information enhanced fault recognition network is trained to obtain the multi-feature information enhanced fault recognition model;

[0037] Based on the test set, the trained multi-feature information enhanced fault recognition model is tested to obtain the multi-feature information enhanced fault recognition model that meets the preset loss function requirements.

[0038] Optionally, the model building module is further used to:

[0039] The historical operation data is annotated according to the fault type to obtain annotated data;

[0040] Preprocessing the labeled data, and converting the preprocessed labeled data into a labeled image;

[0041] A sample set is constructed based on the annotated image, and the sample set includes a training set and a test set divided according to a preset ratio.

[0042] Optionally, the multi-feature information enhanced fault identification network includes:

[0043] A local semantic relationship feature extraction module, used to extract local semantic relationship features from the annotated images of the training set, wherein the local semantic relationship feature extraction module adopts a convolutional neural network (CNN) architecture;

[0044] A global semantic relationship feature extraction module, used to extract global semantic relationship features from the annotated images of the training set, wherein the global semantic relationship feature extraction module adopts a Transformer architecture;

[0045] A feature interaction fusion module, used for performing a splicing operation on the local semantic relationship feature and the global semantic relationship feature along the channel dimension to obtain a fused feature;

[0046] The feature classification and identification module is used to classify and identify fault features of the fused features to obtain and output fault identification results.

[0047] Optionally, the image to be detected is a time-frequency spectrum diagram or a vibration signal diagram.

[0048] In a third aspect, an electronic device is provided, comprising: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the vehicle fault identification method as described in the first aspect by executing the computer instructions.

[0049] In a fourth aspect, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the vehicle fault identification method as described in the first aspect.

[0050] The technical solution provided in the embodiments of the present invention has at least the following technical effects or advantages:

[0051] The embodiment of the present invention provides a vehicle fault identification method, device, equipment and storage medium. First, by acquiring vehicle operation data and converting it into an image to be detected, real-time monitoring of the vehicle operation status is achieved. This conversion process not only makes the data more intuitive and easy to understand, but also facilitates the subsequent fault feature extraction and identification. By real-time monitoring of vehicle operation data, abnormal conditions in vehicle operation can be discovered in time, providing strong support for fault warning and timely processing. Secondly, the present invention adopts a multi-feature information enhanced fault identification model, which can extract multiple fault features including local semantic relationship features and global semantic relationship features from the image to be detected. The local semantic relationship features can capture the detailed information related to a specific fault in the vehicle operation data, while the global semantic relationship features can reflect the change trend of the overall operation status of the vehicle. By extracting and fusing these fault features, the present invention can more comprehensively understand the operation status of the vehicle, thereby improving the accuracy and reliability of fault identification. In addition, the multi-feature information enhanced fault identification model in the present invention also has powerful feature interaction fusion and feature classification recognition capabilities. The model can efficiently interact and fuse the extracted fault features, further enhancing the model's understanding and processing capabilities of fault information. At the same time, the model can also accurately classify and identify the fused features, quickly determine the type and severity of the fault, and provide an important basis for subsequent fault handling and maintenance. In summary, the vehicle fault identification method provided by the present invention has the remarkable characteristics of real-time monitoring, high efficiency and accuracy, and comprehensive reliability. This method not only solves the major challenges currently faced in the field of fault diagnosis, but also provides a strong guarantee for the safe operation and timely maintenance of vehicles.

[0052] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0054] Figure 1 is a flow chart of a vehicle fault identification method provided by an embodiment of the present invention;

[0055] Figure 2 is a flow chart of another vehicle fault identification method provided by an embodiment of the present invention;

[0056] Figure 3 It is a structural diagram of a multi-feature information enhanced fault identification network provided by an embodiment of the present invention;

[0057] Figure 4 It is a structural block diagram of a vehicle fault identification device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0058] In order to better understand the above-mentioned technical scheme, the above-mentioned technical scheme will be described in detail below in combination with the accompanying drawings and specific implementation methods of the specification. It should be understood that the embodiments of the present disclosure and the specific features in the embodiments are detailed descriptions of the technical scheme of the present application, rather than limitations on the technical scheme of the present application. In the absence of conflict, the embodiments of the present application and the technical features in the embodiments can be combined with each other.

[0059] Figure 1 is a flow chart of a vehicle fault identification method provided by an embodiment of the present invention, such as Figure 1 As shown, the vehicle fault identification method includes:

[0060] Step S110, obtaining the data to be detected, where the data to be detected is vehicle operation data.

[0061] In this embodiment, vehicle operation data can be acquired at fixed time intervals. The vehicle operation data includes but is not limited to vehicle driving parameters (such as vehicle speed, mileage, steering angle, etc.), engine-related parameters (such as engine temperature, speed, continuous working time, etc.), electrical system parameters (power battery voltage, motor operating status, etc.), braking system parameters (such as ABS working status, etc.) and other parameters (such as the operating status of various auxiliary equipment on the vehicle such as lighting system, audio system, air-conditioning system, etc.).

[0062] Optionally, after executing step S110 and before executing step S120, the method may further include:

[0063] The data to be detected is interpolated, normalized and denoised.

[0064] By performing the above preprocessing on multiple monitoring data, the quality, completeness and availability of the data can be significantly improved, while enhancing the performance, accuracy and robustness of the model.

[0065] Step S120: convert the data to be detected into an image to be detected.

[0066] By converting the vehicle fault data to be detected into fault images, this conversion process not only makes the data to be detected more intuitive and easy to understand, but also fully utilizes the advantages of computer vision algorithms to efficiently model the fault data. This visual processing method greatly improves the model's ability to retrieve fault data, making the fault identification process faster and more accurate.

[0067] In one implementation, the image to be detected is a time-frequency spectrum diagram or a vibration signal diagram. The time-frequency spectrum diagram and the vibration signal diagram can intuitively display the frequency and time distribution of the vehicle operation data, making the fault characteristics more clearly visible. This helps technicians or algorithm models to more accurately identify the type and location of the fault. Different fault types will show different characteristics on the time-frequency spectrum diagram and the vibration signal diagram. By analyzing these characteristics, the fault type can be accurately identified. By observing the changes in parameters such as the amplitude and frequency of the signal, the degree of the fault can be quantitatively evaluated, providing strong support for subsequent repairs and maintenance. The time-frequency spectrum diagram and the vibration signal diagram can monitor the operating status of the vehicle in real time. Once an abnormal signal is found, fault warning and diagnosis can be performed immediately, which helps to timely discover and handle faults and avoid accidents.

[0068] Step S130: input the image to be detected into the multi-feature information enhanced fault recognition model, extract fault features from the image to be detected by the multi-feature information enhanced fault recognition model, and interactively fuse and classify the fault features to obtain fault recognition results; wherein the fault features include local semantic relationship features and global semantic relationship features.

[0069] Among them, local semantic relationship features are usually related to specific areas or components in the image, and can capture the specific location of the fault or the abnormality of the component. By extracting these features, the model can locate the fault point more accurately and reduce the possibility of false positives and false negatives. The global semantic relationship features reflect the association information between the entire image or different regions, which helps the model understand the impact of the fault on the entire system or component, allowing the model to more comprehensively evaluate the fault situation and improve the accuracy of recognition.

[0070] Optionally, before executing step S110 to step S130, the vehicle fault identification method further includes a process of constructing a multi-feature information enhanced fault identification model. Figure 2 is another flow chart of a vehicle fault identification method provided by an embodiment of the present invention. Figure 2 As shown, the method includes:

[0071] Step S210: Obtain historical vehicle operation data.

[0072] The historical vehicle operation data may be vehicle operation data that can reflect different vehicle faults.

[0073] Step S220: construct a data set based on historical fault data, where the data set includes a training set and a test set.

[0074] In one implementation, step S220 may include:

[0075] The historical operation data are labeled according to the fault type to obtain labeled data; the labeled data are preprocessed and converted into labeled images; a sample set is constructed based on the labeled images, and the sample set includes a training set and a test set divided according to a preset ratio.

[0076] Exemplarily, the fault type may match the type of historical operating data. For example, when the historical operating data is engine-related data, the fault type may be engine-related faults, including engine overheating, engine jitter, etc. When the historical operating data is transmission system-related data, the fault type may be clutch slippage, transmission gear wear, and other transmission system-related faults. The annotated image may be a time-frequency spectrum diagram or a vibration signal diagram, which is the same as the image type of the image to be detected in step S120. The division ratio of the training set and the test set may be, for example, 8:2, etc., which is not limited in this embodiment.

[0077] Step S230: Based on the training set, the pre-constructed multi-feature information enhanced fault recognition network is trained to obtain a multi-feature information enhanced fault recognition model.

[0078] Figure 3 is a schematic diagram of a multi-feature information enhanced fault identification network provided by an embodiment of the present invention, such as Figure 3 As shown, in this embodiment, the multi-feature information enhanced fault identification network includes: a local semantic relationship feature extraction module 310, a global semantic relationship feature extraction module 320, a feature interaction fusion module 330 and a feature classification identification module 340.

[0079] The local semantic relationship feature extraction module 310 is used to extract local semantic relationship features from the annotated images of the training set, and the local semantic relationship feature extraction module 310 adopts the convolutional neural network CNN architecture. The global semantic relationship feature extraction module 320 is used to extract global semantic relationship features from the annotated images of the training set, and the global semantic relationship feature extraction module 320 adopts the Transformer architecture. The feature interaction fusion module 330 is used to splice the local semantic relationship features and the global semantic relationship features along the channel dimension to obtain fused features. The feature classification and identification module 340 is used to classify and identify fault features of the fused features to obtain and output the fault identification results.

[0080] By extracting local and global features simultaneously, the model can learn more diverse fault manifestations, thereby enhancing its generalization ability in different scenarios and conditions. The multi-feature information enhanced fault recognition network simultaneously utilizes the local feature extraction capability of CNN and the global information capture capability of Transformer, which can improve the performance of the trained model in a variety of computer vision tasks.

[0081] Specifically, the local semantic relationship feature extraction module 310 includes a reference submodule 311 and an attention feature enhancement submodule 312, wherein the reference submodule 311 is composed of a convolutional neural network (CNN) structure, and is used to capture local features in the image through convolution operations from the input image, thereby ensuring the effective extraction of basic fault features; the attention feature enhancement submodule 312 is used to mine more significant effective information from the fault features, thereby significantly improving the accuracy and pertinence of fault feature extraction. The global semantic relationship feature extraction module 320 includes a long-range relationship modeling submodule 321, wherein the long-range relationship modeling submodule 321 is composed of a Transformer structure, and is used to capture global relationships in images or data through a self-attention mechanism, thereby achieving a comprehensive understanding of the overall state of the vehicle. The multi-feature information enhanced fault recognition network simultaneously utilizes the local feature extraction capability of CNN and the global information capture capability of Transformer to collaboratively extract the fault features of the vehicle from different levels, thereby improving the effective information available for model construction.

[0082] It should be noted that the local semantic relationship feature extraction module 310 and the global semantic relationship feature extraction module 320 in this embodiment respectively adopt different attention mechanisms. This design enables the model to focus on the key areas and features in the image more flexibly, effectively enhancing the model's ability to learn significant information in fault features. This not only improves the accuracy of the model's fault identification, but also makes the model more adaptable and robust when facing complex and changeable vehicle fault conditions.

[0083] In a specific implementation, step S230 may specifically include the following steps:

[0084] The annotated images of the training set are input into the multi-feature information enhanced fault recognition network, and the local semantic relationship features and the global semantic relationship features are extracted in parallel by the local semantic relationship feature extraction module 310 and the global semantic relationship feature extraction module 320; the local semantic relationship features and the global semantic relationship features are concatenated along the channel dimension by the feature interaction fusion module 330 to obtain fused features; the feature classification recognition module 340 identifies the fault category of the fused features to obtain and output the fault recognition results.

[0085] In one implementation, the local semantic relationship feature extraction module 310 may extract the local semantic relationship feature by:

[0086] The first step is to label the image x∈R in the training set C×H×W Input to the benchmark submodule 311 to obtain the feature map Where C is the number of input channels, C1 is the output channel, H and W are the height and width of the local feature;

[0087] The calculation formula is as follows:

[0088] l x =RL(Conv 3×3 (x)+Conv 5×5 (x)+Conv 7×7 (x)); (1)

[0089] Where Conv(·) represents the convolution operation and RL(·) represents the ReLU activation function.

[0090] The second step is to output features Input the attention feature enhancement submodule 312 to obtain the feature map after attention weighted calculation Where C2 is the output channel, H and W are the height and width of the attention feature;

[0091] The calculation formulas are as follows: Formulas (2) and (3):

[0092] o=l x ×θ(RL(Conv 1×1 (l x ))); (2)

[0093]

[0094] Among them, Conv(·) represents the convolution operation, RL(·) represents the ReLU activation function, and θ(·) represents the Sigmoid function. Indicates accumulation operation;

[0095] In one implementation, the global semantic relationship feature extraction module 320 extracts the global semantic relationship feature, which may specifically include the following implementations:

[0096] The labeled images x∈R in the training set C×H×W Input to the long-range relationship modeling submodule 321, and obtain the global semantic relationship features through feature modeling C is the number of input channels, C3 is the number of output channels of the convolutional layer, H and W are the height and width of the global feature;

[0097] The calculation formula is as follows:

[0098] f global =φ(RL(Conv 3×3 (x))); (4)

[0099] Among them, Conv(·) represents the convolution operation, RL(·) represents the RELU activation function, and φ(·) represents the Transformer architecture.

[0100] In one implementation, the feature interaction fusion module 330 extracts the local semantic relationship features. and global semantic relationship features Perform feature concatenation along the channel dimension to obtain the interactive fusion features Specifically, it can be achieved by the following formula (5):

[0101] output = cat(f local ,f global ); (5)

[0102] Among them, cat(·) represents the feature concatenation operation, f local represents the local semantic relationship feature, f global Represents the global semantic relationship features, and output represents the output features after information interaction.

[0103] In this embodiment, the cross entropy loss function can be used for training during the multi-feature information enhanced fault identification network training process, and the SGD optimization function is used to achieve overall optimization of the gradient. The initial learning rate is set to 0.01 and the momentum is set to 0.9. The loss function used can be:

[0104]

[0105] Among them, L ce is the cross entropy loss function, x i Represents input data;

[0106] Step S240: Based on the test set, the trained multi-feature information enhanced fault recognition model is tested to obtain a multi-feature information enhanced fault recognition model that meets the preset loss function requirements.

[0107] Among them, the loss function, also known as the cost function, objective function or error function, is an indicator used to measure the difference between the model's predicted value and the actual observed value. In supervised learning tasks, the loss function calculates the degree of inconsistency between the model's predicted result for each sample and the true label. Through the supervised training of the entire network by the loss function, the model can achieve efficient and accurate recognition of automobile faults.

[0108] In step S240, after the test set is input into the trained multi-feature information enhanced fault recognition model, the inconsistency between the fault recognition result of each annotated image and the fault recognition result annotated by the annotated image itself can be calculated by the loss function calculation model, so as to determine whether the multi-feature information enhanced fault recognition model meets the preset loss function requirements. If the preset loss function requirements are not met, the parameters of the multi-feature information enhanced fault recognition model, such as the number of iterations, the initial learning rate, etc., can be adjusted to gradually reduce the loss function and finally meet the preset loss function requirements.

[0109] It should be noted that the multi-feature information enhanced fault recognition model in step S130 can be constructed through the above steps S210 to S240, that is, the multi-feature information enhanced fault recognition model in step S130 also includes a local semantic relationship feature extraction module, a global semantic relationship feature extraction module, a feature interaction fusion module and a feature classification recognition module. The local semantic relationship feature extraction module and the global semantic relationship feature extraction module extract fault features, the feature interaction fusion module interactively fuses the fault features, and the feature classification recognition module classifies and recognizes fault features to obtain and output fault recognition results. The feature classification recognition module 340 is a classification layer, which achieves the final fault feature classification and recognition by matching with the annotated images in a given training set.

[0110] The multi-feature information enhanced fault identification model can monitor the vehicle operation status, obtain vehicle operation data in real time, automatically process the images to be detected, extract fault features and classify and identify them. This process does not require human intervention, which greatly improves the speed and efficiency of fault identification. In addition, this method can issue an early warning before a fault occurs, providing maintenance personnel with enough time to troubleshoot and repair the fault. This helps reduce vehicle downtime and maintenance costs caused by faults and improves vehicle operation efficiency and safety.

[0111] Based on the same inventive concept, an embodiment of the present invention further provides a vehicle fault identification device for implementing vehicle fault identification. Figure 4 is a structural block diagram of a vehicle fault identification device provided by an embodiment of the present invention, such as Figure 4 As shown, the vehicle fault identification device 400 includes:

[0112] The acquisition module 410 is used to acquire the data to be detected, where the data to be detected is the vehicle operation data;

[0113] A conversion module 420, used for converting the data to be detected into an image to be detected;

[0114] The recognition module 430 is used to input the image to be detected into the multi-feature information enhanced fault recognition model, extract the fault features from the image to be detected by the multi-feature information enhanced fault recognition model, and interactively fuse and classify the fault features to obtain a fault classification recognition result;

[0115] Among them, fault features include local semantic relationship features and global semantic relationship features.

[0116] Optionally, the vehicle fault identification device 400 further includes a model building module, which is used to:

[0117] Obtain vehicle historical fault data;

[0118] Build a data set based on historical fault data, which includes a training set and a test set;

[0119] Based on the training set, the pre-built multi-feature information enhanced fault recognition network is trained to obtain a multi-feature information enhanced fault recognition model;

[0120] Based on the test set, the trained multi-feature information enhanced fault recognition model is tested to obtain a multi-feature information enhanced fault recognition model that meets the requirements of the preset loss function.

[0121] Optionally, the model building module is also used to:

[0122] The historical operation data is annotated according to the fault type to obtain annotated data;

[0123] Preprocessing the labeled data, and converting the preprocessed labeled data into a labeled image;

[0124] A sample set is constructed based on the annotated images, and the sample set includes a training set and a test set divided according to a preset ratio.

[0125] Optionally, the multi-feature information enhanced fault identification network includes:

[0126] The local semantic relationship feature extraction module is used to extract local semantic relationship features from the annotated images of the training set. The local semantic relationship feature extraction module adopts the convolutional neural network (CNN) architecture.

[0127] The global semantic relationship feature extraction module is used to extract global semantic relationship features from the annotated images in the training set. The global semantic relationship feature extraction module adopts the Transformer architecture.

[0128] The feature interaction fusion module is used to concatenate local semantic relationship features and global semantic relationship features along the channel dimension to obtain fused features;

[0129] The feature classification and recognition module is used to classify and recognize fault features of the fused features to obtain and output fault recognition results.

[0130] Optionally, the image to be detected is a time-frequency spectrum diagram or a vibration signal diagram.

[0131] The specific details of the vehicle fault identification method involved in the above-mentioned device can be understood by corresponding to the relevant descriptions and effects in the above-mentioned vehicle fault identification method embodiment, and will not be repeated here.

[0132] Based on the same inventive concept as the above-mentioned vehicle fault identification method, the present invention also provides an electronic device, which may include a processor and a memory, wherein the processor and the memory may be connected to each other through a bus or other means. The processor may be a central processing unit (CPU). The processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components and other chips, or a combination of the above-mentioned various types of chips. The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs, non-transient computer executable programs and modules, such as program instructions / modules corresponding to the vehicle fault identification method in the embodiment of the present invention. The processor executes various functional applications and data processing of the processor by running the non-transient software programs, instructions and modules stored in the memory, that is, the vehicle fault identification in the above-mentioned method embodiment is realized.

[0133] The memory may include a program storage area and a data storage area, wherein the program storage area may store an operating system and an application required by at least one function; the data storage area may store data created by the processor, etc. In addition, the memory may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. The one or more modules are stored in the memory, and when executed by the processor, the execution is as follows: Figure 1 and 2 The vehicle fault identification method in the illustrated embodiment.

[0134] For details of the above electronic equipment, please refer to Figure 1 and 2 The corresponding related descriptions and effects in the illustrated embodiments can be understood and will not be repeated here.

[0135] Based on the same inventive concept as the vehicle fault identification method, the present invention also provides a computer-readable storage medium, on which computer instructions are stored, and the computer program instructions are used to enable a computer to execute the vehicle fault identification method in the above embodiment.

[0136] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the storage medium can be a read-only memory (ROM), a random access memory (RAM), a flash memory, a hard disk drive (HDD) or a solid-state drive (SSD), etc.; the storage medium can also include a combination of the above-mentioned types of memory.

[0137] The technical solutions provided in the above embodiments of the present application have at least the following technical effects or advantages:

[0138] The embodiment of the present invention provides a vehicle fault identification method, device, equipment and storage medium. First, by acquiring vehicle operation data and converting it into an image to be detected, real-time monitoring of the vehicle operation status is achieved. This conversion process not only makes the data more intuitive and easy to understand, but also facilitates the subsequent fault feature extraction and identification. By real-time monitoring of vehicle operation data, abnormal conditions in vehicle operation can be discovered in time, providing strong support for fault warning and timely processing. Secondly, the present invention adopts a multi-feature information enhanced fault identification model, which can extract multiple fault features including local semantic relationship features and global semantic relationship features from the image to be detected. The local semantic relationship features can capture the detailed information related to a specific fault in the vehicle operation data, while the global semantic relationship features can reflect the change trend of the overall operation status of the vehicle. By extracting and fusing these fault features, the present invention can more comprehensively understand the operation status of the vehicle, thereby improving the accuracy and reliability of fault identification. In addition, the multi-feature information enhanced fault identification model in the present invention also has powerful feature interaction fusion and feature classification recognition capabilities. The model can efficiently interact and fuse the extracted fault features, further enhancing the model's understanding and processing capabilities of fault information. At the same time, the model can also accurately classify and identify the fused features, quickly determine the type and severity of the fault, and provide an important basis for subsequent fault handling and maintenance. In summary, the vehicle fault identification method provided by the present invention has the remarkable characteristics of real-time monitoring, high efficiency and accuracy, and comprehensive reliability. This method not only solves the major challenges currently faced in the field of fault diagnosis, but also provides a strong guarantee for the safe operation and timely maintenance of vehicles.

[0139] In the description provided herein, a large number of specific details are described. However, it is understood that embodiments of the present invention can be practiced without these specific details. In some instances, well-known methods, structures and techniques are not shown in detail so as not to obscure the understanding of this description.

[0140] Similarly, it should be understood that in order to streamline the present disclosure and aid in understanding one or more of the various inventive aspects, in the above description of exemplary embodiments of the present invention, various features of the present invention are sometimes grouped together into a single embodiment, figure, or description thereof. However, this disclosed method should not be interpreted as reflecting the intention that the claimed invention requires more features than those explicitly recited in each claim. More specifically, as reflected in the claims below, inventive aspects lie in less than all the features of the individual embodiments disclosed above. Therefore, the claims that follow the specific embodiment are hereby expressly incorporated into the specific embodiment, with each claim itself serving as a separate embodiment of the present invention.

[0141] It should be noted that the above-mentioned embodiments illustrate rather than limit the invention and that those skilled in the art will be able to design alternative embodiments without departing from the scope of the appended claims.

Claims

1. A vehicle fault identification method, characterized in that: The vehicle fault identification method comprises: Acquiring data to be detected, wherein the data to be detected is vehicle operation data; Converting the data to be detected into an image to be detected; The image to be detected is input into a multi-feature information enhanced fault recognition model, the multi-feature information enhanced fault recognition model extracts fault features from the image to be detected, and interactively fuses and classifies the fault features to obtain a fault recognition result; The fault features include local semantic relationship features and global semantic relationship features.

2. The vehicle fault identification method according to claim 1, characterized in that: Before obtaining the data to be detected, the vehicle fault identification method further includes: Obtain vehicle historical operation data; Building a data set based on the historical operation data, the data set includes a training set and a test set; Based on the training set, a pre-built multi-feature information enhanced fault recognition network is trained to obtain the multi-feature information enhanced fault recognition model; Based on the test set, the trained multi-feature information enhanced fault recognition model is tested to obtain the multi-feature information enhanced fault recognition model that meets the preset loss function requirements.

3. The vehicle fault identification method according to claim 2, characterized in that: The step of constructing a data set based on the historical fault data includes: The historical operation data is annotated according to the fault type to obtain annotated data; Preprocessing the labeled data, and converting the preprocessed labeled data into a labeled image; A sample set is constructed based on the annotated image, and the sample set includes a training set and a test set divided according to a preset ratio.

4. The vehicle fault identification method according to claim 2, characterized in that: The multi-feature information enhanced fault identification network includes: A local semantic relationship feature extraction module, used to extract local semantic relationship features from the annotated images of the training set, wherein the local semantic relationship feature extraction module adopts a convolutional neural network (CNN) architecture; A global semantic relationship feature extraction module, used to extract global semantic relationship features from the annotated images of the training set, wherein the global semantic relationship feature extraction module adopts a Transformer architecture; A feature interaction fusion module, used for performing a splicing operation on the local semantic relationship feature and the global semantic relationship feature along the channel dimension to obtain a fused feature; The feature classification and identification module is used to classify and identify fault features of the fused features to obtain and output fault identification results.

5. The vehicle fault identification method according to claim 1, characterized in that: The image to be detected is a time-frequency spectrum diagram or a vibration signal diagram.

6. The vehicle fault identification method according to claim 1, characterized in that: Before converting the to-be-detected data into an to-be-detected image, the vehicle fault identification method further includes: The data to be detected is interpolated, normalized and denoised.

7. A vehicle fault identification device, characterized in that: The vehicle fault identification device comprises: An acquisition module, used for acquiring data to be detected, wherein the data to be detected is vehicle operation data; A conversion module, used for converting the data to be detected into an image to be detected; A recognition module, used for inputting the image to be detected into a multi-feature information enhanced fault recognition model, extracting fault features from the image to be detected by the multi-feature information enhanced fault recognition model, and interactively fusing and classifying the fault features to obtain a fault classification recognition result; The fault features include local semantic relationship features and global semantic relationship features.

8. The vehicle fault identification device according to claim 7, characterized in that: The vehicle fault identification device further includes a model building module, which is used to: Obtain vehicle historical fault data; Building a data set based on the historical fault data, the data set comprising a training set and a test set; Based on the training set, a pre-built multi-feature information enhanced fault recognition network is trained to obtain the multi-feature information enhanced fault recognition model; Based on the test set, the trained multi-feature information enhanced fault recognition model is tested to obtain the multi-feature information enhanced fault recognition model that meets the preset loss function requirements.

9. An electronic device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the vehicle fault identification method as described in any one of claims 1 to 6 by executing the computer instructions.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the vehicle fault identification method according to any one of claims 1 to 6.

Images