Vehicle fault diagnosis method based on artificial intelligence
Through the vehicle fault diagnosis method based on artificial intelligence, the cloud platform and artificial intelligence model are used to solve the problems of traditional slow diagnosis speed and low accuracy, and the rapid and accurate diagnosis and alarm of vehicle faults are achieved, reducing maintenance costs.
Patent Information
- Application Number
- CN202411973102.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-30
- Publication Date
- 2025-04-18
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
In the prior art, vehicle fault diagnosis speed is slow, low accuracy and high maintenance costs.
Using artificial intelligence-based vehicle fault diagnosis method, artificial intelligence models are built and trained by collecting historical vehicle fault operation data and fault type tags, deploying them on a cloud platform to identify vehicle faults in real time and generate alarm information.
It realizes rapid and accurate diagnosis of vehicle failures, reduces maintenance costs, and improves the accuracy of fault identification.
Smart Images
Figure CN119850185A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of vehicle fault detection, and in particular relates to a vehicle fault diagnosis method based on artificial intelligence. Background Art
[0002] The motor vehicle aftermarket is a relatively vague market. In fact, there is no clear definition. From the moment a consumer purchases a product at a car service station, it can be defined that the vehicle has entered the service scope of the automotive aftermarket. Because cars are also consumables, vehicles will inevitably have problems during daily use. The maintenance and repair of vehicles in the domestic automotive aftermarket must be carried out in professional auto repair shops and by very experienced auto repair staff to ensure the normal and safe driving of the vehicle, especially for special vehicles. Therefore, traditional vehicle fault diagnosis methods mainly rely on manual experience, and there are problems such as slow diagnosis speed, low accuracy, and high maintenance costs. Summary of the invention
[0003] The present invention provides a vehicle fault diagnosis method based on artificial intelligence, which is used to solve the problems of slow vehicle fault diagnosis speed, low accuracy and high maintenance cost in the prior art.
[0004] A vehicle fault diagnosis method based on artificial intelligence, comprising: Collect historical vehicle fault operation data and fault type labels corresponding to the historical vehicle fault operation data to obtain training sample data, and preprocess the training sample data to obtain preprocessed training sample data; Construct an initialization artificial intelligence model, and use the pre-processed training sample data as support to train the initialization artificial intelligence model using a synchronous intelligent optimization algorithm to obtain a vehicle fault diagnosis model; The vehicle fault diagnosis model is deployed on a cloud platform, real-time vehicle fault operation data is collected, and the real-time vehicle fault operation data is transmitted to the cloud platform for identification to obtain a vehicle fault diagnosis result.
[0005] Furthermore, after the real-time vehicle fault operation data is transmitted to the cloud platform for identification and the vehicle fault diagnosis result is obtained, it also includes: when the vehicle fault diagnosis result meets the preset alarm condition, fault alarm information is generated through the cloud platform, and the fault alarm information is transmitted to the user-specified device and the vehicle computer for alarm.
[0006] Furthermore, constructing an initialized artificial intelligence model includes: constructing a convolutional neural network model to obtain an initialized artificial intelligence model.
[0007] Furthermore, based on the pre-processed training sample data, a synchronous intelligent optimization algorithm is used to train the initialized artificial intelligence model to obtain a vehicle fault diagnosis model, including: Initializing the hyperparameters of the initialization artificial intelligence model, obtaining a population, and dividing individuals in the population into a first sub-population and a second sub-population; Based on the preprocessed training sample data, the fitness of each first individual in the first subpopulation and the fitness of each second individual in the second subpopulation are obtained; According to the fitness of each first individual in the first subpopulation, a first best individual and a first worst individual are obtained; according to the fitness of each second individual in the second subpopulation, a second best individual and a second worst individual are obtained; According to the first best individual and the first worst individual, an information interactive search is performed on a second individual in the second subpopulation to obtain a second individual after the information interactive search; According to the second best individual and the second worst individual, an information interaction search is performed on the first individual in the first subpopulation to obtain a first individual after the information interaction search; For the first body after the information interaction search, an adaptive spiral search strategy is used to perform a local encirclement search on the first body to obtain the first body after the local encirclement search; For the second individual after the information interaction search, the individual information interaction search strategy is used to perform a local expansion search on the second individual to obtain the second individual after the local expansion search; The first individual after the local encirclement search and the second individual after the local expansion search are taken as the target individual, and a random perturbation strategy is used to perform a global search on the target individual to obtain the target individual after the global search; Determine whether the training end conditions are met. If so, obtain the global optimal individual based on the target individual after the global search, and use the hyperparameters in the global optimal individual as the final hyperparameters to initialize the artificial intelligence model to obtain the vehicle fault diagnosis model. Otherwise, return to the step of obtaining fitness.
[0008] Further, based on the preprocessed training sample data, the fitness of each first individual in the first subpopulation and the fitness of each second individual in the second subpopulation are obtained, including: For each first individual in the first subpopulation or each second individual in the second subpopulation, applying the hyperparameters contained in the first individual or the second individual to the initialized artificial intelligence model; For the initialized artificial intelligence model to which hyperparameters are applied, the preprocessed historical vehicle fault operation data is used as the input of the initialized artificial intelligence model, and the fault type label corresponding to the historical vehicle fault operation data is used as the expected output of the initialized artificial intelligence model to obtain the cross entropy loss function value; After adding the cross entropy loss function value to the preset constant term, a non-zero numerical term is obtained, and the reciprocal of the non-zero numerical term is taken to obtain the fitness corresponding to the first individual or the second individual.
[0009] Further, according to the first best individual and the first worst individual, an information interaction search is performed on a second individual in the second subpopulation to obtain the second individual after the information interaction search, including: For the second individual in the second subpopulation, according to the first best individual and the first worst individual, the update speed corresponding to the second individual is obtained as follows: ;in, Indicates t During the training i The speed corresponding to the second individual, i =1,2,…,I, I represents the total number of the second individuals, Indicates t +1 training session i The speed corresponding to the second individual, Represents a random number between (0,1), represents the first optimal individual, represents the first worst individual, Represents a random number between (0,1), Indicates t During the training i a second individual; According to the update speed corresponding to the second individual, an information interaction search is performed on the second individual in the second subpopulation, and the second individual after the information interaction search is obtained as: ;in, Represents a random number between (0,1), Represents the second individual after information interaction search .
[0010] Further, according to the second best individual and the second worst individual, an information interaction search is performed on the first individual in the first subpopulation to obtain the first individual after the information interaction search, including: For the first individual in the first subpopulation, according to the second best individual and the second worst individual, the update speed corresponding to the first individual is obtained as follows: ;in, Indicates t During the trainingj The velocity corresponding to the first body, j =1,2,…,J, J represents the total number of the first individual, Indicates t +1 training session j The velocity corresponding to the first body, Represents a random number between (0,1), represents the second best individual, represents the second worst individual, Represents a random number between (0,1), Indicates t During the training j the first individual; According to the update speed corresponding to the first individual, an information interaction search is performed on the first individual in the first subpopulation, and the first individual after the information interaction search is obtained as: ;in, Represents a random number between (0,1), Indicates the first entity after information interaction search .
[0011] Furthermore, for the first body after the information interaction search, an adaptive spiral search strategy is used to perform a local encirclement search on the first body, and the first body after the local encirclement search is obtained, including: Based on the current number of training times, the adaptive inertia weight is obtained as: ;in, represents the adaptive inertia weight, represents the sine function, represents pi, Indicates the preset maximum number of training times. represents a constant term; For the first body after the information interaction search, a local encirclement search is performed on the first body according to the adaptive inertia weight, and the first body after the local encirclement search is obtained as follows: ;in, Indicates t During the training m The first body after the information interaction search, Represents the first volume after the local enclosing search , represents a natural constant, b represents the spiral search constant, l Represents a random number between [-1,1], represents the cosine function, Represents the circumference, represents the first optimal individual.
[0012] Furthermore, for the second individual after the information interaction search, the individual information interaction search strategy is used to perform a local extended search on the second individual, and the second individual after the local extended search is obtained, including: For the second individual after the information interaction search, obtain the mean of the second individual and obtain the central individual of the second population; According to the second optimal individual, the first local extended search item is obtained as: ;in, represents the first local expansion search term, Represents a random number between (0,1), Indicates t During the training n The second individual after the information interaction search, represents the second best individual, represents the local search weight; According to the central individual of the second population, the second local extended search item is obtained as: ,in, represents the second local expansion search term, Represents a random number between (0,1), represents the central individual, ; According to the first local extended search item and the second local extended search item, a local extended search is performed on the second individual, and the second individual after the local extended search is obtained as: ;in, Represents the second individual after the local expansion search .
[0013] Furthermore, the first individual after the local encirclement search and the second individual after the local expansion search are taken as target individuals, and a random perturbation strategy is used to perform a global search on the target individuals to obtain the target individuals after the global search, including: The first individual after the local encirclement search and the second individual after the local expansion search are taken as target individuals; For any target individual, the adaptive variation range is: ;in, represents the maximum variation range, Indicates k The variation range corresponding to each target individual is represents the maximum fitness among the target individuals, Indicates k The fitness corresponding to the target individual is represents a constant term; According to the adaptive variation range, a global search is performed on the target individual, and the target individual after the global search is obtained as follows: ;in, Indicates the tth training process k Target individuals No. d Dimension parameters, Represents the target individual after global search No. d Dimension parameters, represents the adaptive inertia weight, express A random number between Determine whether the fitness of the target individual after the global search has increased. If so, accept the global search. Otherwise, reject the global search and use the original target individual directly as the target individual after the global search.
[0014] The present invention provides an artificial intelligence-based vehicle fault diagnosis method, which uses an artificial intelligence model to learn the data relationship between historical vehicle fault operation data and the fault type labels corresponding to the historical vehicle fault operation data to obtain a vehicle fault diagnosis model, and then uses the vehicle fault diagnosis model to identify real-time vehicle fault operation data, thereby achieving rapid and accurate diagnosis of vehicle faults. At the same time, a synchronous intelligent optimization algorithm is used to train the initialized artificial intelligence model, thereby making data learning more accurate and ensuring the accuracy of vehicle fault identification. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.
[0016] Figure 1 A flowchart of a vehicle fault diagnosis method based on artificial intelligence is provided in an embodiment of the present invention.
[0017] The above drawings have shown clear embodiments of the present invention, which will be described in more detail below. These drawings and text descriptions are not intended to limit the scope of the present invention in any way, but to illustrate the concept of the present invention for those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION
[0018] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Instead, they are merely examples of devices and methods consistent with some aspects of the present invention as detailed in the appended claims.
[0019] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.
[0020] like Figure 1 As shown, an embodiment of the present invention provides a vehicle fault diagnosis method based on artificial intelligence, comprising: S101, collecting historical vehicle fault operation data and fault type labels corresponding to the historical vehicle fault operation data to obtain training sample data, and preprocessing the training sample data to obtain preprocessed training sample data; Historical vehicle fault operation data can be all the operation data on the vehicle, or it can be the operation data of a certain component. Since all the operation data on the vehicle are analyzed, there will be a lot of noise for fault classification, which may lead to a low fault accuracy rate, which is very unfavorable for fault identification of special vehicles. Therefore, the embodiment of the present invention uses the historical vehicle fault operation data of a single component and the fault type label corresponding to the historical vehicle fault operation data for data learning, so as to achieve rapid identification of faults of a single component. For example, the vibration time domain data or frequency domain data of the engine can be collected as the historical vehicle fault operation data, and then the actual fault type is used as the fault type label corresponding to the historical vehicle fault operation data.
[0021] The preprocessing may be to remove missing values and / or abnormal values in the historical vehicle fault operation data and the fault type labels corresponding to the historical vehicle fault operation data.
[0022] S102, constructing an initialization artificial intelligence model, and using the pre-processed training sample data as support, using a synchronous intelligent optimization algorithm to train the initialization artificial intelligence model to obtain a vehicle fault diagnosis model; The initialization artificial intelligence model can preferably be set to a convolutional neural network model, but it is worth noting that other artificial intelligence models can also be used as the initialization artificial intelligence model to achieve relationship recognition between historical vehicle fault operation data and the fault type labels corresponding to the historical vehicle fault operation data.
[0023] In the prior art, the gradient descent intelligent optimization algorithm is often used to train the artificial intelligence model. Although it can achieve a certain training effect, it is very easy to fall into the local optimum, resulting in a low final fault identification accuracy. Therefore, the embodiment of the present invention provides a synchronous intelligent optimization algorithm to improve the solution space traversal and global search capabilities, and ultimately improve the fault diagnosis accuracy, ensuring the normal and safe driving of the vehicle, especially ensuring the normal and safe driving of the characteristic vehicle.
[0024] S103, deploying the vehicle fault diagnosis model on a cloud platform, collecting real-time vehicle fault operation data, and transmitting the real-time vehicle fault operation data to the cloud platform for identification to obtain a vehicle fault diagnosis result.
[0025] Since the processing performance of the vehicle's computer is relatively low, if the vehicle fault diagnosis model is deployed on the vehicle's computer, it may cause the computer to get stuck or the recognition speed to be slow. Therefore, the embodiment of the present invention preferably deploys the vehicle fault diagnosis model on a cloud platform, utilizing the high data processing performance of the cloud platform and the convenience of the mobile network to improve the convenience of fault diagnosis.
[0026] The present invention provides an artificial intelligence-based vehicle fault diagnosis method, which uses an artificial intelligence model to learn the data relationship between historical vehicle fault operation data and the fault type labels corresponding to the historical vehicle fault operation data to obtain a vehicle fault diagnosis model, and then uses the vehicle fault diagnosis model to identify real-time vehicle fault operation data, thereby achieving rapid and accurate diagnosis of vehicle faults. At the same time, a synchronous intelligent optimization algorithm is used to train the initialized artificial intelligence model, thereby making data learning more accurate and ensuring the accuracy of vehicle fault identification.
[0027] In an embodiment of the present invention, after the real-time vehicle fault operation data is transmitted to the cloud platform for identification and the vehicle fault diagnosis result is obtained, it also includes: when the vehicle fault diagnosis result meets the preset alarm condition, fault alarm information is generated through the cloud platform, and the fault alarm information is transmitted to the user-specified device and the vehicle computer for alarm.
[0028] For example, the fault alarm information can be transmitted to the user's mobile phone or vehicle computer, so that the user can quickly troubleshoot and repair the vehicle fault, thereby ensuring the safe driving of the vehicle.
[0029] In an embodiment of the present invention, constructing an initialized artificial intelligence model includes: constructing a convolutional neural network model to obtain an initialized artificial intelligence model.
[0030] In an embodiment of the present invention, the initialization artificial intelligence model is trained with the pre-processed training sample data as support, and a synchronous intelligent optimization algorithm is used to obtain a vehicle fault diagnosis model, including: A1. Initialize the hyperparameters of the initialization artificial intelligence model, obtain a population, and divide the individuals in the population into a first sub-population and a second sub-population; For example, the hyperparameters of the initialized artificial intelligence model all have corresponding upper and lower limits. Therefore, they can be randomly initialized between the upper and lower limits of the hyperparameters and encoded as vectors, so that an individual can be obtained. After multiple initializations, multiple different individuals can be obtained repeatedly. Furthermore, the chaotic initial mapping method can be used to initialize the population, thereby improving the uniformity of the distribution of the initial solution in the solution space, which is conducive to improving the training speed.
[0031] The individuals in the population may be divided into a first sub-population and a second sub-population by random allocation.
[0032] A2. Based on the preprocessed training sample data, obtain the fitness of each first individual in the first subpopulation and the fitness of each second individual in the second subpopulation; A3. According to the fitness of each first individual in the first subpopulation, obtain a first best individual and a first worst individual; according to the fitness of each second individual in the second subpopulation, obtain a second best individual and a second worst individual; A4. Performing information interactive search on the second individual in the second subpopulation according to the first best individual and the first worst individual to obtain the second individual after the information interactive search; A5. Perform information interaction search on the first individual in the first subpopulation according to the second best individual and the second worst individual to obtain the first individual after the information interaction search; A6. For the first body after the information interaction search, an adaptive spiral search strategy is used to perform a local encirclement search on the first body to obtain the first body after the local encirclement search; A7. For the second individual after the information interaction search, the individual information interaction search strategy is used to perform a local expansion search on the second individual to obtain the second individual after the local expansion search; A8. Take the first individual after the local encirclement search and the second individual after the local expansion search as the target individual, and use the random perturbation strategy to perform a global search on the target individual to obtain the target individual after the global search; A9. Determine whether the training end conditions are met (such as the number of training times reaches the maximum number of training times). If so, obtain the global optimal individual based on the target individual after the global search, and use the hyperparameters in the global optimal individual as the final hyperparameters for initializing the artificial intelligence model to obtain the vehicle fault diagnosis model. Otherwise, return to the step of obtaining fitness.
[0033] In an embodiment of the present invention, the fitness of each first individual in the first subpopulation and the fitness of each second individual in the second subpopulation are obtained based on the preprocessed training sample data, including: For each first individual in the first subpopulation or each second individual in the second subpopulation, applying the hyperparameters contained in the first individual or the second individual to the initialized artificial intelligence model; For the initialized artificial intelligence model to which hyperparameters are applied, the preprocessed historical vehicle fault operation data is used as the input of the initialized artificial intelligence model, and the fault type label corresponding to the historical vehicle fault operation data is used as the expected output of the initialized artificial intelligence model to obtain the cross entropy loss function value; After adding the cross entropy loss function value to a preset constant term (such as 0.001), a non-zero numerical term is obtained, and the reciprocal of the non-zero numerical term is taken to obtain the fitness corresponding to the first individual or the second individual.
[0034] In the embodiment of the present invention, performing information interactive search on a second individual in a second subpopulation according to the first best individual and the first worst individual to obtain the second individual after the information interactive search includes: For the second individual in the second subpopulation, according to the first best individual and the first worst individual, the update speed corresponding to the second individual is obtained as follows: ;in, Indicates t During the training i The speed corresponding to the second individual, i =1,2,…,I, I represents the total number of the second individuals, Indicates t +1 training session i The speed corresponding to the second individual, Represents a random number between (0,1), represents the first optimal individual, represents the first worst individual, Represents a random number between (0,1), Indicates t During the training i a second individual; According to the update speed corresponding to the second individual, an information interaction search is performed on the second individual in the second subpopulation, and the second individual after the information interaction search is obtained as: ;in, Represents a random number between (0,1), Represents the second individual after information interaction search .
[0035] In the embodiment of the present invention, performing information interaction search on the first individual in the first subpopulation according to the second best individual and the second worst individual to obtain the first individual after the information interaction search includes: For the first individual in the first subpopulation, according to the second best individual and the second worst individual, the update speed corresponding to the first individual is obtained as follows: ;in, Indicates t During the training j The velocity corresponding to the first body, j =1,2,…,J, J represents the total number of the first individual, Indicates t +1 training session j The velocity corresponding to the first body, Represents a random number between (0,1), represents the second best individual, represents the second worst individual, Represents a random number between (0,1), Indicates t During the training j the first individual; According to the update speed corresponding to the first individual, an information interaction search is performed on the first individual in the first subpopulation, and the first individual after the information interaction search is obtained as: ;in, Represents a random number between (0,1), Indicates the first entity after information interaction search .
[0036] The synchronous intelligent optimization algorithm provided by the embodiment of the present invention adopts a dual-group search method, and the dual groups perform information interaction, which can effectively realize the search of more solution spaces and effectively avoid the algorithm from falling into a local optimum.
[0037] In the embodiment of the present invention, for the first volume after the information interaction search, an adaptive spiral search strategy is used to perform a local encirclement search on the first volume to obtain the first volume after the local encirclement search, including: Based on the current number of training times, the adaptive inertia weight is obtained as: ;in, represents the adaptive inertia weight, represents the sine function, represents pi, Indicates the preset maximum number of training times. Represents a constant term, which can be set to 1; For the first body after the information interaction search, a local encirclement search is performed on the first body according to the adaptive inertia weight, and the first body after the local encirclement search is obtained as follows: ;in, Indicates t During the training m The first body after the information interaction search, Represents the first volume after the local enclosing search , represents a natural constant, b represents the spiral search constant, l Represents a random number between [-1,1], represents the cosine function, Represents the circumference, represents the first optimal individual.
[0038] The adaptive spiral search strategy provided by the embodiment of the present invention can enable the first subpopulation to quickly surround a local optimal position, thereby improving the search speed of the algorithm, and the spiral search method is more conducive to jumping out of the local optimal solution, while improving the search accuracy in the later stage of the algorithm.
[0039] In the embodiment of the present invention, for the second individual after the information interaction search, the individual information interaction search strategy is used to perform a local extended search on the second individual to obtain the second individual after the local extended search, including: For the second individual after the information interaction search, obtain the mean of the second individual and obtain the central individual of the second population; According to the second optimal individual, the first local extended search item is obtained as: ;in, represents the first local expansion search term, Represents a random number between (0,1), Indicates t During the training n The second individual after the information interaction search, represents the second best individual, represents the local search weight; According to the central individual of the second population, the second local extended search item is obtained as: ,in, represents the second local expansion search term, Represents a random number between (0,1), represents the central individual, ; According to the first local extended search item and the second local extended search item, a local extended search is performed on the second individual, and the second individual after the local extended search is obtained as: ;in, Represents the second individual after the local expansion search .
[0040] The individual information interaction search strategy provided in the embodiment of the present invention can effectively interact with the information between the second individuals, thereby improving the traversal of the solution space, ensuring the individual diversity in the search process, improving the global search capability in the early stage of the algorithm, and ensuring the search for a better local optimal position in the later stage of the algorithm.
[0041] In an embodiment of the present invention, the first individual after the local encirclement search and the second individual after the local expansion search are taken as target individuals, and a random perturbation strategy is used to perform a global search on the target individuals to obtain the target individuals after the global search, including: The first individual after the local encirclement search and the second individual after the local expansion search are taken as target individuals; For any target individual, the adaptive variation range is: ;in, represents the maximum variation range, Indicates k The variation range corresponding to each target individual is represents the maximum fitness among the target individuals, Indicates k The fitness corresponding to the target individual is represents a constant term; According to the adaptive variation range, a global search is performed on the target individual, and the target individual after the global search is obtained as follows: ;in, Indicates the tth training process k Target individuals No. d Dimension parameters, Represents the target individual after global search No. d Dimension parameters, represents the adaptive inertia weight, express A random number between Determine whether the fitness of the target individual after the global search has increased. If so, accept the global search. Otherwise, reject the global search and use the original target individual directly as the target individual after the global search.
[0042] The random perturbation strategy provided by the embodiment of the present invention can effectively improve the global search capability of the algorithm, and at the same time greedily process the global search results, thereby avoiding the search stagnation problem of the algorithm.
[0043] Optionally, after each search, out-of-bounds processing can be performed on the parameters in the individual to ensure the validity of the search.
[0044] Those skilled in the art will readily appreciate other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. The present invention is intended to cover any variations, uses or adaptations of the present invention that follow the general principles of the present invention and include common knowledge or customary techniques in the art that are not disclosed by the present invention. It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from the scope thereof. The scope of the present invention is limited only by the appended claims.
Claims
1. A vehicle fault diagnosis method based on artificial intelligence, characterized in that: include: Collect historical vehicle fault operation data and fault type labels corresponding to the historical vehicle fault operation data to obtain training sample data, and preprocess the training sample data to obtain preprocessed training sample data; Construct an initialization artificial intelligence model, and use the pre-processed training sample data as support to train the initialization artificial intelligence model using a synchronous intelligent optimization algorithm to obtain a vehicle fault diagnosis model; The vehicle fault diagnosis model is deployed on a cloud platform, real-time vehicle fault operation data is collected, and the real-time vehicle fault operation data is transmitted to the cloud platform for identification to obtain a vehicle fault diagnosis result.
2. The vehicle fault diagnosis method based on artificial intelligence according to claim 1 is characterized in that: The real-time vehicle fault operation data is transmitted to the cloud platform for identification. After the vehicle fault diagnosis result is obtained, it also includes: when the vehicle fault diagnosis result meets the preset alarm condition, fault alarm information is generated through the cloud platform, and the fault alarm information is transmitted to the user-specified device and the vehicle computer for alarm.
3. The vehicle fault diagnosis method based on artificial intelligence according to claim 1 is characterized in that: Constructing an initialized artificial intelligence model, including: constructing a convolutional neural network model to obtain an initialized artificial intelligence model.
4. The vehicle fault diagnosis method based on artificial intelligence according to claim 3 is characterized in that: Based on the pre-processed training sample data, the synchronous intelligent optimization algorithm is used to train the initialized artificial intelligence model to obtain the vehicle fault diagnosis model, including: Initializing the hyperparameters of the initialization artificial intelligence model, obtaining a population, and dividing individuals in the population into a first sub-population and a second sub-population; Based on the preprocessed training sample data, the fitness of each first individual in the first subpopulation and the fitness of each second individual in the second subpopulation are obtained; According to the fitness of each first individual in the first subpopulation, a first best individual and a first worst individual are obtained; according to the fitness of each second individual in the second subpopulation, a second best individual and a second worst individual are obtained; According to the first best individual and the first worst individual, an information interactive search is performed on a second individual in the second subpopulation to obtain a second individual after the information interactive search; According to the second best individual and the second worst individual, an information interaction search is performed on the first individual in the first subpopulation to obtain a first individual after the information interaction search; For the first body after the information interaction search, an adaptive spiral search strategy is used to perform a local encirclement search on the first body to obtain the first body after the local encirclement search; For the second individual after the information interaction search, the individual information interaction search strategy is used to perform a local expansion search on the second individual to obtain the second individual after the local expansion search; The first individual after the local encirclement search and the second individual after the local expansion search are taken as the target individual, and a random perturbation strategy is used to perform a global search on the target individual to obtain the target individual after the global search; Determine whether the training end conditions are met. If so, obtain the global optimal individual based on the target individual after the global search, and use the hyperparameters in the global optimal individual as the final hyperparameters to initialize the artificial intelligence model to obtain the vehicle fault diagnosis model. Otherwise, return to the step of obtaining fitness.
5. The vehicle fault diagnosis method based on artificial intelligence according to claim 4 is characterized in that: Based on the preprocessed training sample data, the fitness of each first individual in the first subpopulation and the fitness of each second individual in the second subpopulation are obtained, including: For each first individual in the first subpopulation or each second individual in the second subpopulation, applying the hyperparameters contained in the first individual or the second individual to the initialized artificial intelligence model; For the initialized artificial intelligence model to which hyperparameters are applied, the preprocessed historical vehicle fault operation data is used as the input of the initialized artificial intelligence model, and the fault type label corresponding to the historical vehicle fault operation data is used as the expected output of the initialized artificial intelligence model to obtain the cross entropy loss function value; After adding the cross entropy loss function value to the preset constant term, a non-zero numerical term is obtained, and the reciprocal of the non-zero numerical term is taken to obtain the fitness corresponding to the first individual or the second individual.
6. The vehicle fault diagnosis method based on artificial intelligence according to claim 5 is characterized in that: According to the first best individual and the first worst individual, performing information interactive search on a second individual in the second subpopulation to obtain the second individual after the information interactive search, comprising: For the second individual in the second subpopulation, according to the first best individual and the first worst individual, the update speed corresponding to the second individual is obtained as follows: ;in, Indicates t During the training i The speed corresponding to the second individual, i =1,2,…,I, I represents the total number of the second individuals, Indicates t +1 training session i The speed corresponding to the second individual, Represents a random number between (0,1), represents the first optimal individual, represents the first worst individual, Represents a random number between (0,1), Indicates t During the training i a second individual; According to the update speed corresponding to the second individual, an information interaction search is performed on the second individual in the second subpopulation, and the second individual after the information interaction search is obtained as: ;in, Represents a random number between (0,1), Represents the second individual after information interaction search .
7. The vehicle fault diagnosis method based on artificial intelligence according to claim 6 is characterized in that: According to the second best individual and the second worst individual, an information interaction search is performed on the first individual in the first subpopulation to obtain the first individual after the information interaction search, including: For the first individual in the first subpopulation, according to the second best individual and the second worst individual, the update speed corresponding to the first individual is obtained as follows: ;in, Indicates t During the training j The velocity corresponding to the first body, j =1,2,…,J, J represents the total number of the first individual, Indicates t +1 training session j The velocity corresponding to the first body, Represents a random number between (0,1), represents the second best individual, represents the second worst individual, Represents a random number between (0,1), Indicates t During the training j the first individual; According to the update speed corresponding to the first individual, an information interaction search is performed on the first individual in the first subpopulation, and the first individual after the information interaction search is obtained as: ;in, Represents a random number between (0,1), Indicates the first entity after information interaction search .
8. The vehicle fault diagnosis method based on artificial intelligence according to claim 7 is characterized in that: For the first body after the information interaction search, an adaptive spiral search strategy is used to perform a local encirclement search on the first body, and the first body after the local encirclement search is obtained, including: Based on the current number of training times, the adaptive inertia weight is obtained as: ;in, represents the adaptive inertia weight, represents the sine function, represents pi, Indicates the preset maximum number of training times. represents a constant term; For the first body after the information interaction search, a local encirclement search is performed on the first body according to the adaptive inertia weight, and the first body after the local encirclement search is obtained as follows: ;in, Indicates t During the training m The first body after the information interaction search, Represents the first volume after the local enclosing search , represents a natural constant, b represents the spiral search constant, l Represents a random number between [-1,1], represents the cosine function, Represents the circumference, represents the first optimal individual.
9. The vehicle fault diagnosis method based on artificial intelligence according to claim 8 is characterized in that: For the second individual after the information interaction search, the individual information interaction search strategy is used to perform a local extension search on the second individual, and the second individual after the local extension search is obtained, including: For the second individual after the information interaction search, obtain the mean of the second individual and obtain the central individual of the second population; According to the second optimal individual, the first local extended search item is obtained as: ;in, represents the first local expansion search term, Represents a random number between (0,1), Indicates t During the training n The second individual after the information interaction search, represents the second best individual, represents the local search weight; According to the central individual of the second population, the second local extended search item is obtained as: ,in, represents the second local expansion search term, Represents a random number between (0,1), represents the central individual, ; According to the first local extended search item and the second local extended search item, a local extended search is performed on the second individual, and the second individual after the local extended search is obtained as: ;in, Represents the second individual after the local expansion search .
10. The vehicle fault diagnosis method based on artificial intelligence according to claim 9 is characterized in that: The first individual after the local encirclement search and the second individual after the local expansion search are taken as the target individuals, and a random perturbation strategy is used to perform a global search on the target individuals to obtain the target individuals after the global search, including: The first individual after the local encirclement search and the second individual after the local expansion search are taken as target individuals; For any target individual, the adaptive variation range is obtained as: ;in, represents the maximum variation range, Indicates k The variation range corresponding to each target individual is represents the maximum fitness among the target individuals, Indicates k The fitness corresponding to the target individual is represents a constant term; According to the adaptive variation range, a global search is performed on the target individual, and the target individual after the global search is obtained as follows: ;in, Indicates the tth training process k Target individuals No. d Dimension parameters, Represents the target individual after global search No. d Dimension parameters, represents the adaptive inertia weight, express A random number between Determine whether the fitness of the target individual after the global search has increased. If so, accept the global search. Otherwise, reject the global search and use the original target individual directly as the target individual after the global search.