A VR-based vehicle maintenance assistance system

Abstract

A VR-based vehicle maintenance assistance system, comprising a vehicle data acquisition module, a vehicle fault database, a virtual fault repair database, a data processing module, a vehicle fault diagnosis module, and a virtual reality demonstration module. The beneficial effects of the present invention are as follows: The BP neural network is applied to vehicle fault diagnosis, and the BP neural network is trained using the fault types of the vehicle and the corresponding vehicle operation data, thereby establishing a mapping between the vehicle operation data and the fault types, so that the fault types of the vehicle can be diagnosed in real time according to the obtained vehicle operation data; The virtual reality technology is applied to vehicle maintenance. By reproducing the vehicle maintenance process through virtual reality technology, maintenance personnel can more intuitively understand the maintenance process, laying a foundation for vehicle maintenance.

Description

Technical Field

[0001] The present invention relates to the field of vehicle maintenance, and particularly to a VR-based vehicle maintenance assistance system. Background Art

[0002] The rapid economic development has significantly improved people's living standards, and the demand for vehicles is also increasing, which has promoted the relatively fast development of China's automobile industry. Especially in recent years, with the rapid changes in science and technology, the architecture of automobiles has become increasingly complex. Therefore, the requirements for vehicle maintenance are becoming more and more strict. It is precisely due to the continuous development of new technologies that the vehicle fault diagnosis technology has been continuously updated and improved. Nowadays, artificial intelligence has become a field with numerous practical applications and active research topics, and it is developing rapidly. Improving the existing vehicle fault diagnosis technology through artificial intelligence methods is of great significance for enhancing the level of vehicle fault diagnosis technology.

[0003] In addition, the internal structure of the vehicle is complex, and there are many maintenance steps and maintenance tools involved in the vehicle maintenance process. Applying virtual reality technology to vehicle maintenance, by reproducing the vehicle maintenance process through virtual reality technology, maintenance personnel can more intuitively understand the safety conditions, the use of maintenance tools, and the maintenance steps involved in the maintenance process, laying a foundation for vehicle maintenance. Summary of the Invention

[0004] In view of the above problems, the present invention aims to provide a VR-based vehicle maintenance assistance system.

[0005] The object of the present invention is achieved by the following technical solutions:

[0006] A VR-based vehicle maintenance assistance system includes a vehicle data acquisition module, a vehicle fault database, a virtual fault repair database, a data processing module, a vehicle fault diagnosis module, and a virtual reality demonstration module. The vehicle data acquisition module is used to acquire vehicle operation data that can reflect the vehicle state. The vehicle fault database stores the fault types of the vehicle and the vehicle operation data corresponding to the fault types. The virtual fault repair database stores the fault types of the vehicle and the repair methods corresponding to the fault types. The data processing module is used to perform normalization processing on the data in the vehicle data acquisition module and the vehicle fault database. The vehicle fault diagnosis module trains a BP neural network diagnosis model using the data in the normalized vehicle fault database and inputs the vehicle operation data of the normalized vehicle data acquisition module into the trained BP neural network diagnosis model for vehicle fault diagnosis. The output result of the BP neural network diagnosis model is the fault type of the vehicle. The virtual reality demonstration module is used to retrieve the repair method corresponding to the fault type stored in the virtual fault repair database according to the fault type diagnosed by the vehicle fault diagnosis module and demonstrate the repair method using virtual reality technology.

[0007] The beneficial effects of the present invention: The BP neural network is applied to vehicle fault diagnosis, and the BP neural network is trained using the fault types of the vehicle and the corresponding vehicle operation data, so as to establish a mapping between the vehicle operation data and the fault types, and the fault types of the vehicle can be diagnosed in real time according to the obtained vehicle operation data. The virtual reality technology is applied to vehicle maintenance, and the vehicle maintenance process is reproduced through virtual reality technology. By watching the virtual maintenance process demonstration, maintenance personnel can more intuitively understand the maintenance process, laying a foundation for vehicle maintenance. Brief Description of the Drawings

[0008] The invention is further described with reference to the accompanying drawings. However, the embodiments in the drawings do not constitute any limitation to the present invention. For those of ordinary skill in the art, other drawings can be obtained according to the following drawings without creative efforts.

[0009] Figure 1 It is a schematic structural diagram of the present invention. Detailed Embodiments

[0010] The present invention is further described in conjunction with the following embodiments.

[0011] See Figure 1, an auxiliary system for vehicle maintenance based on VR in this embodiment includes a vehicle data acquisition module, a vehicle fault database, a virtual fault repair database, a data processing module, a vehicle fault diagnosis module, and a virtual reality demonstration module. The vehicle data acquisition module is used to acquire vehicle operation data that can reflect the vehicle state. The vehicle fault database stores the fault types of the vehicle and the vehicle operation data corresponding to the fault types. The virtual fault repair database stores the fault types of the vehicle and the repair methods corresponding to the fault types. The data processing module is used to perform normalization processing on the data in the vehicle data acquisition module and the vehicle fault database. The vehicle fault diagnosis module trains a BP neural network diagnosis model using the data in the normalized vehicle fault database, and inputs the vehicle operation data of the normalized vehicle data acquisition module into the trained BP neural network diagnosis model for vehicle fault diagnosis. The output result of the BP neural network diagnosis model is the fault type of the vehicle. The virtual reality demonstration module is used to retrieve the repair method corresponding to the fault type stored in the virtual fault repair database according to the fault type diagnosed by the vehicle fault diagnosis module, and demonstrate the repair method using virtual reality technology.

[0012] In this preferred embodiment, the BP neural network is applied to vehicle fault diagnosis. The BP neural network is trained using the fault types of the vehicle and the corresponding vehicle operation data, so as to establish a mapping between the vehicle operation data and the fault types. Thus, the fault types of the vehicle can be diagnosed in real time according to the obtained vehicle operation data. The virtual reality technology is applied to vehicle maintenance. The vehicle maintenance process is reproduced through virtual reality technology. By watching the virtual maintenance process demonstration, maintenance personnel can understand the maintenance process more intuitively, laying a foundation for vehicle maintenance.

[0013] Preferably, the particle swarm optimization algorithm is used to optimize the weights and thresholds of the BP neural network used in the vehicle fault diagnosis module. The expression of the fitness function f of the particle swarm optimization algorithm is defined as:

[0014]

[0015] In the formula, n represents the number of samples, Y ij represents the j-th ideal output value of the i-th sample, and y ij represents the j-th actual output value of the i-th sample, and p represents the number of output nodes.

[0016] In this preferred embodiment, the particle swarm optimization algorithm and the BP neural network are combined. The particle swarm optimization algorithm is used to optimize the weights and thresholds of the BP neural network. The optimization of the particle swarm optimization algorithm helps the BP neural network jump out of the local minimum, and better improves the performance of fault diagnosis. The smaller the value of the defined fitness function, the better the optimization result of the particle.

[0017] Preferably, three different particle update modes are defined in the particle swarm optimization algorithm. The first particle update mode is as follows:

[0018]

[0019] X i (t + 1) = X i (t) + V i (t + 1)

[0020] The second particle update mode is as follows:

[0021]

[0022] X i (t + 1) = X i (t) + V i (t + 1)

[0023] The third particle update mode is as follows:

[0024]

[0025] X i (t + 1) = X i (t) + V i (t + 1)

[0026] In the formula, X i (t) and V i (t) respectively represent the position and step size of particle i after the t-th iterative update. X i (t + 1) and V i (t + 1) respectively represent the position and step size of particle i after the (t + 1)-th iterative update. represents the inertia weight factor corresponding to the first update mode when particle i is updated after the t-th iterative update. represents the inertia weight factor corresponding to the second update mode when particle i is updated after the t-th iterative update. represents the inertia weight factor corresponding to the third update mode when particle i is updated after the t-th iterative update. rand() represents a random number between 0 and 1. c1, c2, c3, c4, c5, and c6 are respectively learning factors, and the values of c1, c2, c3, c4, c5, c6 are set to 2. B(t) represents the global optimal solution of the particle swarm after the t-th iterative update. P i (t) represents the individual optimal solution of particle i after the t-th iterative update. L i (t) represents the local reference solution of particle i after the t-th iterative update. And θ(Xi (t), X l (t)) is a judgment function, and wherein, X l (t) represents the position of particle l after the t-th iteration update, f i (t) represents the fitness function value corresponding to particle i after the t-th iteration update, f l (t) represents the fitness function value corresponding to particle l after the t-th iteration update, d(X i (t), X l (t)) represents the position X of particle i after the t-th iteration update i (t) and the position X of particle l after the t-th iteration update l (t) The distance between them is R i (t) is the local reference radius corresponding to particle i after the t-th iteration update, and represents the position X of particle i after the t-th iteration update i (t) and the minimum value of the distances between the positions of other particles in the particle swarm, and represents the position X of particle i after the t-th iteration update i (t) and the median of the distances between the positions of other particles in the particle swarm, and K i (t) represents the global reference solution of particle i after the t-th iteration update, and d(P i (t), B(t)) represents the distance between the individual optimal solution P of particle i after the t-th iteration update i (t) and the global optimal solution B(t), represents the position X of particle i after the t-th iteration update i (t) and the maximum value of the distances between the positions of other particles in the particle swarm, and

[0027] This preferred embodiment defines three different particle update modes for the particle swarm, which improves the diversity of the particle swarm algorithm and avoids the algorithm falling into local optima. In the first particle update mode, the particle is made to learn towards the global optimal solution of the particle swarm, which improves the global search ability of the particle swarm algorithm. In the second particle update mode, the particle is made to learn from its individual optimal solution and the particles within its local reference radius whose fitness function values are better than its own fitness value, which improves the local search ability of the particle swarm algorithm. In the third update mode, the individual optimal solution of the particle and the global optimal solution of the particle swarm are introduced, taking into account both the global search ability and the local search ability of the particle swarm algorithm. At the same time, a global reference solution is introduced in the third update mode. When the difference between the individual optimal solution of the particle and the global optimal solution is large, the global reference solution added in the third update mode can effectively increase the value of V i (t + 1), thereby effectively improving the convergence speed of the particle and being conducive to global search.

[0028] Preferably, define to represent the priority of particle i selecting the j-th update mode for the (t + 1)-th iterative update, and the expression of is:

[0029]

[0030] In the formula, represents the total number of iterations of particle i using the j-th update mode for update from initialization to the t-th iterative update, represents the iteration number corresponding to particle i using the j-th update mode for the l-th update, represents particle i at the -th iterative update corresponding fitness function value, represents particle i at the -th iterative update corresponding fitness function value, is a comparison function. When , then When , then represents the iteration number corresponding to particle i's most recent use of the j-th update mode for update, represents particle i at the -th iterative update corresponding fitness function value, represents particle i at the -th iterative update corresponding fitness function value, f i max (t) and f i min(t) represent the maximum fitness function value and the minimum fitness function value of particle i after the t-th iteration update from initialization, is the correction factor of the priority for particle i to be updated using the j-th update mode after the t-th iteration update. When f i (t) ≤ H1(f), then When H1(f) < f i (t) ≤ H2(f), then When f i (t) > H2(f), then where H1(f) and H2(f) are the given first fitness function threshold and second fitness function threshold, and where N is the total number of particles in the particle swarm, is the average fitness function value of the particle swarm after the t-th iteration update, is the comparison function, then When then

[0031] Particle i finally selects the update mode with the maximum priority for the (t + 1)-th iteration update.

[0032] In this preferred embodiment, the priorities for the particle to select the first, second, and third update modes for update are calculated to determine which update mode the particle selects for iterative update. When the particle performs iterative update, first, the priorities for the particle to select the first, second, and third update modes for update are calculated. The calculation formula of the priority includes an exponential function part and a correction factor part. The value of the exponential function part is jointly determined by the change in the fitness function value after the particle last updated using this update mode and the success rate of the particle using this update mode for update. When the particle last updated using this update mode, the smaller the fitness function value of the particle compared to the fitness function value after the previous iterative update, the better the optimization result of the particle using this update mode, and the larger the value of the exponential function part. When the success rate of the particle using this update mode for update is higher, it indicates that the optimization result of the particle using this update mode for update is better, and the value of the exponential function part is also larger, that is, the priority of the particle using this update mode is greater; the correction factor in the priority formula is considered from the current fitness function value of the particle. The smaller the fitness function value of the particle after the current iterative update, the better the optimization ability of the particle. Let the particle with better optimization ability explore a better search area globally, thereby driving the search of other particles. Therefore, the correction factor increases the priority of the particle with better optimization ability to select the first update mode for update. The larger the fitness function value of the particle after the current iterative update, the worse the optimization ability of the particle. Let the particle with worse optimization ability strengthen local search, that is, the correction factor increases the priority of the particle with worse optimization ability to select the second update mode for update. When the fitness function value of the particle after the current iterative update is between the first fitness function threshold and the second fitness function threshold, it indicates that the particle is a particle with average optimization ability in the particle swarm and has better global search ability and local search ability. Therefore, the correction factor increases the priority of the particle with average optimization ability to select the third update mode for update; finally, the particle selects the update mode with the largest priority for update, and the selected update mode is more adaptable to the particle's own characteristics, improving the optimization ability of the particle.

[0033] Preferably, let represent the inertia weight factor corresponding to the particle i using the j-th update mode for update after the t-th iterative update, and j = 1, 2, 3, The value of is determined in the following manner:

[0034] (1) Arrange the particles in the particle swarm that select the j-th update mode for the (t + 1)-th iterative update after the t-th iterative update in ascending order of their fitness values to form a set J(t), and obtain the ranking of the particle i in the set J(t)

[0035] (2) Given the ideal fitness function value F(X), define the population optimization degree detection factor of the particle swarm after the t-th iteration update as β(t), then the expression of β(t) is:

[0036]

[0037] where f(P i (t)) represents the fitness function value corresponding to the individual optimal solution P i (t) of particle i after the t-th iteration update, f max (t) represents the maximum fitness function value of the particle swarm after the t-th iteration update, f min (t) represents the minimum fitness function value of the particle swarm after the t-th iteration update, F(X) is the given ideal fitness function value, and the value of F(X) can be taken as 0.0001. The position X corresponding to the ideal fitness function value F(X) is regarded as the ideal optimal solution;

[0038] Then The value of is:

[0039]

[0040] In the formula, ω max and ω min are the given maximum and minimum inertia weight factors, is the maximum inertia weight factor corresponding to the update using the j-th update mode, and N j is the number of particles in the set J(t).

[0041] This preferred embodiment is used to determine the inertia weight factor corresponding to the particles that select the j-th update mode for update after the t-th iteration update in the particle swarm. The set inertia weight factor value is jointly determined by the fitness function value of the particles and the population optimization degree detection factor of the particle swarm. First, the particles that select the same update mode are sorted from small to large according to their fitness values to form a set J(t), and the sorting of the particle fitness values is obtained. This sorting reflects the quality of the optimization result of the particle compared with the optimization results of other particles that select the same update mode. The earlier the sorting, the smaller the fitness function value, indicating that the particle has better optimization ability compared with other particles that select the same update mode. For this type of particle, a larger inertia weight factor is set so that it can explore a better search area globally. For the particles with a later sorting, their fitness values are larger, indicating that the particle has worse optimization ability compared with other particles that select the same update mode. For this type of particle, a smaller inertia weight factor is set so that it can converge to a better area faster and strengthen the local search in this area. The sorting of the particle fitness values can only reflect the optimization state of the particle itself and cannot reflect the overall optimization state of the particle swarm. When the particle swarm has converged to near the ideal optimal solution as a whole, if a larger inertia weight value is used for the particles with better optimization ability, it is very likely that the particle with better optimization ability will escape from the ideal optimal solution, thus reducing the optimization performance of the particle. To avoid this phenomenon, this preferred embodiment introduces a population optimization degree detection factor. The population optimization detection factor judges the overall optimization state of the particle swarm by comparing the fitness function values of each particle in the particle swarm with the given ideal fitness function value. The larger the value of β(t), the closer the particle swarm as a whole is to the ideal optimal solution. The smaller the value of β(t), the farther the particle swarm as a whole is from the ideal optimal solution. The inertia weight factor of the particles is constrained according to the population optimization degree detection factor, thus avoiding the phenomenon that a larger inertia weight factor is set for the particles with better optimization ability and causing the particle to escape from the ideal optimal solution, and increasing the optimization performance of the particle swarm algorithm.

[0042] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the protection scope of the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the essence and scope of the technical solutions of the present invention.

Claims

1. A vehicle maintenance and repair assistance system based on VR, characterized in that it includes a vehicle data acquisition module, a vehicle fault database, a virtual fault repair database, a data processing module, a vehicle fault diagnosis module, and a virtual reality demonstration module. The vehicle data acquisition module is used to acquire vehicle operation data that can reflect the vehicle state. The vehicle fault database stores the fault types of the vehicle and the vehicle operation data corresponding to the fault types. The virtual fault repair database stores the fault types of the vehicle and the repair methods corresponding to the fault types. The data processing module is used to perform normalization processing on the data in the vehicle data acquisition module and the vehicle fault database. The vehicle fault diagnosis module trains a BP neural network diagnosis model using the data in the normalized vehicle fault database, and inputs the vehicle operation data of the normalized vehicle data acquisition module into the trained BP neural network diagnosis model for vehicle fault diagnosis. The output result of the BP neural network diagnosis model is the fault type of the vehicle. The virtual reality demonstration module is used to retrieve the repair method corresponding to the fault type stored in the virtual fault repair database according to the fault type diagnosed by the vehicle fault diagnosis module, and demonstrate the repair method using virtual reality technology; Characterized in that the weights and thresholds of the BP neural network used in the vehicle fault diagnosis module are optimized by the particle swarm algorithm, and the fitness function of the particle swarm algorithm is defined The expression of is: In the formula, Represents the number of samples, Represents the th sample's th ideal output value, Represents the th sample's th actual output value, Represents the number of output nodes; Three different particle update modes are defined in the particle swarm algorithm. The first particle update mode is: The second particle update mode is: The third particle update mode is: In the formula, And Respectively represent particles At the position and step size after the -th iteration update, respectively represent the position and step size of the particle after the -th iteration update, represents the inertial weight factor corresponding to the first update mode when the particle is updated after the -th iteration update, represents the inertial weight factor corresponding to the second update mode when the particle is updated after the -th iteration update, represents the inertial weight factor corresponding to the third update mode when the particle is updated after the represents generating a to random number, , , , , and are learning factors respectively, and , , , , , are set to , represents the global optimal solution after the -th iteration update of the particle swarm, represents the individual optimal solution of the particle after the -th iteration update, represents the local reference solution of the particle after the -th iteration update, and , is a judgment function, and , where represents the position of the particle after the -th iteration update, represents the fitness function value corresponding to the particle after the -th iteration update, represents the fitness function value corresponding to the particle after the -th iteration update, represents the fitness function value corresponding to the particle At the position after the th iteration update and the particle position after the th iteration update, the local reference radius corresponding to the particle after the th iteration update, and , represents the minimum distance between the position of the particle after the th iteration update and the positions of other particles in the particle swarm, and , represents the median distance between the position of the particle after the th iteration update and the positions of other particles in the particle swarm, and , represents the global reference solution of the particle after the th iteration update, and , represents the distance between the individual optimal solution of the particle after the th iteration update and the global optimal solution, represents the maximum distance between the position of the particle after the th iteration update and the positions of other particles in the particle swarm, and ; Define to represent the priority of the th update mode selected for the th iteration update by the particle , and the expression of is: In the formula, represents the total number of iterations of the particle from initialization to the th iteration update using the th update mode, represents the iteration number corresponding to the th use of the th update mode by the particle , represents the particle at the The fitness function value corresponding to the represents the particle at the -th iteration update, is a comparison function. When , then , when , then . represents the particle in the most recent iteration corresponding to the update using the -th update mode, represents the particle at the -th iteration update, represents the particle at the -th iteration update, and respectively represent the maximum and minimum fitness function values of the particle from initialization to the -th iteration update, is the correction factor for the priority of the particle at the -th iteration update when using the -th update mode. When , then , when , then , when , then , where and are the given first fitness function threshold and second fitness function threshold, and . , where is the total number of particles in the particle swarm, is the average fitness function value of the particle swarm at the -th iteration update, is a comparison function. When , then , when , then ; The particle finally selects the update mode with the highest priority for the -th iteration update; Its feature is that let represent the particle at the After the -th iteration update, the corresponding inertia weight factor is updated using the -th update mode, and is determined in the following way:

1. Arrange the particles that are updated in the -th iteration update and select the -th update mode for the -th iteration update in ascending order of their fitness values to form a set , and obtain the sorting of particle in the set ; 2. Given the ideal fitness function value , define the population optimization degree detection factor of the particle swarm after the -th iteration update as , then is expressed as: where represents the fitness function value corresponding to the individual optimal solution of particle after the -th iteration update, represents the maximum fitness function value of the particle swarm after the -th iteration update, represents the minimum fitness function value of the particle swarm after the -th iteration update; then is: In the formula, and are the given maximum and minimum inertia weight factors, is the maximum inertia weight factor corresponding to the update using the -th update mode, and , is the number of particles in the set .

Images