Simulation parameter decision-making method and device based on deep learning representation and medium

Abstract

The invention discloses a simulation parameter decision-making method and device based on deep learning representation and a medium, and relates to the field of simulation parameter decision-making, and the method comprises the steps: obtaining simulation data of historical vehicle design modeling, and extracting vehicle design parameters and vehicle performance parameters from the simulation data; key vehicle design parameters are screened out from the design parameter set by applying a CFS method; the key vehicle design parameters serve as network input, the vehicle performance parameters serve as network output, a neural network KAN is trained, and then a nonlinear equation set between the key vehicle design parameters and the vehicle performance parameters is obtained; and solving the nonlinear equation set by using a multi-objective optimization algorithm to obtain an optimal vehicle design parameter value by taking the condition that each vehicle performance parameter meets a set objective as a parameter optimization objective. By applying the method provided by the invention, the problems of inaccurate decision-making result and low efficiency caused by long time period in the existing simulation parameter decision-making method can be solved.

Description

Technical Field

[0001] The present application relates to the field of simulation parameter decision making, and in particular to a simulation parameter decision making method, device and medium based on deep learning representation. Background Art

[0002] In recent years, the automotive industry has actively followed the trend of industrial development, strengthened the position of automotive enterprises as the main body of innovation, and become a highland leading the innovation and development of the global automotive industry. Among them, the development of the new energy vehicle industry has achieved positive results, injecting strong momentum into the electrification transformation of the global automotive industry. It integrates a number of revolutionary technologies and has become an important carrier for the implementation of new technologies.

[0003] The development of new energy vehicles is an iterative process of design, testing and optimization. The interaction of complex multi-physical systems in automobiles brings challenges to the development of new energy vehicles. Modeling and simulation technology provides a convenient means to check and optimize the complex interactions in electric vehicles. The system-level automobile development of simulation software involves vehicle performance, performance integration, energy management, PE performance development, thermal management, technical planning, new energy system development, powertrain, fuel cell simulation and other aspects. Reducing the number of required experiments through simulation technology has become an auxiliary development method for automobile companies to reduce development costs, shorten product time to market and improve product quality.

[0004] The traditional simulation parameter decision method can find the best design parameter combination by systematically searching within the predefined design parameter range, but this method has a large computational overhead. There are also control strategies that enable the control system to respond quickly to changes in the external environment through real-time feedback and parameter adjustment, that is, error-based control strategies to eliminate errors. For example, PID Tuner adjusts the output of the system through a combination of three links: proportional (P), integral (I), and differential (D) to track the reference input and suppress interference. In addition, the more common simulation parameter decision method is to list all parameters flatly. Engineers rely on their own design experience or existing established rules to define the design parameters, and gradually adjust them according to the simulation results until they are adjusted to reasonable results. However, the existing simulation decision-making methods have some disadvantages: (1) The method of real-time feedback and parameter adjustment is subject to the particularity of the simulation system. For example, for nonlinear systems, this method may not provide the best control effect. On the other hand, it is sensitive to noise and external disturbances, especially the differential term, which may cause the simulation system output to become unstable when facing high-frequency noise. In addition, there is an integral saturation problem (integral overflow). When the simulation system cannot reach the set point for a long time, the integral term may accumulate too much, causing the controller to saturate and affect the control effect. (2) For simulation systems that cannot use the feedback adjustment method, if the most traditional manual parameter adjustment method is relied upon, if engineers lack prior experience, it will take a long time to repeatedly adjust and optimize multiple vehicle design parameters to achieve the best performance by customizing the vehicle design parameters.

[0005] Therefore, the existing simulation parameter decision-making method has the problems of inaccurate decision results and low efficiency due to a long time period. Summary of the invention

[0006] The purpose of this application is to provide a simulation parameter decision method, device and medium based on deep learning representation, which can solve the problems of inaccurate decision results and low efficiency due to long time period in existing simulation parameter decision methods.

[0007] To achieve the above objectives, this application provides the following solutions.

[0008] In a first aspect, the present application provides a simulation parameter decision method based on deep learning representation, comprising the following steps: Acquire simulation data of historical vehicle design modeling, extract vehicle design parameters and vehicle performance parameters from the simulation data, and obtain a design parameter set and a performance parameter set; Applying the CFS method to select key vehicle design parameters from the design parameter set to obtain a key design parameter set; Taking each key vehicle design parameter in the key design parameter set as a network input and each corresponding vehicle performance parameter in the performance parameter set as a network output, training the neural network KAN to obtain a trained neural network KAN; A nonlinear equation group between key vehicle design parameters and various vehicle performance parameters is obtained according to the network structure of the trained neural network KAN; a nonlinear equation in the nonlinear equation group is a nonlinear relationship formula between a vehicle performance parameter and various key vehicle design parameters; Taking the performance parameters of each vehicle satisfying the set targets as the parameter optimization target, a multi-objective optimization algorithm is used to solve the nonlinear equation group to obtain the optimal vehicle design parameter values.

[0009] In a second aspect, the present application provides a simulation parameter decision device based on deep learning representation, comprising the following modules: A simulation data acquisition module is used to acquire simulation data of historical vehicle design modeling, extract vehicle design parameters and vehicle performance parameters from the simulation data, and obtain a design parameter set and a performance parameter set; A key vehicle design parameter screening module, used for screening key vehicle design parameters from the design parameter set using a CFS method to obtain a key design parameter set; A neural network KAN training module is used to train the neural network KAN using each key vehicle design parameter in the key design parameter set as a network input and each corresponding vehicle performance parameter in the performance parameter set as a network output to obtain a trained neural network KAN; A nonlinear equation group construction module is used to obtain a nonlinear equation group between key vehicle design parameters and various vehicle performance parameters according to the network structure of the trained neural network KAN; a nonlinear equation in the nonlinear equation group is a nonlinear relationship formula between a vehicle performance parameter and various key vehicle design parameters; The solution module is used to solve the nonlinear equations using a multi-objective optimization algorithm with the performance parameters of each vehicle meeting the set target as the parameter optimization target, so as to obtain the optimal vehicle design parameter value.

[0010] In a third aspect, the present application provides a computer device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-mentioned simulation parameter decision method based on deep learning representation.

[0011] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned simulation parameter decision method based on deep learning representation.

[0012] According to the specific embodiments provided in this application, this application discloses the following technical effects: The present application provides a simulation parameter decision method, device and medium based on deep learning representation, obtains simulation data of historical vehicle design modeling, extracts vehicle design parameters input by the simulation system and vehicle performance parameters output by the simulation system from the simulation data; applies CFS method to screen out key vehicle design parameters from the extracted vehicle design parameters; trains neural network KAN with key vehicle design parameters as network input and output vehicle performance parameters as network output; obtains nonlinear equations between key vehicle design parameters and vehicle performance parameters according to the network structure of trained neural network KAN; takes each vehicle performance parameter satisfying the set target as parameter optimization target, solves nonlinear equations using multi-objective optimization algorithm, and obtains optimal vehicle design parameter values. The present invention takes into account the large amount of expert experience and design knowledge contained in the existing vehicle established parameter settings and the data generated by multiple simulations, which has certain significance for vehicle development. Therefore, the present invention makes full use of these existing vehicle simulation data, mines and reuses the knowledge and rules implied in the relationship between vehicle design parameters and vehicle performance, thereby reducing the actual number of simulations, thereby improving iterative design efficiency and reducing simulation development costs. In addition, key vehicle design parameters are extracted from numerous vehicle design parameters, redundant design parameters are reduced, and the computational efficiency of simulation parameter decision-making can be improved. In addition, the present invention uses key vehicle design parameters and vehicle performance parameters to train the neural network KAN, uses a deep learning algorithm to characterize the simulation process, and transforms the simulation parameter recommendation problem into a multi-objective optimization problem to solve the global optimal solution of the simulation process. Since the training of the neural network KAN is not subject to the particularity of the simulation system, and the spline function activation mechanism in the neural network enables it to more accurately capture the nonlinear relationship between vehicle design parameters and vehicle performance parameters, improving the fitting accuracy. Therefore, compared with the existing simulation decision-making method, the present invention not only ensures the efficiency of simulation parameter decision-making, but also ensures the accuracy of simulation parameter decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0014] Figure 1 This is an application environment diagram of a simulation parameter decision method based on deep learning representation in one embodiment of the present application; Figure 2 A flowchart of a simulation parameter decision method based on deep learning representation provided in one embodiment of the present application; Figure 3 A schematic diagram of the network structure of a neural network KAN provided in one embodiment of the present application; Figure 4 A schematic diagram of functional modules of a simulation parameter decision device based on deep learning representation provided in one embodiment of the present application; Figure 5 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0015] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0016] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0017] The simulation parameter decision method based on deep learning representation provided in the embodiment of the present application can be applied to Figure 1In the application environment shown in the figure, the terminal communicates with the server through the network. The data storage system can store the data that the server needs to process. The data storage system can be set up separately, integrated on the server, or placed on the cloud or other servers. The terminal can send the simulation data of the historical vehicle design modeling to the server. After the server receives the simulation data of the historical vehicle design modeling, the server extracts the vehicle design parameters and vehicle performance parameters from the simulation data to obtain the design parameter set and the performance parameter set; the CFS method is used to screen the key vehicle design parameters from the design parameter set to obtain the key design parameter set; the key vehicle design parameters in the key design parameter set are used as network inputs, and the vehicle performance parameters in the performance parameter set are used as network outputs to train the neural network KAN to obtain the trained neural network KAN; the nonlinear equation group between the key vehicle design parameters and the vehicle performance parameters is obtained according to the network structure of the trained neural network KAN; the parameter optimization target is that each vehicle performance parameter meets the set target, and the nonlinear equation group is solved by using the multi-objective optimization algorithm to obtain the optimal vehicle design parameter value. The server can feedback the obtained optimal vehicle design parameter value to the terminal. In addition, in some embodiments, the simulation parameter decision method based on deep learning representation can also be implemented independently by a server or a terminal. For example, the terminal can directly calculate the optimal vehicle design parameter value for the simulation data of historical vehicle design modeling, or the server can obtain the simulation data of historical vehicle design modeling from the data storage system and calculate the optimal vehicle design parameter value.

[0018] The terminal may be, but is not limited to, various desktop computers, laptops, smart phones, tablet computers, IoT devices, and portable wearable devices. The server may be implemented as an independent server or a server cluster consisting of multiple servers, or may be a cloud server.

[0019] In an exemplary embodiment, Figure 2 As shown, a simulation parameter decision method based on deep learning representation is provided. The method is executed by a computer device, and can be executed by a computer device such as a terminal or a server alone, or by a terminal and a server together. In the embodiment of the present application, the method is applied to Figure 1 The server in is taken as an example to illustrate, including the following steps 101 to 105.

[0020] Step 101 , obtaining simulation data of historical vehicle design modeling, extracting vehicle design parameters and vehicle performance parameters from the simulation data, and obtaining a design parameter set and a performance parameter set.

[0021] Vehicle design parameters include size parameters (such as vehicle length, width, height, wheelbase, track, minimum ground clearance, etc.), weight parameters (such as curb weight, gross mass, load capacity, etc.), body parameters (such as body type, number of doors, number of seats, etc.), power system parameters (such as engine type, engine displacement, engine layout, drive mode, etc.), suspension system parameters (such as front suspension type, rear suspension type, etc.), braking system parameters (such as front brake type, rear brake type, etc.), etc.

[0022] Vehicle performance parameters include power performance parameters (such as the maximum power of the engine, the maximum torque of the engine, the time required for the vehicle to accelerate from a standstill to 100 km / h, the maximum speed, etc.), braking performance parameters (such as 100-0 km / h braking distance, braking system response time, etc.), fuel performance parameters (such as comprehensive fuel consumption, cruising range, etc.), etc.

[0023] Step 102: Apply the CFS method to filter out key vehicle design parameters from the design parameter set to obtain a key design parameter set. The key vehicle design parameters are part of the vehicle design parameters filtered out from the design parameter set by applying the CFS method.

[0024] Step 103 , using each key vehicle design parameter in the key design parameter set as a network input and each vehicle performance parameter in the performance parameter set as a network output, training the neural network KAN to obtain a trained neural network KAN.

[0025] By introducing the neural network KAN improved by MLP to characterize the simulation process, the extracted vehicle performance parameters are used as the output of the neural network KAN, and the key vehicle design parameters input by simulation are used as the input of the neural network KAN to train the neural network KAN.

[0026] The neural network KAN has learnable activation functions on the edges between nodes, which are built using basis functions, so the training structure does not need to specify the activation function. It introduces the Kolmogorov-Arnold Representation Theorem (KAT) and B-spline functions, allowing the neural network KAN to improve its data processing method in each training iteration, thereby continuously improving its accuracy and efficiency. Figure 3 is a network structure of a neural network KAN, x is the m input of the neural network KAN, KAN(x) is the n output of the neural network KAN, , is a nonlinear learnable activation function, ADAM is used as the optimizer, and RMSE is used as the loss function. The loss function is as follows.

[0027] .

[0028] in, is the calculated output value of the neural network KAN, that is, the vehicle performance parameter output by the network; , , ..., , ..., is the actual simulation output result in the simulation data, that is, the vehicle performance parameter in the simulation data; i refers to the i-th vehicle performance parameter; n is the number of observed vehicle performance parameters.

[0029] After the neural network KAN is trained, some edges between neurons are below the threshold and will be pruned, making the communication of the neural network KAN more efficient. Based on the particularity of the neural network KAN, its structure after network training can be extracted into a symbolic formula and used for the optimization of subsequent simulation system design parameters.

[0030] Step 104, deriving a nonlinear equation group between key vehicle design parameters and each vehicle performance parameter based on the network structure of the trained neural network KAN; a nonlinear equation in the nonlinear equation group is a nonlinear relationship formula between a vehicle performance parameter and each key vehicle design parameter.

[0031] The symbolic formula obtained by the trained neural network KAN is used to recommend the simulation design parameters of the simulation system, which is derived from the KAT theorem on which the neural network KAN is based. Therefore, for each vehicle performance parameter, the trained neural network KAN can extract a nonlinear formula calculated by the key vehicle design parameters, thereby obtaining a set of nonlinear equations for each vehicle performance parameter.

[0032] Step 105, taking the vehicle performance parameters satisfying the set targets as the parameter optimization target, solving the nonlinear equations using a multi-objective optimization algorithm to obtain the optimal design parameter values, that is, the optimal numerical combination of key vehicle design parameters.

[0033] In implementing the above steps 101 to 105, the present application takes into account the large amount of expert experience and design knowledge contained in the existing vehicle established parameter settings and the data generated by multiple simulations, which has a certain significance for vehicle development. Therefore, the present application makes full use of these existing vehicle simulation data, mines and reuses the knowledge and rules implied in the relationship between vehicle design parameters and vehicle performance, thereby reducing the actual number of simulations, thereby improving the efficiency of iterative design and reducing the cost of simulation development. In addition, key vehicle design parameters are extracted from numerous vehicle design parameters, reducing redundant design parameters, and improving the computational efficiency of simulation parameter decision making. In addition, the present application uses key vehicle design parameters and vehicle performance parameters output by simulation to train the neural network KAN, characterizes the simulation process with a deep learning algorithm, transforms the simulation parameter recommendation problem into a multi-objective optimization problem, and solves the global optimal solution of the simulation process. Since the training of the neural network KAN is not subject to the particularity of the simulation system, and the spline function activation mechanism in the neural network enables it to more accurately capture the nonlinear relationship between vehicle design parameters and vehicle performance parameters, improving the fitting accuracy. Therefore, compared with the existing simulation decision-making method, the present invention not only ensures the efficiency of simulation parameter decision making, but also ensures the accuracy of simulation parameter decision making.

[0034] In another exemplary embodiment of the present application, for step 101, the vehicle design parameters of the simulation input and the vehicle performance parameters of the simulation output are extracted from the historical vehicle design modeling simulation data, and the parameter data is cleaned to eliminate samples with missing values ​​and abnormal values, so as to get rid of the influence of abnormal data on the subsequent training process. The vehicle performance parameters are then standardized and converted into a unified unit format to facilitate subsequent data mining. Therefore, step 101, obtaining the simulation data of historical vehicle design modeling, extracting the vehicle design parameters and vehicle performance parameters from the simulation data, and obtaining the design parameter set and the performance parameter set, specifically includes the following steps.

[0035] (1) Obtain simulation data for historical vehicle design modeling.

[0036] (2) Preprocessing the simulation data to obtain preprocessed simulation data.

[0037] (3) Standardizing the preprocessed simulation data to obtain standardized simulation data.

[0038] (4) Extracting vehicle design parameters and vehicle performance parameters from the standardized simulation data to obtain a design parameter set and a performance parameter set.

[0039] In another exemplary embodiment of the present application, for step 102, the key parameter selection attempts to find the key vehicle design parameters that affect the output of the simulation results, and eliminates the redundant vehicle design parameters that have no effect on the simulation results. The vehicle design parameters are mainly screened by considering the correlation between the vehicle design parameters and between the vehicle design parameters and the target variable (vehicle performance parameter) at the same time, and the CFS (Correlation-based Feature Selection) method is adopted, that is, a heuristic feature selection method based on statistical measurement, which is based on the following assumption: a good feature subset contains features that are highly correlated with the results but unrelated to each other. The CFS method first calculates the correlation matrix between the vehicle design parameters and the vehicle performance parameters and between the vehicle design parameters from the design parameter set; in the process of feature selection, the CFS method aims to find a vehicle design parameter subset S, which contains k vehicle design parameters. These vehicle design parameters should have a high correlation with the target variable and a low correlation with each other. The CFS method uses an evaluation function (Merit) to evaluate the quality of the vehicle design parameter subset S, so the Merit (evaluation value) of the vehicle design parameter subset S composed of k vehicle design parameters can be calculated. The CFS method will traverse all possible vehicle design parameter subsets and calculate their Merit evaluation values, and finally select the vehicle design parameter subset with the highest Merit evaluation value as the final key design parameter set.

[0040] The form of the Merit function is shown below.

[0041] ; in, is the average value of the correlation between all vehicle design parameters and vehicle performance parameters; is the average value of the correlations between all vehicle design parameters and vehicle design parameters; Denote a subset S consisting of k vehicle design parameters.

[0042] The CFS method obtains a better subset of vehicle design parameters by continuously iteratively updating the subset of vehicle design parameters and calculating its Merit evaluation value. The iterative update of the subset of vehicle design parameters uses the best first search to search the space of the subset of vehicle design parameters. At the beginning, the subset of vehicle design parameters consists of all possible single vehicle design parameters; the estimated value of the vehicle design parameter (represented by the Merit evaluation value) is calculated, and the vehicle design parameter with the largest Merit evaluation value is selected to enter the subset S, and then the vehicle design parameter with the second largest Merit evaluation value is selected to enter the subset S. If the Merit evaluation value of the subset composed of these two vehicle design parameters is less than the Merit evaluation value of the original subset, the vehicle design parameter with the second largest Merit evaluation value is not added to the subset S, and then the next vehicle design parameter is selected. In this way, the combination of vehicle design parameters that maximizes the Merit evaluation value is obtained, which is the key design parameter set. Therefore, in step 102, the CFS method is applied to screen out the key vehicle design parameters from the design parameter set to obtain the key design parameter set, which specifically includes: (1) Calculating a first correlation between the key vehicle design parameter and the vehicle performance parameter and a second correlation between the key vehicle design parameters.

[0043] (2) Calculating a Merit evaluation value of each vehicle design parameter in the design parameter set according to the first correlation and the second correlation.

[0044] (3) Filter out the vehicle design parameters with the largest Merit evaluation value as key vehicle design parameters. Add the selected key vehicle design parameters to the key design parameter set, and calculate the Merit evaluation value of the current key design parameter set; initially, the key design parameter set is an empty set.

[0045] (4) The vehicle design parameter with the largest Merit evaluation value in the current remaining design parameter set and all the key vehicle design parameters in the current key design parameter set are taken as a temporary subset, and the Merit evaluation value of the temporary subset is calculated; the remaining design parameter set refers to the set of parameters in the design parameter set except the current key vehicle design parameters.

[0046] (5) Determine whether the Merit evaluation value of the temporary subset is greater than the Merit evaluation value of the current key design parameter set.

[0047] (6) If so, the temporary subset is used as the current key design parameter set, and the process returns to step “using the vehicle design parameter with the largest Merit evaluation value in the current remaining design parameter set and all the key vehicle design parameters in the current key design parameter set as the temporary subset”.

[0048] (7) If not, the parameters in the current remaining design parameter set excluding the vehicle design parameters added to the temporary subset constitute the current remaining design parameter set, and return to the step of "taking the vehicle design parameters with the largest Merit evaluation value in the current remaining design parameter set and all the key vehicle design parameters in the current key design parameter set as the temporary subset".

[0049] (8) Until the maximum value of the Merit evaluation value of the key design parameter set is found, the final key design parameter set is obtained.

[0050] In another exemplary embodiment of the present application, in order to prevent the CFS method from falling into a local optimal solution, a K-fold cross-validation is performed on the original design parameter set, that is, the original design parameter set is divided into K subsets of equal size. Each time, one of the subsets is selected as the validation set, and the remaining K-1 subsets are used as training sets. This process is repeated K times, and a different subset is selected as the validation set each time. In each iteration, the MLP model is trained using the training set, and the performance of the MLP model is evaluated using the validation set, and the performance evaluation indicator uses the mean absolute error MAE. According to the results of cross-validation, the impact of different training sets on the performance of the MLP model is analyzed. Usually, the training set with the best performance on the validation set is selected to compare with the performance of the key design parameter set to evaluate the performance of the key design parameter set. If the performance of the training set with the best performance on the validation set is better than the performance of the key design parameter set, the training set with the best performance on the validation set will be used as the key design parameter set.

[0051] Therefore, after executing step 102 "applying the CFS method to filter out key vehicle design parameters from the design parameter set to obtain a key design parameter set", the simulation parameter decision method based on deep learning characterization also includes: evaluating the key design parameter set, specifically including the following steps.

[0052] (1) Divide the design parameter set into K subsets.

[0053] (2) Each subset is used as a validation set, and the remaining K-1 subsets are used as training sets.

[0054] (3) Using the training set to train the MLP model, and obtaining a first trained MLP model.

[0055] (4) Using the validation set to validate the first trained MLP model, and obtaining a first validation evaluation indicator.

[0056] (5) Determine the optimal first verification evaluation indicator.

[0057] (6) Using the key design parameter set as a training set, training the MLP model to obtain a second trained MLP model.

[0058] (7) Using the remaining design parameter set as a validation set, verify the second trained MLP model and obtain a second validation evaluation indicator.

[0059] (8) Determine whether the second verification evaluation indicator is better than the optimal first verification evaluation indicator.

[0060] (9) If so, retain the critical design parameter set.

[0061] (10) If not, use the training set corresponding to the optimal first verification evaluation indicator as the key design parameter set.

[0062] In another exemplary embodiment of the present application, the problem of solving the optimal vehicle design parameters of the simulation system becomes the problem of seeking the optimal solution of the nonlinear equation group, that is, it can be solved using a multi-objective optimization algorithm such as a genetic algorithm and a particle swarm algorithm. Therefore, step 105, taking each vehicle performance parameter satisfying the set target as the parameter optimization target, using a multi-objective optimization algorithm to solve the nonlinear equation group to obtain the optimal vehicle design parameter value, specifically includes the following steps.

[0063] (1) The objective function is determined by taking the performance parameters of each vehicle as the parameter optimization target and satisfying the set target.

[0064] (2) Solving the nonlinear equations by using the genetic algorithm or the particle swarm algorithm to obtain optimal vehicle design parameter values; in the genetic algorithm or the particle swarm algorithm, the individuals in the population refer to the numerical combinations of the key vehicle design parameters that satisfy the nonlinear equations and the objective function.

[0065] In this embodiment, the key vehicle design parameters are first selected by the vehicle design parameter mining method, and the performance of the key design parameter set is evaluated by cross-validation. Then, based on these parameters, the simulation process is characterized by a deep learning algorithm. After the simulation parameter recommendation problem is converted into a multi-objective optimization problem, the global optimal solution of the simulation process is solved to explore the underlying knowledge in the massive simulation data and provide support for engineers' vehicle design parameter decisions. Among them, the key vehicle design parameter selection step can reduce the problem of too long model training time caused by redundant parameters, and eliminate design parameters that do not affect the results, thereby improving the model training efficiency. Cross-validation evaluation can prevent the selected key vehicle design parameters from falling into local optimality and ensure that the parameters input into the neural network are representative. The simulation process is characterized by deep learning, the deep learning neural network is trained using key vehicle design parameters, and the symbolic formula is extracted using an interpretable deep learning model, laying the foundation for solving the optimal vehicle design parameters. Finally, the symbolic formula obtained by deep learning is used for vehicle design parameter recommendation.

[0066] In this embodiment, deep learning is an end-to-end learning process, and the training of the model is not subject to the particularity of the simulation system. The spline function activation mechanism in the neural network enables it to more accurately capture the nonlinear relationship in the data and improve the fitting accuracy. For noise and external interference in the data, this application adopts different technical means to improve the noise robustness of the method, such as the early data preprocessing method, and when processing high-dimensional data, the neural network KAN can remove noise while retaining key information by decomposing complex functions, thereby improving the efficiency of subsequent analysis.

[0067] The present application also provides an application scenario, which applies the above-mentioned simulation parameter decision method based on deep learning representation. Specifically: the simulation parameter decision method based on deep learning representation provided in this embodiment can be applied in a vehicle design scenario. The scenario includes a data acquisition link, a data processing link and a vehicle design link; the data acquisition link is used to obtain simulation data for historical vehicle design modeling; the data processing link is used to process the simulation data to obtain the optimal vehicle design parameter value; the vehicle design link is used to design the vehicle with reference to the optimal vehicle design parameter value. The simulation parameter decision method based on deep learning representation provided in this embodiment belongs to the data processing link.

[0068] Based on the same inventive concept, the embodiment of the present application also provides a simulation parameter decision device based on deep learning representation for implementing the simulation parameter decision method based on deep learning representation involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more simulation parameter decision device embodiments based on deep learning representation provided below can refer to the limitations of the simulation parameter decision method based on deep learning representation above, and will not be repeated here.

[0069] In an exemplary embodiment, Figure 4 As shown, a simulation parameter decision device based on deep learning representation is provided, which includes the following modules.

[0070] The simulation data acquisition module M1 is used to acquire simulation data of historical vehicle design modeling, extract vehicle design parameters and vehicle performance parameters from the simulation data, and obtain a design parameter set and a performance parameter set.

[0071] The key vehicle design parameter screening module M2 is used to screen out key vehicle design parameters from the design parameter set using the CFS method to obtain a key design parameter set.

[0072] The neural network KAN training module M3 is used to train the neural network KAN using the key vehicle design parameters in the key design parameter set as network input and the corresponding vehicle performance parameters in the performance parameter set as network output to obtain a trained neural network KAN.

[0073] The nonlinear equation group construction module M4 is used to obtain a nonlinear equation group between key vehicle design parameters and various vehicle performance parameters based on the network structure of the trained neural network KAN; a nonlinear equation in the nonlinear equation group is a nonlinear relationship formula between a vehicle performance parameter and various key vehicle design parameters.

[0074] The solving module M5 is used to solve the nonlinear equations by using a multi-objective optimization algorithm with the performance parameters of each vehicle meeting the set target as the parameter optimization target, so as to obtain the optimal vehicle design parameter value.

[0075] In an exemplary embodiment, a computer device (i.e. Figure 5 The computer device in the figure can be a server or a terminal, and its internal structure diagram can be as follows: Figure 5As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store optimal design parameter value data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a simulation parameter decision method based on deep learning representation is implemented.

[0076] Those skilled in the art will understand that Figure 5 The structure shown is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components. In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the above-mentioned simulation parameter decision method based on deep learning representation is implemented.

[0077] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, which, when executed by a processor, implements the above-mentioned simulation parameter decision method based on deep learning representation.

[0078] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0079] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0080] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0081] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0082] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A simulation parameter decision method based on deep learning representation, characterized in that: The simulation parameter decision method based on deep learning representation includes: Acquire simulation data of historical vehicle design modeling, extract vehicle design parameters and vehicle performance parameters from the simulation data, and obtain a design parameter set and a performance parameter set; Applying the CFS method to select key vehicle design parameters from the design parameter set to obtain a key design parameter set; Taking each key vehicle design parameter in the key design parameter set as a network input and each corresponding vehicle performance parameter in the performance parameter set as a network output, training the neural network KAN to obtain a trained neural network KAN; A nonlinear equation group between key vehicle design parameters and various vehicle performance parameters is obtained according to the network structure of the trained neural network KAN; a nonlinear equation in the nonlinear equation group is a nonlinear relationship formula between a vehicle performance parameter and various key vehicle design parameters; Taking the performance parameters of each vehicle satisfying the set targets as the parameter optimization target, a multi-objective optimization algorithm is used to solve the nonlinear equation group to obtain the optimal vehicle design parameter values.

2. The simulation parameter decision method based on deep learning representation according to claim 1, characterized in that: Acquire simulation data of historical vehicle design modeling, extract vehicle design parameters and vehicle performance parameters from the simulation data, and obtain a design parameter set and a performance parameter set, specifically including: Access simulation data for historical vehicle design modeling; Preprocessing the simulation data to obtain preprocessed simulation data; Performing standardization processing on the preprocessed simulation data to obtain standardized simulation data; Vehicle design parameters and vehicle performance parameters are extracted from the standardized simulation data to obtain a design parameter set and a performance parameter set.

3. The simulation parameter decision method based on deep learning representation according to claim 1, characterized in that: The CFS method is applied to screen out key vehicle design parameters from the design parameter set to obtain a key design parameter set, which specifically includes: calculating a first correlation between a key vehicle design parameter and the vehicle performance parameter and a second correlation between the key vehicle design parameters; Calculate a Merit evaluation value of each vehicle design parameter in the design parameter set according to the first correlation and the second correlation; The vehicle design parameters with the largest Merit evaluation value are selected as key vehicle design parameters; the selected key vehicle design parameters are added to the key design parameter set, and the Merit evaluation value of the current key design parameter set is calculated; initially, the key design parameter set is an empty set; The vehicle design parameter with the largest Merit evaluation value in the current remaining design parameter set and all the key vehicle design parameters in the current key design parameter set are taken as a temporary subset, and the Merit evaluation value of the temporary subset is calculated; the remaining design parameter set refers to a set of parameters in the design parameter set except the current key vehicle design parameters; Determine whether the Merit evaluation value of the temporary subset is greater than the Merit evaluation value of the current key design parameter set; If yes, the temporary subset is used as the current key design parameter set, and the process returns to step "using the vehicle design parameter with the largest Merit evaluation value in the current remaining design parameter set and all key vehicle design parameters in the current key design parameter set as the temporary subset"; If not, the parameters in the current remaining design parameter set excluding the vehicle design parameters added to the temporary subset constitute the current remaining design parameter set, and return to the step of "taking the vehicle design parameter with the largest Merit evaluation value in the current remaining design parameter set and all the key vehicle design parameters in the current key design parameter set as the temporary subset"; Until the maximum value of the Merit evaluation value of the key design parameter set is found, the final key design parameter set is obtained.

4. The simulation parameter decision method based on deep learning representation according to claim 3 is characterized in that: After executing the step of "using the CFS method to screen out key vehicle design parameters from the design parameter set to obtain a key design parameter set", the simulation parameter decision method based on deep learning representation further includes: evaluating the key design parameter set, specifically: Dividing the design parameter set into K subsets; Each subset is used as a validation set, and the remaining K-1 subsets are used as training sets; Using the training set to train the MLP model, to obtain a first trained MLP model; Using the validation set to validate the first trained MLP model, and obtaining a first validation evaluation index; Determining an optimal first verification evaluation metric; Using the key design parameter set as a training set, training an MLP model to obtain a second trained MLP model; Using the remaining design parameter set as a validation set, validating the second trained MLP model, and obtaining a second validation evaluation index; Determining whether the second verification evaluation indicator is better than the optimal first verification evaluation indicator; If yes, retaining the key design parameter set; If not, the training set corresponding to the optimal first verification evaluation indicator is used as the key design parameter set.

5. The simulation parameter decision method based on deep learning representation according to claim 1 is characterized in that: Taking the performance parameters of each vehicle meeting the set target as the parameter optimization target, the nonlinear equations are solved by using a multi-objective optimization algorithm to obtain the optimal vehicle design parameter values, including: Taking the performance parameters of each vehicle meeting the set target as the parameter optimization target, determine the objective function; The nonlinear equations are solved by using a genetic algorithm or a particle swarm algorithm to obtain optimal vehicle design parameter values; in the genetic algorithm or the particle swarm algorithm, the individuals in the population refer to the numerical combinations of the key vehicle design parameters that satisfy the nonlinear equations and the objective function.

6. A simulation parameter decision device based on deep learning representation, characterized in that: The simulation parameter decision device based on deep learning representation includes: A simulation data acquisition module is used to acquire simulation data of historical vehicle design modeling, extract vehicle design parameters and vehicle performance parameters from the simulation data, and obtain a design parameter set and a performance parameter set; A key vehicle design parameter screening module, used for screening key vehicle design parameters from the design parameter set using a CFS method to obtain a key design parameter set; A neural network KAN training module is used to train the neural network KAN using each key vehicle design parameter in the key design parameter set as a network input and each corresponding vehicle performance parameter in the performance parameter set as a network output to obtain a trained neural network KAN; A nonlinear equation group construction module is used to obtain a nonlinear equation group between key vehicle design parameters and various vehicle performance parameters according to the network structure of the trained neural network KAN; a nonlinear equation in the nonlinear equation group is a nonlinear relationship formula between a vehicle performance parameter and various key vehicle design parameters; The solution module is used to solve the nonlinear equations using a multi-objective optimization algorithm with the performance parameters of each vehicle meeting the set target as the parameter optimization target, so as to obtain the optimal vehicle design parameter value.

7. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the simulation parameter decision method based on deep learning representation as described in any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the simulation parameter decision method based on deep learning representation described in any one of claims 1 to 5 is implemented.

Images