Vehicle control parameter determination method and system and electronic equipment

Abstract

The invention discloses a vehicle control parameter determination method and system and electronic equipment, and the method comprises the steps: simulating a dynamic simulation model of a vehicle, and achieving the dynamic simulation of simulation models of different vehicle types under different driving conditions; besides, the prediction model is obtained based on full-connection neural network training, and the dynamic simulation model of the target vehicle model is input into the prediction model to predict the control parameters of the target vehicle model, so that determination of the vehicle control parameters in the vehicle automatic driving process under different vehicle models is realized, and the control parameters of the vehicle are predicted through the prediction model. The efficiency and accuracy of control parameter prediction are improved, and the safety and stability of the vehicle are ensured.

Description

Technical Field

[0001] The present application relates to the field of autonomous driving, and in particular to a method, system and electronic device for determining vehicle control parameters. Background Art

[0002] Autonomous driving technology refers to the technology that enables a car to drive and complete driving tasks autonomously without a human driver. In autonomous driving technology, the controller inside the vehicle automatically generates control parameters to control the vehicle to drive and complete driving tasks autonomously.

[0003] The vehicle's control parameters need to be optimized during driving or before leaving the factory. The optimization of vehicle control parameters is usually completed through an iterative cycle of manual adjustment and testing. The iterative efficiency and adjustment effect are limited by manual labor. Summary of the invention

[0004] In view of this, the present application provides a vehicle control parameter determination method, system and electronic device, and the specific scheme thereof is as follows:

[0005] A method for determining a vehicle control parameter, comprising:

[0006] Obtaining dynamic parameters of a target vehicle model for which control parameters are to be determined;

[0007] Perform simulation based on the dynamic parameters of the target vehicle model to obtain a dynamic simulation model of the target vehicle model;

[0008] Inputting the dynamics simulation model of the target vehicle model into a pre-trained prediction model to obtain an output result of the prediction model, wherein the output result can characterize the prediction control parameters of the target vehicle model;

[0009] The prediction model is obtained by training a fully connected neural network in advance through dynamic simulation models of multiple sample vehicle models and control parameters of each sample vehicle model in the multiple sample vehicle models.

[0010] Furthermore, it also includes:

[0011] training the prediction model;

[0012] Wherein, the training of the prediction model comprises:

[0013] Obtain dynamic simulation models of multiple sample vehicle models and control parameters of each sample vehicle model;

[0014] Build an initial fully connected neural network model;

[0015] Determine the number of model layers and nodes;

[0016] The initial fully connected neural network model is trained based on the dynamics simulation models of the multiple sample vehicle models and the control parameters of each sample vehicle model to obtain a trained prediction model that meets the number of model layers and nodes.

[0017] Furthermore, obtaining the control parameters of each sample vehicle model includes:

[0018] Obtain multiple groups of control parameters for each sample vehicle model;

[0019] Based on the fitness function, multiple groups of control parameters of each sample vehicle model are optimized to obtain optimized control parameters of each sample vehicle model, so as to use the optimized control parameters for model training.

[0020] Furthermore, the optimization of the multiple control parameters of each sample vehicle model based on the fitness function to obtain the optimized control parameters of each sample vehicle model after optimization includes:

[0021] Determine the fitness function;

[0022] Using the fitness function to evaluate the multiple groups of control parameters of each of the sample vehicle models, respectively, to obtain a first evaluation result corresponding to each group of control parameters in the multiple groups of control parameters;

[0023] The multiple groups of control parameters are optimized based on the first evaluation results respectively corresponding to each group of control parameters in the multiple groups of control parameters to obtain a group of optimized control parameters for each sample vehicle model.

[0024] Furthermore, the optimizing the multiple groups of control parameters based on the first evaluation results respectively corresponding to each group of control parameters in the multiple groups of control parameters to obtain a group of optimized control parameters for each sample vehicle model includes:

[0025] Based on the first evaluation results respectively corresponding to each group of control parameters in the multiple groups of control parameters, selecting a target number of control parameters whose first evaluation results satisfy a first target condition from the multiple groups of control parameters;

[0026] Cross-combining the control parameters of the selected target groups to obtain multiple groups of new control parameters;

[0027] Using the fitness function to evaluate the multiple groups of new control parameters respectively, to obtain second evaluation results corresponding to each group of new control parameters;

[0028] Based on the second evaluation result, a control parameter satisfying a second target condition is determined from the multiple groups of new control parameters as an optimized control parameter.

[0029] Further, the determining, based on the second evaluation result, from the multiple groups of new control parameters, a control parameter that meets the second target condition as an optimized control parameter includes:

[0030] Determine a control parameter that meets a second target condition from the multiple groups of new control parameters based on the second evaluation result, and determine the control parameter that meets the second target condition as a control parameter to be determined;

[0031] Using a dynamics simulation model of a sample vehicle model corresponding to the control parameter to be determined, the control parameter to be determined is tested to determine whether the control parameter to be determined satisfies the driving condition of the sample vehicle model corresponding to the control parameter to be determined;

[0032] If it is determined that the control parameter to be determined satisfies the driving condition of the sample vehicle model corresponding to the control parameter to be determined, the control parameter to be determined is determined as the optimized control parameter.

[0033] A vehicle control parameter determination system, comprising:

[0034] An obtaining unit, used for obtaining dynamic parameters of a target vehicle model for which control parameters are to be determined;

[0035] A simulation unit, used for performing simulation based on the dynamic parameters of the target vehicle model to obtain a dynamic simulation model of the target vehicle model;

[0036] A prediction unit, used for inputting the dynamics simulation model of the target vehicle model into a pre-trained prediction model to obtain an output result of the prediction model, wherein the output result can characterize the prediction control parameters of the target vehicle model;

[0037] The prediction model is obtained by training a fully connected neural network in advance through dynamic simulation models of multiple sample vehicle models and control parameters of each sample vehicle model in the multiple sample vehicle models.

[0038] Furthermore, it also includes:

[0039] A training unit, used for training the prediction model;

[0040] Wherein, the training unit is used for:

[0041] Obtain dynamic simulation models of multiple sample vehicle models and control parameters of each sample vehicle model; construct an initial fully connected neural network model; determine the number of model layers and the number of nodes; perform model training on the initial fully connected neural network model based on the dynamic simulation models of the multiple sample vehicle models and the control parameters of each sample vehicle model, and obtain a trained prediction model that meets the number of model layers and the number of nodes.

[0042] An electronic device, comprising:

[0043] A processor, configured to obtain dynamic parameters of a target vehicle model for which control parameters are to be determined; perform simulation based on the dynamic parameters of the target vehicle model to obtain a dynamic simulation model of the target vehicle model; input the dynamic simulation model of the target vehicle model into a pre-trained prediction model to obtain an output result of the prediction model, wherein the output result can characterize the predicted control parameters of the target vehicle model; wherein the prediction model is obtained by pre-training a fully connected neural network using dynamic simulation models of multiple sample vehicle models and control parameters of each sample vehicle model in the multiple sample vehicle models;

[0044] The memory is used to store the program for the processor to execute the above processing process.

[0045] A readable storage medium for storing at least one set of instructions;

[0046] The instruction set is used to be called and at least execute the vehicle control parameter determination method as described in any one of the above items.

[0047] It can be seen from the above technical scheme that the vehicle control parameter determination method, system and electronic device disclosed in the present application obtain the dynamic parameters of the target vehicle model whose control parameters are to be determined; simulate based on the dynamic parameters of the target vehicle model to obtain the dynamic simulation model of the target vehicle model; input the dynamic simulation model of the target vehicle model into the pre-trained prediction model to obtain the output result of the prediction model, and the output result can characterize the predicted control parameters of the target vehicle model; wherein the prediction model is obtained by pre-training the dynamic simulation models of multiple sample vehicle models and the control parameters of each sample vehicle model in the multiple sample vehicle models using a fully connected neural network. This scheme simulates the dynamic simulation model of the vehicle to achieve dynamic simulation of simulation models of different vehicle models under different driving conditions; in addition, the prediction model is obtained based on the training of the fully connected neural network, and the dynamic simulation model of the target vehicle model is input into the prediction model to predict the control parameters of the target vehicle model, thereby achieving the determination of the vehicle control parameters during the automatic driving process of the vehicle under different vehicle models, and predicting the control parameters of the vehicle through the prediction model, thereby improving the efficiency and accuracy of the control parameter prediction, and ensuring the safety and stability of the vehicle. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related technologies, the drawings required for use in the embodiments or the related technical descriptions are 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.

[0049] Figure 1A flow chart of a method for determining vehicle control parameters disclosed in an embodiment of the present application;

[0050] Figure 2 A flow chart of a method for determining vehicle control parameters disclosed in an embodiment of the present application;

[0051] Figure 3 A flow chart of a method for determining vehicle control parameters disclosed in an embodiment of the present application;

[0052] Figure 4 A schematic diagram of the structure of a vehicle control parameter determination system disclosed in an embodiment of the present application;

[0053] Figure 5 A schematic diagram of the structure of an electronic device disclosed in an embodiment of the present application. DETAILED DESCRIPTION

[0054] The following describes the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. The terms used in the implementation method section of the present application are only used to explain the specific embodiments of the present application, and are not intended to limit the present application.

[0055] The embodiments of the present application are described below in conjunction with the accompanying drawings. Those skilled in the art will appreciate that, with the development of technology and the emergence of new scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0056] The terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and need not be used to describe a specific order or sequential order. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances, which is only to describe the distinction mode adopted by the objects of the same attributes when describing in the embodiments of the present application. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, so that the process, method, system, product or equipment comprising a series of units need not be limited to those units, but may include other units that are not clearly listed or inherent to these processes, methods, products or equipment.

[0057] The present application discloses a method for determining vehicle control parameters, and its flow chart is as follows: Figure 1 As shown, including:

[0058] Step S11, obtaining the dynamic parameters of the target vehicle model for which the control parameters are to be determined;

[0059] Step S12: performing simulation based on the dynamic parameters of the target vehicle model to obtain a dynamic simulation model of the target vehicle model;

[0060] Step S13, inputting the dynamics simulation model of the target vehicle model into the pre-trained prediction model to obtain the output result of the prediction model, and the output result can characterize the prediction control parameters of the target vehicle model; wherein the prediction model is pre-trained using a fully connected neural network through the dynamics simulation models of multiple sample vehicle models and the control parameters of each sample vehicle model in the multiple sample vehicle models.

[0061] Autonomous driving technology refers to the technology that enables a car to drive and complete driving tasks autonomously without a human driver. In autonomous driving technology, the controller inside the vehicle automatically generates control parameters to control the vehicle to drive and complete driving tasks autonomously.

[0062] Perception is the first step in autonomous driving, which mainly involves collecting and understanding information about the surrounding environment. It can obtain information about the surrounding environment in real time through a variety of sensors to form a comprehensive understanding of the surrounding environment; positioning is to determine the precise position of the vehicle on the map; decision-making is to formulate the vehicle's driving strategy based on perception and positioning information; planning is to generate a specific driving path and motion trajectory based on the decision results; control is to convert the planned trajectory into actual accelerator, brake and steering to control the vehicle's movement.

[0063] Vehicle control is achieved through control parameters. The vehicle control parameters need to be optimized during driving or before leaving the factory. The optimization of vehicle control parameters is usually completed through an iterative cycle of manual adjustment and testing. The iterative efficiency and adjustment effect are limited by manual labor.

[0064] Based on this, in this solution, the dynamics simulation model of the target vehicle model is determined based on the dynamics parameters of the target vehicle model, and is input into the prediction model to obtain the predicted control parameters of the target vehicle model, wherein the prediction model is obtained by training the dynamics simulation models of multiple sample vehicle models and the control parameters of each sample vehicle model using a fully connected neural network. The control parameters of the target vehicle model are predicted by the prediction model, which improves the prediction efficiency and accuracy of the control parameters and avoids the problem of accuracy and efficiency being limited by manual labor caused by manual debugging and testing to determine the control parameters in the prior art.

[0065] Specifically, first determine the target vehicle model for which the control parameters are to be determined. The target vehicle model may be a new vehicle model, i.e., there were no control parameters for this vehicle model before, and the control parameters of the new vehicle model need to be determined based on this solution, and the new vehicle model is determined as the target vehicle model; alternatively, the target vehicle model is a vehicle model that needs to have its control parameters updated, i.e., the vehicle model already had control parameters before, but currently needs to have the control parameters updated, and therefore, the vehicle model can be determined as the target vehicle model.

[0066] After determining the target vehicle model, the dynamic parameters of the target vehicle model are collected. The dynamic parameters may include vehicle parameters and tire model parameters.

[0067] Taking the bicycle model as an example, the vehicle parameters may include: vehicle Z-axis moment of inertia, vehicle gross mass, vehicle wheelbase, center of mass position, distance from center of mass to vehicle front axle, distance from center of mass to vehicle rear axle, etc. Among them, vehicle Z-axis moment of inertia can reflect the vehicle's ability to rotate in the Z-axis direction and affect the vehicle's handling stability; vehicle gross mass can directly affect the vehicle's acceleration and braking performance; vehicle wheelbase, that is, the distance between the front and rear axles of the vehicle, can affect the vehicle's steering performance and stability; center of mass position, that is, the distance between the center of mass and the front and rear axles, can affect the vehicle's dynamic response characteristics.

[0068] The tire model parameters mainly include the corresponding relationship curve between the tire lateral force and the sideslip angle. This curve describes the lateral force generated by the tire at different sideslip angles and can directly affect the vehicle's handling performance and stability.

[0069] After obtaining the dynamic parameters of the target vehicle model, the target vehicle model can be simulated to obtain the dynamic simulation model of the target vehicle model. Specifically, the target vehicle model can be simulated through simulation software, and the dynamic parameters of the target vehicle model can be input into the simulation software to obtain the dynamic simulation model of the target vehicle model. The dynamic simulation model of the target vehicle model can simulate the movement under various driving conditions in the simulation software.

[0070] By inputting the dynamic simulation model of the target vehicle model into the pre-trained prediction model, the output result of the prediction model can be obtained. The output result of the prediction model is the predicted control parameter of the target vehicle model, thereby realizing the prediction of the control parameters of the target vehicle model.

[0071] The prediction model is obtained in advance through the dynamic simulation models of multiple sample vehicle models and the control parameters of each sample vehicle model, using a fully connected neural network for training.

[0072] Among them, the fully connected neural network, namely the multilayer perceptron MLP (Multilayer Perceptron), is a fully connected feedforward neural network model, which is composed of a combination of multiple perceptron models. The parameters to be learned of the overall model are the sum of the parameters of each layer of perceptron.

[0073] In addition, in this embodiment, the following can also be done: when training the prediction model, the dynamic parameters of multiple sample vehicle models and the control parameters of each sample vehicle model are directly used, and the prediction model is obtained by training a fully connected neural network. When using the prediction model to predict the control parameters of a target vehicle model, the dynamic parameters of the target vehicle model are directly input into the prediction model to obtain the predicted control parameters of the target vehicle model. In this process, there is no need to perform model simulation on the vehicle model, thereby achieving the purpose of simplifying the operation steps.

[0074] In this embodiment, by systematically collecting the dynamic parameters of the target vehicle model including vehicle parameters and tire model parameters, accurate vehicle dynamic characteristic data can be obtained, providing a reliable basis for modeling and simulation, and ensuring that the simulation model can truly reflect the performance of the vehicle in actual driving; in addition, the prediction model obtained by fully connected neural network training can input the new dynamic simulation model into the prediction model when the dynamics in the vehicle change or a new model needs to be installed, and can quickly adapt to the vehicle characteristics of the new model, so that the vehicle can be efficiently and accurately controlled during the automatic driving process.

[0075] The vehicle control parameter determination method disclosed in this embodiment obtains the dynamic parameters of the target vehicle model whose control parameters are to be determined; performs simulation based on the dynamic parameters of the target vehicle model to obtain the dynamic simulation model of the target vehicle model; inputs the dynamic simulation model of the target vehicle model into a pre-trained prediction model to obtain the output result of the prediction model, and the output result can characterize the predicted control parameters of the target vehicle model; wherein the prediction model is obtained by pre-training the dynamic simulation models of multiple sample vehicle models and the control parameters of each sample vehicle model in the multiple sample vehicle models using a fully connected neural network. This solution simulates the dynamic simulation model of the vehicle to achieve dynamic simulation of simulation models of different vehicle models under different driving conditions; in addition, the prediction model is obtained based on the training of the fully connected neural network, and the dynamic simulation model of the target vehicle model is input into the prediction model to predict the control parameters of the target vehicle model, thereby achieving the determination of vehicle control parameters during the automatic driving process of the vehicle under different vehicle models, and predicting the control parameters of the vehicle through the prediction model improves the efficiency and accuracy of the control parameter prediction, and ensures the safety and stability of the vehicle.

[0076] This embodiment discloses a method for determining vehicle control parameters, and its flow chart is as follows: Figure 2 As shown, including:

[0077] Step S21, training a prediction model;

[0078] Step S22, obtaining the dynamic parameters of the target vehicle model for which the control parameters are to be determined;

[0079] Step S23, performing simulation based on the dynamic parameters of the target vehicle model to obtain a dynamic simulation model of the target vehicle model;

[0080] Step S24, inputting the dynamics simulation model of the target vehicle model into the pre-trained prediction model to obtain the output result of the prediction model, and the output result can characterize the prediction control parameters of the target vehicle model; wherein the prediction model is pre-trained using a fully connected neural network through the dynamics simulation models of multiple sample vehicle models and the control parameters of each sample vehicle model in the multiple sample vehicle models.

[0081] Pre-training the prediction model may include:

[0082] Obtain dynamic simulation models of multiple sample vehicle models and control parameters of each sample vehicle model; construct an initial fully connected neural network model; determine the number of model layers and the number of nodes; perform model training on the initial fully connected neural network model based on the dynamic simulation models of multiple sample vehicle models and the control parameters of each sample vehicle model, and obtain a trained prediction model that meets the number of model layers and the number of nodes.

[0083] When performing model training, training data is first prepared, and the training data includes dynamic simulation models of multiple sample vehicle models as training samples and control parameters of each sample vehicle model.

[0084] Among them, the dynamics simulation model of each sample vehicle model can be obtained in the following way: collect the dynamics parameters of each sample vehicle model, including: vehicle parameters and tire model parameters of each sample vehicle model, and then input the dynamics parameters of each sample vehicle model into the simulation software, simulate the sample vehicle model, and then obtain the dynamics simulation model of the sample vehicle model. In this way, the dynamics simulation model of each sample vehicle model among multiple sample vehicle models is obtained.

[0085] The control parameters of each sample vehicle model can be obtained by collecting the control parameters of each sample vehicle model during driving, and determining the dynamic response of the dynamic simulation model corresponding to different sample vehicle models under corresponding control parameters.

[0086] After obtaining the training data required for model training, the model needs to be constructed. During the model construction process, the initial fully connected neural network model, that is, the initial MLP model, needs to be constructed. In addition, the number of model layers and nodes needs to be determined.

[0087] After the model is built, the initial MLP model is trained using the training data to adjust the model parameters so that the prediction model finally trained can meet the requirements of the number of model layers and nodes, and can meet a certain prediction accuracy while meeting the above requirements.

[0088] In addition, in the vehicle control parameter determination method disclosed in this embodiment, other deep learning models or other types of neural network models, or non-neural network models, such as support vector machines, polynomial regression, etc., can also be used. However, the MLP model in this embodiment has the advantages of no need for data conversion, time independence, and simple model compared to other neural network models.

[0089] In this embodiment, by simulating the dynamic simulation models of multiple sample vehicle models, the dynamic behaviors of different vehicle models under various driving conditions are simulated; and, by training the MLP model, it is able to learn the relationship between complex vehicle dynamic behaviors and control parameters, improve the prediction ability, and achieve accurate prediction and dynamic adaptation of vehicle control parameters, which not only improves the responsiveness of the vehicle's automatic driving system in various complex environments, but also significantly reduces the probability of accidents and improves the overall safety and reliability.

[0090] The vehicle control parameter determination method disclosed in this embodiment includes training a prediction model; obtaining the dynamic parameters of a target vehicle model whose control parameters are to be determined; performing simulation based on the dynamic parameters of the target vehicle model to obtain a dynamic simulation model of the target vehicle model; inputting the dynamic simulation model of the target vehicle model into a pre-trained prediction model to obtain an output result of the prediction model, and the output result can characterize the predicted control parameters of the target vehicle model; wherein the prediction model is obtained by pre-training a fully connected neural network through dynamic simulation models of multiple sample vehicle models and control parameters of each sample vehicle model in the multiple sample vehicle models. This solution trains the prediction model in advance so that the trained prediction model can learn the relationship between complex vehicle dynamic behaviors and control parameters, and can quickly output corresponding control parameters when facing a dynamic simulation model corresponding to a new vehicle model, thereby improving the intelligent level of the vehicle's automatic driving and ensuring the safe driving of the vehicle during the automatic driving process.

[0091] This embodiment discloses a method for determining vehicle control parameters, and its flow chart is as follows: Figure 3 As shown, including:

[0092] Step S31, obtaining dynamic simulation models of multiple sample vehicle models and multiple groups of control parameters for each sample vehicle model;

[0093] Step S32: Optimizing multiple groups of control parameters of each sample vehicle model based on the fitness function to obtain optimized control parameters of each sample vehicle model;

[0094] Step S33, constructing an initial fully connected neural network model;

[0095] Step S34, determining the number of model layers and nodes;

[0096] Step S35: training the initial fully connected neural network model based on the dynamics simulation models of multiple sample vehicle models and the optimized control parameters of each sample vehicle model to obtain a trained prediction model that meets the number of model layers and nodes;

[0097] Step S36, obtaining the dynamic parameters of the target vehicle model for which the control parameters are to be determined;

[0098] Step S37, performing simulation based on the dynamic parameters of the target vehicle model to obtain a dynamic simulation model of the target vehicle model;

[0099] Step S38, input the dynamics simulation model of the target vehicle model into the pre-trained prediction model to obtain the output result of the prediction model, and the output result can characterize the prediction control parameters of the target vehicle model; the prediction model is pre-trained using a fully connected neural network through the dynamics simulation models of multiple sample vehicle models and the control parameters of each sample vehicle model in the multiple sample vehicle models.

[0100] If there are control parameters of a target vehicle model to be predicted, the dynamics simulation model of the target vehicle model needs to be input into the prediction model in order to obtain the predicted control parameters of the target vehicle model output by the prediction model. The prediction model is pre-trained using the dynamics simulation models of multiple sample vehicle models and the control parameters of each sample vehicle model.

[0101] Among them, when training the prediction model, the control parameters of each sample model among multiple sample models are obtained. The control parameters of each sample model can be collected from multiple vehicles of the same model, so there will be multiple groups of control parameters, and each group of control parameters is used to control the dynamic behavior of the vehicle of this model.

[0102] Multiple sets of control parameters are obtained for each sample vehicle model. Afterwards, the multiple sets of control parameters for each sample vehicle model can be optimized using a fitness function to obtain optimized control parameters for each sample vehicle model, and the optimized control parameters are used for model training so that when the model is used to predict the control parameters, the predicted control parameters can be more accurate.

[0103] Among them, the fitness function is a function used in the genetic algorithm to evaluate the quality of the results. In this embodiment, the fitness function can describe the influence of the control parameters on the vehicle posture. Therefore, the fitness function needs to be defined in this embodiment. The fitness function can be designed to evaluate the advantages and disadvantages of different groups of control parameters based on the dynamic response and control accuracy of the vehicle.

[0104] The fitness function can be used to evaluate the quality of each of the multiple control parameter groups in the process of controlling the vehicle. Each of the multiple control parameter groups can be input into the fitness function to obtain the value of the fitness function, and the quality of each control parameter group can be determined based on the value.

[0105] Each vehicle model corresponds to a dynamics simulation model, and accordingly, each dynamics simulation model corresponds to a fitness function, that is, the fitness function is related to the vehicle model.

[0106] After determining the quality of each group of control parameters based on the value of the fitness function, the control parameters of the vehicle model corresponding to the multiple groups of control parameters can be optimized based on the quality of each group of control parameters in the multiple groups of control parameters, thereby obtaining a set of optimized control parameters, which are the control parameters corresponding to the vehicle model.

[0107] Based on the quality of each set of control parameters, the control parameters of the vehicle models corresponding to the multiple sets of control parameters are optimized. The best set of control parameters is directly selected according to the quality, and it is directly determined as the optimized control parameters. Then, the optimized control parameters corresponding to each sample vehicle model obtained in this way are directly used for model training.

[0108] Specifically: determine the fitness function; use the fitness function to evaluate multiple groups of control parameters of each sample vehicle model respectively, and obtain the first evaluation results corresponding to each group of control parameters in the multiple groups of control parameters; optimize the multiple groups of control parameters based on the first evaluation results corresponding to each group of control parameters in the multiple groups of control parameters, so as to obtain a set of optimized control parameters for each sample vehicle model.

[0109] Based on the first evaluation result, multiple groups of control parameters are optimized. In addition to the above-described method of directly selecting the best group of control parameters as the optimized control parameters according to the quality of each group of control parameters in the first evaluation result, the following method can also be used:

[0110] Based on the first evaluation results respectively corresponding to each group of control parameters in the multiple groups of control parameters, a target number of control parameters whose first evaluation results satisfy the first target condition are selected from the multiple groups of control parameters; the selected target number of control parameters are cross-combined to obtain multiple groups of new control parameters; the multiple groups of new control parameters are respectively evaluated using the fitness function to obtain second evaluation results respectively corresponding to each group of new control parameters; and the control parameters satisfying the second target condition are determined from the multiple groups of new control parameters as optimized control parameters based on the second evaluation results.

[0111] After using the fitness function to evaluate each group of control parameters in the multiple groups of control parameters separately to obtain the corresponding first evaluation results, the quality of each group of control parameters can be determined based on the first evaluation results. At this time, a certain number of control parameter groups can be selected from the multiple groups of control parameters based on the quality represented by the first evaluation results, that is, the target number of control parameter groups are determined.

[0112] The target number of control parameters is a control parameter that satisfies the first target condition among multiple control parameter groups. The first target condition can be satisfied by: sorting the multiple control parameter groups according to the quality of the first evaluation result, and selecting the target number of control parameters with the highest sorting as the control parameter that satisfies the first target condition; or determining the control parameter whose quality of the first evaluation result exceeds a certain value among multiple control parameter groups as the control parameter that satisfies the first target condition. Regardless of which the first target condition is, the control parameter that satisfies the first target condition is a relatively excellent control parameter selected from the multiple control parameter groups in the first evaluation result.

[0113] After determining the target number of control parameters that meet the first target condition, it can be directly determined whether there is a set of control parameters whose first evaluation result reaches a certain target value. If so, the set of control parameters whose first evaluation result reaches the target value is directly determined as the optimized control parameters;

[0114] Alternatively, the control parameters of the selected target number of groups are crossovered and mutated to obtain multiple groups of new control parameters. After obtaining the multiple groups of new control parameters, the multiple groups of new control parameters are evaluated separately using the fitness function to obtain second evaluation results corresponding to each group of new control parameters. The second evaluation results represent the quality of each group of new control parameters, so that the multiple groups of new control parameters can be compared based on the second evaluation results to determine whether there are control parameters that meet the second target conditions.

[0115] Among them, satisfying the second target condition can be: there exists a set of new control parameters whose second evaluation results reach a certain target value. If so, the new control parameters are determined as the optimized control parameters. If not, a combination of control parameters with a higher degree of excellence can be continuously selected from multiple sets of new control parameters based on the second evaluation results, and the control parameters in the selected control parameter combination are crossovered and mutated to regenerate new control parameters, and the regenerated new control parameters are evaluated using the fitness function. Iterative optimization is performed in this way until a set of control parameters whose evaluation results reach a certain target value is determined and determined as the optimized control parameters and stops.

[0116] By determining the optimized control parameters in this way, the optimized control parameters of each sample vehicle model used for model training can be the optimal control parameter set obtained through iterative optimization, thereby ensuring the accuracy of the control parameters. When the control parameters are predicted by the prediction model obtained through model training, the prediction accuracy can be improved to ensure the driving safety of the vehicle during autonomous driving.

[0117] Furthermore, in the vehicle control parameter determination method disclosed in this embodiment, determining the control parameter that meets the second target condition from multiple groups of new control parameters based on the second evaluation result as the optimized control parameter includes:

[0118] Based on the second evaluation result, control parameters that meet the second target condition are determined from multiple groups of new control parameters, and the control parameters that meet the second target condition are determined as control parameters to be determined; using a dynamic simulation model of a sample vehicle model corresponding to the control parameters to be determined, the control parameters to be determined are tested to determine whether the control parameters to be determined meet the driving conditions of the sample vehicle model corresponding to the control parameters to be determined; if it is determined that the control parameters to be determined meet the driving conditions of the sample vehicle model corresponding to the control parameters to be determined, the control parameters to be determined are determined as optimized control parameters.

[0119] That is, after optimizing the control parameters, the optimized control parameters are determined as the control parameters to be determined, and the control parameters to be determined are further tested. Only when the test passes, they are determined as the optimized control parameters, and the optimized control parameters are used for model training.

[0120] After determining the control parameters to be determined, the simulation software is used to determine the control parameters to be determined. The simulation software can simulate various vehicle postures of the sample vehicle model. Therefore, after determining the control parameters to be determined, the control parameters to be determined are input into the simulation software, and the dynamics simulation model of the sample vehicle model corresponding to the control parameters to be determined is used to simulate the vehicle posture based on the control parameters to be determined, so as to determine that the control parameters to be determined can meet the driving conditions of the dynamics simulation model of the sample vehicle model, and determine its effectiveness and stability under different driving conditions.

[0121] The vehicle control parameter determination method disclosed in the present embodiment obtains dynamic simulation models of multiple sample vehicle models and multiple groups of control parameters for each sample vehicle model; optimizes the multiple groups of control parameters for each sample vehicle model based on a fitness function to obtain optimized control parameters for each sample vehicle model after optimization; constructs an initial fully connected neural network model; determines the number of model layers and the number of nodes; performs model training on the initial fully connected neural network model based on the dynamic simulation models of multiple sample vehicle models and the optimized control parameters for each sample vehicle model to obtain a trained prediction model that meets the number of model layers and the number of nodes; obtains dynamic parameters of a target vehicle model whose control parameters are to be determined; performs simulation based on the dynamic parameters of the target vehicle model to obtain a dynamic simulation model of the target vehicle model; inputs the dynamic simulation model of the target vehicle model into a pre-trained prediction model to obtain an output result of the prediction model, and the output result can characterize the predicted control parameters of the target vehicle model; the prediction model is obtained by pre-training the dynamic simulation models of multiple sample vehicle models and the control parameters of each sample vehicle model in the multiple sample vehicle models using a fully connected neural network. When this scheme is training the model, after obtaining the control parameters of each sample vehicle model, it is necessary to optimize multiple groups of control parameters of each sample vehicle model based on the fitness function so as to use the optimized control parameters for model training, so that when the control parameters are predicted using the prediction model obtained by model training with the optimized control parameters, the predicted control parameters can improve the safety and stability of the target vehicle during driving, thereby ensuring the driving safety of the vehicle.

[0122] This embodiment discloses a vehicle control parameter determination system, and its structural diagram is shown in FIG. Figure 4 As shown, including:

[0123] An acquisition unit 41 , a simulation unit 42 and a prediction unit 43 .

[0124] The obtaining unit 41 is used to obtain the dynamic parameters of the target vehicle model for which the control parameters are to be determined;

[0125] The simulation unit 42 is used to perform simulation based on the dynamic parameters of the target vehicle model to obtain a dynamic simulation model of the target vehicle model;

[0126] The prediction unit 43 is used to input the dynamic simulation model of the target vehicle model into the pre-trained prediction model to obtain the output result of the prediction model, and the output result can represent the prediction control parameters of the target vehicle model;

[0127] The prediction model is obtained by training a fully connected neural network in advance through a dynamic simulation model of multiple sample vehicle models and control parameters of each sample vehicle model in the multiple sample vehicle models.

[0128] Furthermore, the vehicle control parameter determination system disclosed in this embodiment may also include:

[0129] A training unit, used to train the prediction model;

[0130] Among them, the training unit is used to:

[0131] Obtain dynamic simulation models of multiple sample vehicle models and control parameters of each sample vehicle model; construct an initial fully connected neural network model; determine the number of model layers and the number of nodes; perform model training on the initial fully connected neural network model based on the dynamic simulation models of multiple sample vehicle models and the control parameters of each sample vehicle model, and obtain a trained prediction model that meets the number of model layers and the number of nodes.

[0132] Furthermore, the training unit is used to:

[0133] Obtain multiple groups of control parameters for each sample vehicle model; optimize the multiple groups of control parameters for each sample vehicle model based on the fitness function to obtain optimized control parameters for each sample vehicle model, so as to use the optimized control parameters for model training.

[0134] Furthermore, the training unit is used to:

[0135] Determine a fitness function; use the fitness function to evaluate the multiple groups of control parameters of each sample vehicle model respectively, and obtain a first evaluation result corresponding to each group of control parameters in the multiple groups of control parameters; optimize the multiple groups of control parameters based on the first evaluation result corresponding to each group of control parameters in the multiple groups of control parameters, so as to obtain a set of optimized control parameters for each sample vehicle model.

[0136] Furthermore, the training unit is used to:

[0137] Based on the first evaluation results respectively corresponding to each group of control parameters in the multiple groups of control parameters, a target number of control parameters whose first evaluation results satisfy the first target condition are selected from the multiple groups of control parameters; the selected target number of control parameters are cross-combined to obtain multiple groups of new control parameters; the multiple groups of new control parameters are respectively evaluated using the fitness function to obtain second evaluation results respectively corresponding to each group of new control parameters; and the control parameters satisfying the second target condition are determined from the multiple groups of new control parameters as optimized control parameters based on the second evaluation results.

[0138] Furthermore, the training unit is used to:

[0139] Based on the second evaluation result, control parameters that meet the second target condition are determined from multiple groups of new control parameters, and the control parameters that meet the second target condition are determined as control parameters to be determined; using a dynamic simulation model of a sample vehicle model corresponding to the control parameters to be determined, the control parameters to be determined are tested to determine whether the control parameters to be determined meet the driving conditions of the sample vehicle model corresponding to the control parameters to be determined; if it is determined that the control parameters to be determined meet the driving conditions of the sample vehicle model corresponding to the control parameters to be determined, the control parameters to be determined are determined as optimized control parameters.

[0140] The vehicle control parameter determination system disclosed in this embodiment is implemented based on the vehicle control parameter determination method disclosed in the above embodiment, which will not be described in detail here.

[0141] The vehicle control parameter determination system disclosed in this embodiment obtains the dynamic parameters of the target vehicle model whose control parameters are to be determined; performs simulation based on the dynamic parameters of the target vehicle model to obtain the dynamic simulation model of the target vehicle model; inputs the dynamic simulation model of the target vehicle model into a pre-trained prediction model to obtain the output result of the prediction model, and the output result can characterize the predicted control parameters of the target vehicle model; wherein the prediction model is obtained by pre-training the dynamic simulation models of multiple sample vehicle models and the control parameters of each sample vehicle model in the multiple sample vehicle models using a fully connected neural network. This solution simulates the dynamic simulation model of the vehicle to achieve dynamic simulation of simulation models of different vehicle models under different driving conditions; in addition, the prediction model is obtained based on the training of the fully connected neural network, and the dynamic simulation model of the target vehicle model is input into the prediction model to predict the control parameters of the target vehicle model, thereby achieving the determination of vehicle control parameters during the automatic driving process of the vehicle under different vehicle models, and predicting the control parameters of the vehicle through the prediction model improves the efficiency and accuracy of the control parameter prediction, and ensures the safety and stability of the vehicle.

[0142] This embodiment discloses an electronic device, and its structural diagram is as follows: Figure 5 As shown, including:

[0143] Processor 51 and memory 52.

[0144] The processor 51 is used to obtain the dynamic parameters of the target vehicle model whose control parameters are to be determined; perform simulation based on the dynamic parameters of the target vehicle model to obtain a dynamic simulation model of the target vehicle model; input the dynamic simulation model of the target vehicle model into a pre-trained prediction model to obtain an output result of the prediction model, and the output result can characterize the prediction control parameters of the target vehicle model; wherein the prediction model is obtained by pre-training a fully connected neural network through dynamic simulation models of multiple sample vehicle models and control parameters of each sample vehicle model in the multiple sample vehicle models;

[0145] The memory 52 is used to store the program for the processor to execute the above-mentioned processing process.

[0146] The electronic device disclosed in this embodiment is implemented based on the vehicle control parameter determination method disclosed in the above embodiment, which will not be described in detail here.

[0147] The electronic device disclosed in this embodiment obtains the dynamic parameters of the target vehicle model whose control parameters are to be determined; performs simulation based on the dynamic parameters of the target vehicle model to obtain the dynamic simulation model of the target vehicle model; inputs the dynamic simulation model of the target vehicle model into a pre-trained prediction model to obtain the output result of the prediction model, and the output result can characterize the predicted control parameters of the target vehicle model; wherein the prediction model is obtained by pre-training the dynamic simulation models of multiple sample vehicle models and the control parameters of each sample vehicle model in the multiple sample vehicle models using a fully connected neural network. This solution simulates the dynamic simulation model of the vehicle to achieve dynamic simulation of simulation models of different vehicle models under different driving conditions; in addition, the prediction model is obtained based on the training of the fully connected neural network, and the dynamic simulation model of the target vehicle model is input into the prediction model to predict the control parameters of the target vehicle model, thereby achieving the determination of the vehicle control parameters during the automatic driving process of the vehicle under different vehicle models, and predicting the control parameters of the vehicle through the prediction model improves the efficiency and accuracy of the control parameter prediction, thereby ensuring the safety and stability of the vehicle.

[0148] An embodiment of the present application also provides a readable storage medium on which a computer program is stored. The computer program is loaded and executed by a processor to implement the steps of the above-mentioned vehicle control parameter determination method. The specific implementation process can refer to the description of the corresponding part of the above-mentioned embodiment, which will not be repeated in this embodiment.

[0149] The present application also proposes a computer program product or a computer program, which includes a computer instruction, and the computer instruction is stored in a computer-readable storage medium. The processor of the electronic device reads the computer instruction from the computer-readable storage medium, and the processor executes the computer instruction, so that the electronic device executes the method provided in various optional implementations of the vehicle control parameter determination method or the vehicle control parameter determination system. The specific implementation process can refer to the description of the corresponding embodiment above, and will not be repeated.

[0150] It should also be noted that the device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed over multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the drawings of the device embodiments provided by the present application, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines.

[0151] Through the description of the above implementation mode, the technicians in the field can clearly understand that the present application can be implemented by means of software plus necessary general hardware, and of course, it can also be implemented by special hardware including special integrated circuits, special CPUs, special memories, special components, etc. In general, all functions completed by computer programs can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be various, such as analog circuits, digital circuits or special circuits. However, for the present application, software program implementation is a better implementation mode in more cases. Based on such an understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a readable storage medium, such as a computer floppy disk, a U disk, a mobile hard disk, a ROM, a RAM, a disk or an optical disk, etc., including a number of instructions to enable a computer device (which can be a personal computer, a training device, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0152] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments may be implemented in the form of a computer program product.

[0153] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from a website site, a computer, a training device, or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, training device, or data center. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a training device, a data center, etc. that includes one or more available media integrations. The available medium may be a magnetic medium, (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive (SSD)), etc.

Claims

1. A method for determining vehicle control parameters, characterized in that: include: Obtaining dynamic parameters of a target vehicle model for which control parameters are to be determined; Perform simulation based on the dynamic parameters of the target vehicle model to obtain a dynamic simulation model of the target vehicle model; Inputting the dynamics simulation model of the target vehicle model into a pre-trained prediction model to obtain an output result of the prediction model, wherein the output result can characterize the prediction control parameters of the target vehicle model; The prediction model is obtained by training a fully connected neural network in advance through dynamic simulation models of multiple sample vehicle models and control parameters of each sample vehicle model in the multiple sample vehicle models.

2. The method according to claim 1, characterized in that Also includes: training the prediction model; Wherein, the training of the prediction model comprises: Obtain dynamic simulation models of multiple sample vehicle models and control parameters of each sample vehicle model; Build an initial fully connected neural network model; Determine the number of model layers and nodes; The initial fully connected neural network model is trained based on the dynamics simulation models of the multiple sample vehicle models and the control parameters of each sample vehicle model to obtain a trained prediction model that meets the number of model layers and nodes.

3. The method according to claim 2, characterized in that The step of obtaining the control parameters of each sample vehicle model includes: Obtain multiple groups of control parameters for each sample vehicle model; Based on the fitness function, multiple groups of control parameters of each sample vehicle model are optimized to obtain optimized control parameters of each sample vehicle model, so as to use the optimized control parameters for model training.

4. The method according to claim 3, characterized in that The step of optimizing the multiple control parameters of each sample vehicle model based on the fitness function to obtain the optimized control parameters of each sample vehicle model includes: Determine the fitness function; Using the fitness function to evaluate the multiple groups of control parameters of each of the sample vehicle models, respectively, to obtain a first evaluation result corresponding to each group of control parameters in the multiple groups of control parameters; The multiple groups of control parameters are optimized based on the first evaluation results respectively corresponding to each group of control parameters in the multiple groups of control parameters to obtain a group of optimized control parameters for each sample vehicle model.

5. The method according to claim 4, characterized in that The optimizing the multiple groups of control parameters based on the first evaluation results respectively corresponding to each group of control parameters in the multiple groups of control parameters to obtain a group of optimized control parameters for each sample vehicle model includes: Based on the first evaluation results respectively corresponding to each group of control parameters in the multiple groups of control parameters, selecting a target number of control parameters whose first evaluation results satisfy a first target condition from the multiple groups of control parameters; Cross-combining the control parameters of the selected target groups to obtain multiple groups of new control parameters; Using the fitness function to evaluate the multiple groups of new control parameters respectively, to obtain second evaluation results corresponding to each group of new control parameters; Based on the second evaluation result, a control parameter satisfying a second target condition is determined from the multiple groups of new control parameters as an optimized control parameter.

6. The method according to claim 5, characterized in that The step of determining, based on the second evaluation result, a control parameter satisfying a second target condition from the multiple groups of new control parameters as an optimized control parameter comprises: Determine a control parameter that meets a second target condition from the multiple groups of new control parameters based on the second evaluation result, and determine the control parameter that meets the second target condition as a control parameter to be determined; Using a dynamics simulation model of a sample vehicle model corresponding to the control parameter to be determined, the control parameter to be determined is tested to determine whether the control parameter to be determined satisfies the driving condition of the sample vehicle model corresponding to the control parameter to be determined; If it is determined that the control parameter to be determined satisfies the driving condition of the sample vehicle model corresponding to the control parameter to be determined, the control parameter to be determined is determined as the optimized control parameter.

7. A vehicle control parameter determination system, characterized in that: include: An obtaining unit, used for obtaining dynamic parameters of a target vehicle model for which control parameters are to be determined; A simulation unit, used for performing simulation based on the dynamic parameters of the target vehicle model to obtain a dynamic simulation model of the target vehicle model; A prediction unit, used for inputting the dynamics simulation model of the target vehicle model into a pre-trained prediction model to obtain an output result of the prediction model, wherein the output result can characterize the prediction control parameters of the target vehicle model; The prediction model is obtained by training a fully connected neural network in advance through dynamic simulation models of multiple sample vehicle models and control parameters of each sample vehicle model in the multiple sample vehicle models.

8. The system according to claim 7, characterized in that Also includes: A training unit, used for training the prediction model; Wherein, the training unit is used for: Obtain dynamic simulation models of multiple sample vehicle models and control parameters of each sample vehicle model; construct an initial fully connected neural network model; determine the number of model layers and the number of nodes; perform model training on the initial fully connected neural network model based on the dynamic simulation models of the multiple sample vehicle models and the control parameters of each sample vehicle model, and obtain a trained prediction model that meets the number of model layers and the number of nodes.

9. An electronic device, characterized in that: include: A processor, used to obtain dynamic parameters of a target vehicle model for which control parameters are to be determined; Simulating based on the dynamic parameters of the target vehicle model to obtain a dynamic simulation model of the target vehicle model; inputting the dynamic simulation model of the target vehicle model into a pre-trained prediction model to obtain an output result of the prediction model, wherein the output result can characterize the prediction control parameters of the target vehicle model; wherein the prediction model is obtained by pre-training a fully connected neural network using dynamic simulation models of multiple sample vehicle models and control parameters of each sample vehicle model in the multiple sample vehicle models; The memory is used to store the program for the processor to execute the above processing process.

10. A readable storage medium, characterized in that: for storing at least one set of instruction sets; The instruction set is used to be called and at least execute the vehicle control parameter determination method according to any one of claims 1-6.

Images