基于神经网络逆系统的车辆动力学模型解耦法

Decoupling Method of Vehicle Dynamics Model Based on Neural Network Inverse System

为了消除车辆各系统纵横向之间的耦合影响,对车辆动力学模型进行了神经网络逆系统解耦控制。选用的研究对象为四轮驱动、前轮转向的无人驾驶车辆。首先,将包含侧向运动和横摆运动两个自由度的车辆动力学模型通过Interactor算法进行可逆性分析;其次,搭建卷积神经网络(Convolutional Neural Networks,CNN)和长短期记忆神经网络(Long Short-Term Memory,LSTM)逆系统结构构建逆系统,并验证该方法的可行性;将该解耦方法应用于无人驾驶车辆的轨迹跟踪控制设计中,通过CarSim和Matlab/Simulink联合仿真试验证明,设计的CNN+LSTM神经网络逆系统解耦控制在多种工况下都具较好的跟踪特性及稳定性。

In order to eliminate the coupling effect between vertical and horizontal directions of vehicle systems,neural network inverse system decoupling control is carried out for vehicle dynamics model.The research object is driverless vehicle with four-wheel drive and front wheel steering.Firstly,the vehicle dynamics model with two degrees of freedom including lateral motion and yaw motion is analyzed by interactor algorithm,Secondly,the inverse system structures of Convolutional Neural Networks(CNN)and Long Short Term Memory(LSTM)neural networks are built to replace the traditional inverse system decoupling strategy,and the feasibility of this method is verified,The decoupling method is applied to the trajectory tracking control design of driverless vehicle.The tracking effect is judged by observing the output response curve of the vehicle,and then the feasibility and stability of the method are proved,Finally,through the joint simulation test of CarSim and MATLAB/Simulink,it is proved that the CNN+LSTM neural network inverse system decoupling control designed in this paper has good tracking characteristics and stability under various working conditions.