RBF神经网络的主动前轮转向滑模控制

Sliding Mode Control of Active Front Steering on RBF Neural-Network

针对汽车主动前轮转向控制存在的非线性和参数时变不确定性,传统的滑模控制会产生"抖振"现象。为此,提出了一种基于RBF神经网络的主动前轮转向滑模控制策略。基于RBF神经网络对其切换增益进行实时调节,以抑制滑模变结构控制带来的"抖振"现象。利用CarSim和Matlab/Simulink联合仿真平台,对该控制策略进行仿真验证。研究结果表明:提出的RBF神经网络的滑模控制能够有效抑制"抖振"现象,相对于传统的滑模控制具有更强的鲁棒性和更高的控制效果,提高车辆转向时的操纵稳定性。

Aiming to the nonlinear and parameter time-varying characteristics of vehicle active front steering control,the conventional sliding mode control,a sliding mode variable structure using RBF neural network is proposed based on control of active front steering.In order to effectively reduce the chattering of using sliding mode variable structure,a real-time adjustment of the switch gain is employed based on RBF neural network.The simulation test of control strategy used by the co-simulation of Matlab/Simulink and CarSim under step response and double lane change operating conditions.From the simulation test,it can be concluded that the proposed sliding mode control of RBF neural network can significantly suppresses the chattering and more robustness than the conventional sliding mode control.It also improve vehicle stability during cornering.