线控转向系统的自适应神经网络跟踪控制

Adaptive Neural Network Tracking Control for SbW System

针对车辆线控转向(SbW)控制系统受外部扰动和结构不确定等因素影响,结合神经网络系统的逼近特性,提出一种鲁棒自适应神经网络控制器来实现SbW系统的在线建模与转角控制。该控制器不需要SbW系统结构和参数完全已知的条件,而是通过神经网络逼近SbW系统的未知动力学,以及鲁棒方法消弱时变扰动和逼近误差对控制性能的影响。同时,该方法可由Lyapunov函数导出自适应律,从而保证了系统收敛性与稳定性。最后,通过稳定性分析表明了系统跟踪误差可渐近收敛至原点附近可调的邻域内;数值仿真和硬件在环仿真实验表明了该方法的有效性与优越性。

Considering the external disturbance and uncertainty of structure in the steer-by-wire(SbW)system,a robust adaptive neural network controller is proposed in this paper to realize online modeling and the front wheel control of SbW system by combining the approximation ability of neural network.The controller does not require the condition that the structure and parameters of the SbW system are completely known.Instead,a neural network is used to approximate the unknown dynamics of the SbW system and the control performance influence caused by time-varying disturbances and the approximation errors is eliminated by the robust method.Besides,the adaptive law can be derived from the Lyapunov function to ensure the convergence and stability of the system.Finally,the stability analysis shows that the system tracking error can asymptotically converge to the adjustable neighborhood near the origin.Numerical simulation and hardware-in-the-loop simulation experiments demonstrate the effectiveness and superiority of the method.