无人水面艇目标路径跟踪优化控制研究

Research of Target Path Tracking Optimization Control of Unmanned Surface Vehicle

无人水面艇(USV)目标路径跟踪优化控制研究是为了实现自主、准确、快速的路径跟踪。时变风干扰是无人艇目标路径跟踪过程中不可忽略的环境因素,在时变风干扰的影响下采用变船长比(LOS-PD)控制方法会产生较大的跟踪误差,因此无人水面艇目标路径跟踪优化控制的难点技术之一是有效抵抗时变风干扰的影响。针对上述问题,提出了一种基于神经网络的电压控制量学习算法。通过在时变风干扰下采用LOS-PD控制方法经过多次调参后得到大量先验数据,利用这些数据对神经网络进行训练,得到了一组从距离偏差、角速度、风速大小、目标路径方向的倾斜角、航向角到无人艇左右电压控制量之间的最佳权值矩阵。通过仿真,得到两种算法在不同风干扰下的平均跟踪误差、跟踪时间和平均跟踪能耗。仿真结果验证了基于神经网络的电压控制量学习算法对于抵抗时变风干扰的有效性。

The research on target path tracking optimization control of unmanned surface vehicle( USV) is to realize autonomous,accurate and rapid path tracking. The time-varying wind disturbance is an environmental factor that can not be neglected in the target path tracking of USV,and the tracking error under time-varying wind disturbance is large by using the variable ship length ratio( LOS-PD) control method,so one of the most difficult techniques of target path tracking optimization control of USV is to effectively resist the influence of time-varying wind disturbance. In order to solve this problem,a learning algorithm for voltage control based on Neural Network is proposed. Under time-varying wind disturbance,a large amount of prior data was obtained by using LOS-PD control method after multiple parameters adjustment. Then,a neural network was trained by using the prior data. The optimal weight matrices were obtained,which relate distance deviation,angular velocity,wind speed,the inclination angle of the target path direction,and heading angle to the output voltage on both sides of USV. By simulation,the average tracking error,tracking time and average tracking energy consumption of the two algorithms under different time-varying wind disturbances were obtained. The simulation result validates the effectiveness of the learning algorithm for voltage control based on neural network to resist time-varying wind disturbance.