基于干扰观测器的直播机路径跟踪快速终端滑模控制

针对农田中存在不确定性干扰时，会导致建立的无人水稻直播机运动学模型精确度不高，从而使路径跟踪收敛时间长，跟踪效果较差等问题，提出由直播机运动学模型建立相应的非线性干扰观测器从而实现对不确定性复合干扰的精确观测。将观测到的干扰值补偿到运动学模型中以提高模型的精度，降低滑模控制的抖振。为了控制直播机沿指定路径作业并提高路径跟踪收敛时间，设计了一种快速终端滑模控制算法。基于李雅普诺夫判据检验了算法闭环系统的稳定性。使用Matlab/Simulink建立了农机运动学仿真模型，仿真结果表明，非线性干扰观测器能精确观测出系统的不确定性干扰。与不带干扰观测器的滑模控制算法相比，本文控制算法可有效减少收敛时间，抑制干扰带来的抖振。无人直播机水田作业实验表明，采用本文所提出的算法以1.2m/s的速度高效作业时，横向平均绝对偏差为0.0247m，均方差为0.0311m。并且转弯收敛速度快，无超调，路径跟踪精度满足实际作业要求。

Fast Terminal Sliding Mode Control for Autonomous Rice Seeding Machine Based on Disturbance Observer

Due to the existence of uncertainty disturbance, the accuracy of the established kinematics model of the autonomous rice seeding machine was not of highly accuracy, which would lead to problems such as slow convergence time and poor path tracking effect. Firstly, in order to improve the accuracy of the kinematics model and reduce the convergence time, the fast terminal sliding mode control algorithm was proposed based on the unmanned rice seeding machine. To realize the accurate observation of the unknown disturbance, the corresponding nonlinear disturbance observer was established according to the kinematics model, and the disturbance observed would be compensated to the controller. Next, the stability of the control system was proved according to Lyapunov criterion. Then, the kinematics simulation model of the agricultural machinery was built by Matlab/Simulink. The simulation results showed that the uncertain disturbance can be accurately observed by the disturbance observer. The control algorithm had better performance than sliding mode control algorithm without disturbance observer, which can effectively reduce the convergence time and almost had no chattering phenomenon. Finally, an experiment in a paddy field was carried out. When machine worked efficiently at the speed of 1.2m/s, the absolute average lateral deviation was 0.0247m, and the mean square error was 0.0311m. The turning speed was also fast and there was no overshot phenomenon. The results showed that the path tracking effect of the algorithm proposed was of high performance and can satisfy the actual operation requirements well.