基于自适应模型预测控制的拖拉机路径跟踪研究

Research on path tracking of tractors based on adaptive model predictive control

自主化作业的拖拉机由于作业速度和跟踪路径曲率的不断变化,基于固定参数模型预测控制的路径跟踪器不能达到理想效果。为提高控制器的自适应性,提出基于改进粒子群优化的自适应模型预测控制算法。该算法将作业场景与粒子群算法相结合,对模型预测控制中的预测时域进行自适应调整,当作业场景发生改变时,则用粒子群优化算法选取理想预测时域参数。为提高粒子群优化算法的寻优效果,采用分段函数的方式对惯性权重进行改进。以东方红-X1304拖拉机为研究对象,对作业速度为1、2 m/s和变速,跟踪路径为直线和曲线等情况进行仿真实验,并对比分析基于固定预测时域和自适应预测时域的控制器。结果表明,相对于基于三个固定时域的控制器,基于自适应预测时域控制器的跟踪精度和收敛速度分别提高了2%~44%和2%~71%。

Due to the varying working speed and tracking path curvature for autonomous tractors,the path tracking controller based on model predictive control(MPC)with fixed parameters cannot achieve the best effect.To improve the adaptability of path tracking controller,an adaptive MPC using improved particle swarm optimization(IPSO)is proposed.The algorithm combines the working scene and particle swarm optimization(PSO)algorithm to adaptively adjust the parameter of prediction horizon in MPC.The IPSO algorithm is used to optimize the prediction horizon when working scene changes.To improve the search effect of PSO,inertia weight is designed as a piecewise function.Simulation experiments are carried out under various conditions with different speeds(i.e.,1 m/s,2 m/s and varying)and paths(i.e.,straight and curve)for the Dongfanghong-X1304 tractor.And the controllers based on fixed prediction time domain and adaptive prediction time domain are compared and analyzed.The results show that compared with the controllers based on three fixed time domains,the tracking accuracy and convergence speed of the adaptive predictive time domain controller are improved by 2%~44%and 2%~71%,respectively.