考虑误差和时间最优的柔性关节驱动电机抗饱和控制  

Anti-windup control for flexible joint drive motor considering error and time optimization

作　　者：宋传明 杜钦君[1] 冯晗 庞浩 李存贺 SONG Chuan-ming;DU Qin-jun;FENG Han;PANG Hao;LI Cun-he(School of Electrical and Electronic Engineering,Shandong University of Technology,Zibo 255049,China)

机构地区：[1]山东理工大学电气与电子工程学院,山东淄博255049

出　　处：《电机与控制学报》2023年第1期148-158,共11页Electric Machines and Control

基　　金：国家自然科学基金(62076152);山东省自然科学基金(ZR2020MF096);北京理工大学智能机器人与系统高精尖创新中心资助项目。

摘　　要：针对PID控制下柔性关节驱动电机的饱和问题,研究了一种考虑误差和时间最优的抗饱和控制方法。构建以系统跟踪误差和调节时间为优化目标,以驱动电机饱和上限为约束条件的优化模型,提出基于多目标模拟退火算法(MOSA)和莱文贝格-马夸特算法(LM)的MOSA-LM-PID组合优化方法;先利用MOSA的全局收敛能力得出PID参数最优解所在区域,再利用LM的局部快速收敛能力将近似最优解拟合为全局最优解,避免了MOSA的无效搜索,提高了收敛速度和精确度;针对MOSA算法一次随机权重可能造成优化偏重的问题,设计了一种考虑当前周期各目标优化情况的目标函数权重赋值方法。仿真与实验结果表明,MOSA-LM-PID的调节时间小于0.5 s,最大跟踪误差小于1.5%,且控制力矩的饱和度小于50%,均优于PID和MOSA-PID控制,有效改善了柔性关节的动态性能,提高了系统抗饱和能力。In order to solve the saturation problem of flexible joint drive motor under PID control, an anti-windup control method considering error and time optimization was studied. A combinatorial optimization model with joint tracking error and system adjustment time as optimization objectives and upper saturation limit of drive motor as constraints was constructed, and a MOSA-LM-PID combinatorial optimization method based on multi-objective simulated annealing algorithm(MOSA) and Levinberg Marquart algorithm(LM) was proposed. Firstly, the region of the optimal solution of PID parameters was obtained by using the global convergence ability of MOSA algorithm, and then the approximate optimal solution was fit to the global optimal solution by using the fast local convergence ability of LM algorithm, which avoids the invalid search of MOSA and improves the convergence speed and accuracy. For the traditional MOSA algorithm may cause optimization bias due to one-time random weights, an objective function weight assignment method considering the changing trend of each objective in the current cycle was proposed. The simulation and experimental results show that the adjustment time of MOSA-LM-PID is less than 0.5 s, the maximum tracking error is less than 1.5%, and the maximum control torque is about 50% smaller than the allowable maximum input torque, all of which are better than PID and MOSA-PID control. The design method effectively improves the dynamic performance of the flexible joint and improves the anti-saturation ability of the system.

关 键 词：柔性关节 驱动电机抗饱和 多目标模拟退火算法 莱文贝格—马夸特算法 组合优化 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]