基于强化学习算法的串联液压机械臂运动优化控制研究  

Research on Motion Optimal Control of Series Hydraulic Manipulator Based on Reinforcement Learning Algorithm

作　　者：张娟荣[1] 王宇璐 ZHANG Rong-juan;WANG Yu-lu(School of Mechanical and Electrical,Xianyang Polytechnic Institute,Xianyang 712000)

机构地区：[1]咸阳职业技术学院机电(技师)学院,咸阳712000

出　　处：《环境技术》2023年第12期86-90,共5页Environmental Technology

基　　金：陕西省自然科学基础研究计划——基于Bayesian机器学习的振荡扑翼集群设计控制一体方法研究,项目编号:2023-JCQN-0063。

摘　　要：串联机械臂是一种融合机电液特征的连杆机构,针对复杂多变运行工况时运动稳定性能下降的缺陷,在串联机械臂运动系统结构和运动动力学模型基础上,对串联机械臂运动控制系统控制拓扑结构进行了重构,设计了基于强化学习算法的运动控制策略,通过在线标定优化相关参数,实现运动控制的稳定性,在构建的变工况运动控制试验台架中验证算法的可行性,实验结果表明,设计的强化学习算法能够有效的提升串联机械臂的运动稳定性。A series robotic arm is a linkage mechanism that integrates electromechanical and hydraulic characteristics.In response to the defect of reduced motion stability performance under complex and variable operating conditions,the control topology structure of the series robotic arm motion control system was reconstructed based on the structure and dynamic model of the series robotic arm motion system.A motion control strategy based on reinforcement learning algorithm was designed,and relevant parameters were optimized through online calibration,To achieve stability in motion control,the feasibility of the algorithm was verified in the constructed variable working condition motion control test bench.The experimental results showed that the designed reinforcement learning algorithm can effectively improve the motion stability of the s eries robotic arm.

关 键 词：串联机械臂 连杆机构 拓扑结构 强化学习 在线标定 变工况 

分 类 号：TH137[机械工程—机械制造及自动化]