摘要
针对存在不确定性执行器故障的可重构机器人系统,提出一种基于自适应动态规划(ADP)的在线故障补偿控制。将利用策略迭代算法求解HJB方程获取的近似最优反馈控制器,基于可重构机器人动力学信息的名义控制器,以及无需故障诊断环节的在线补偿器,三者有效融合构建全局容错控制器,实现兼顾最优性的安全可靠性控制。仿真实验验证了该故障补偿策略的有效性,能够保证可重构机器人系统的稳定性与良好的跟踪性能。
In this paper,an online fault compensation control scheme is proposed for modular reconfigurable robots(MRRs)with uncertain actuator failures based on adaptive dynamic programming(ADP).In order to realize optimal safety and reliability control,the global fault tolerant controller is composed of three parts,i.e.,the approximate optimal feedback controller obtained by solving the HJB equation via the policy iterative algorithm,the nominal controller based on MRRs dynamics information,and the online compensator without fault diagnosis link.Simulation experiments verify the effectiveness of the fault compensation strategy,which ensures the system stability and good tracking performance of the MRR system.
作者
夏宏兵
黄迎辉
XIA Hong-bing;HUANG Ying-hui(School of Electronic and Electrical Engineering,Bengbu University,Bengbu 233030,China)
出处
《武汉轻工大学学报》
2020年第5期96-103,共8页
Journal of Wuhan Polytechnic University
基金
安徽省高校自然科学研究重点项目(KJ2019A0854)
蚌埠学院自然科学研究重点项目(2019ZR01zd).
关键词
自适应动态规划
补偿控制
容错控制
神经网络
可重构机器人
adaptive dynamic programming
compensation control
fault tolerant control
neural networks
modular reconfigurable robots