摘要
基于Lyapunov稳定性理论和可重构机械臂的模块化属性,针对非故障系统设计了分散反演神经网络控制器,并采用自适应神经网络系统补偿子系统关联项对系统控制精度的影响。通过引入一阶滤波器将传感器故障转化成伪执行器故障,从而得到增广故障子系统模型,进而采用多步时延技术补偿并发故障,实现容错。两种不同构形的2-DOF可重构机械臂系统的仿真结果表明了所设计分散容错控制方法的有效性。
According to Lyapunov stable theory and the modularity of reconfigurable manipulator,a decentralized backstepping neural network controller is designed for the fault free system,and the adaptive neural network system is employed to compensate the reduced control accuracy from the interconnection term.An augmented subsystem is obtained by introducing a filter,which maps the sensor fault into pseudo actuator fault,and then the multi-step time-delay technique is adopted to compensate the concurrent failures and realize the fault-tolerant.Simulations illustrated by two 2-DOF reconfigurable manipulators with different configurations demonstrate the effectiveness of the proposed scheme.
出处
《吉林大学学报(工学版)》
EI
CAS
CSCD
北大核心
2015年第6期1874-1880,共7页
Journal of Jilin University:Engineering and Technology Edition
基金
国家自然科学基金项目(61374051
60974010)
吉林省科技发展计划项目(20150520112JH)
关键词
自动控制技术
可重构机械臂
分散容错控制
多步时延
并发故障
automatic control technique
reconfigurable manipulator
decentralized fault-tolerant control
multi-step time-delay
concurrent failure