摘要
针对BP优化PID神经网络(BP-PDNN)易陷入局部极小的不足,提出了一种变尺度混沌优化PID神经网络设计方法,即MSCOA-PIDNN,将其应用于机械臂轨迹跟踪控制中。利用混沌运动的遍历性优化网络权值,通过压缩优化变量取值区间提高搜索效率。采用MSCOA-PIDNN建立机械臂系统的预测模型,以多步预测性能指标为目标函数,优化PID神经网络控制器,从而实现机械臂系统轨迹跟踪的有效控制。仿真结果表明,MSCOAPIDNN在机械臂轨迹跟踪控制中性能优于BP-PIDNN。
Aiming at the defects of PID neural network optimized by BP algorithm, a kind of PID neural network training method based on mutative scale chaos optimization algorithm （MSCOA） is proposed and applied in manipulator trajectory tracking control. The neural network weights can be optimized by making use of the ergodicity of chaos,and the search efficiency can be increased through narrowing solution space. Through establishing the predictive model of manipulator by utilizing MSCOA-PIDNN, and by using the multi step predictive objective function to train the weights of PIDNN controller,manipulator trajectory tracking prediction control can be realized. The simulation results show that the performances of MSCOA-PIDNN are better than those of BP-PIDNN in manipulator traje ctory tracking control.
出处
《山东科技大学学报（自然科学版）》
CAS
2013年第5期84-89,95,共7页
Journal of Shandong University of Science and Technology(Natural Science)
基金
国家自然科学基金项目(50675186)
关键词
混沌优化
PID神经网络
机械臂
轨迹跟踪
预测控制
chaos optimization
PID neural network
manipulator
trajectory tracking
predictive control
