摘要
针对X-Y定位平台中摩擦等非线性部分对控制精度的影响问题,提出了基于自适应神经网络的鲁棒控制策略。设计神经网络控制器对摩擦及干扰等不确定部分进行补偿,其网络逼近误差作为外界扰动通过鲁棒控制器消除,保证X-Y平台的定位精度;设计神经网络参数学习算法,保证权值的在线自适应实时调整。基于H∞的HJI理论证明了控制系统的稳定性,并保证了系统L2增益小于给定的指标。试验结果表明所提控制方法能够很好补偿摩擦模型,提高了定位精度,具有重要工程应用价值。
Control problems are considered for X-Y position table system with nonlinear friction and interferences, robust control scheme based on neural network was proposed. The neural network controller is designed to compensate and approach uncertain parts such as friction and interference of the system. Network approximation errors as external disturbances are eliminated by robust controller to ensure the control accuracy; adaptive learning algorithm of neural network is designed to ensure that the online real-time adjustment of weights; The stability of control system based on H∞ theory is proved, and the system the L2 gain is less than the given target. Experimental results show that the proposed control method has high control precision, has important engineering value.
出处
《机械设计与制造》
北大核心
2015年第12期153-156,共4页
Machinery Design & Manufacture
基金
国家科技支撑计划课题(2013BAC16B02)
浙江省科技厅项目(2015C31160)
浙江省自然科学基金资助项目(Y14F030005)
丽水市公益技术应用项目(2014GYX047)
关键词
神经网络
X-Y定位平台
摩擦补偿
鲁棒控制
自适应控制
Neural Network
X-Y Position Table
Friction Compensition
Robust Control
Adaptive Control
