关键词: 微操作平台, 单神经元PID控制, 运动跟踪, RBF神经网络, 压电驱动器

Abstract: Considering hysteresis nonlinearity and time variation of micro-manipulation stage, a single neuron PID control strategy was proposed for motion tracking control to improve motion accuracy and response of the stage. Gradient informations of the micro-manipulation stage might be obtained online by RBF neural network identifier, and learning algorithm of single neuron network was applied to achieve online self-tuning of PID parameters to achieve the adaptive motion tracking control of the micro-manipulation stage. In order to illustrate the feasibility of the proposed control method, experimental comparative analyses with ordinary PID control were carried out. Experimental results show that, displacement error ranges of the single neuron PID are as -0.5~0.5μm, and the adjustment time of that is as 0.1s, while displacement error ranges of ordinary PID control are as  -2.5~2.5μm, and the adjustment time of that is as 0.4s. It shows that the proposed control method has better control accuracy, response speed, and stronger adaptability.

Key words: micro-manipulation stage, single neuron PID control, motion tracking, radial basis function(RBF) neural network, piezoelectric actuator