扫描隧道显微镜微位移工作台的 神经网络PID控制方法研究

提出了一种基于神经网络理论的微位移工作台控制方案。该工作台以压电陶瓷作为微位移驱动元件,对伺服电机大位移进行位移补偿。分析了压电陶瓷微位移驱动器的原理,建立了工作台的数学模型。神经网络PID控制器对工作台进行闭环控制,利用BP网络的自学习和自适应能力,实时调整网络加权值,改变PID控制器的控制系数,减小工作台的位移误差。采用专用的压电陶瓷驱动电源对工作台的位移进行了实验,相对于常规PID控制器,微位移为11.41 μm时的响应时间从1.5 s缩短到1 s,稳态位移误差从3.13％减小到1.05％,工作台的稳定性和定位精度得以提高,改善了扫描隧道显微镜的工作性能。

Study on neural network PID control for micro-displacement stage of Scanning Tunneling Microscope

A control scheme for micro-displacement stage of Scanning Tunneling Microscope(STM) based on the neural network was proposed,in which piezoelectric ceramics is used as the micro-displacement actuator of stage to compensate the rough displacement of the servo mechanism. The principle of the actuator was analyzed and the mathematical model was set up. With the stage controlled by the neural-network PID controller in close loop,the weights of BP network and the parameters of PID controller could be adjusted to reduce the displacement error of stage by the function of self-learning and adaptability in real time. Experiments of stage displacement using the special electronic ceramics power were conducted. The results show that the response time for a micro displacement of 11.41 μm is shortened from 1.5 s to 1 s, and the stable error is reduced from 3.13% to 1.05%. The stability and positioning precision are improved and the performance of Scanning Tunneling Microscope is enhanced compared with the traditional PID controller.