基于模糊RBF神经网络的超精密定位系统

Ultra Precision Positioning Control Based on Fuzzy Neural Network

设计了一种基于激光莫尔信号的超精密平面定位系统。针对精密定位系统存在非线性难以建立精确控制模型的缺陷,将不基于模型的模糊控制理论应用于精密定位控制系统中,并应用径向基函数(RBF)神经网络实现模糊控制器的输入输出映射关系,构造了神经网络模糊控制器模型,通过对光强及光强变化率的映射,得到电机驱动信号。改造后的模糊控制系统具有了知识自动获取功能,能更好地适应工况环境。实验结果表明,使用模糊神经网络控制,控制响应快、稳定性好及鲁棒性强,可有效提高定位精度及定位速度,系统可获得±0.5μm的定位精度。

An ultra-precision positioning .set based on laser Moiré signals is designed. It can achieve high-accuracy position detection and full automatic precision plane alignment. In view of the existing non-linearity and the difficulty in establishing precision controlled model,fuzzy control method is used in the precision positioning control. The RBF neural network is used to realize the relationship of input and output in fuzzy control. Fuzzy neural network controller（FNNC） model is set up. Driving signal is obtained by mapping Moiré signals intensity and its intensity rate. The improved fuzzy control system can obtain corresponding operation knowledge according to working environment. The experimental result shows FNNC can perform excellent control effect, such as,good stability, and strong robustness. The positioning speed and positioning accuracy can be effectively improved. The device based on fuzzy neural network can achieve the positioning accuracy of ±0.5μm.