摘要
Due to the non-linearity behavior of the precision positioning system, an accurate mathematical control model is difficult to set up, a novel control method for ultra-precision alignment is presented. This method relies on neural network and alignment marks that are in the form of 100μm pitch gratings. The 0-th order Moire signals＇ intensity and its intensity rate are chosen as input variables of the neural network. The characteristics of the neural network make it possible to perform self-training and self-adjusting so as to achieve automatic precision alignment. A neural network model for precision positioning is set up. The model is composed of three neural layers, i.e. input layer, hidden layer and output layer. Driving signal is obtained by mapping Moire signals＇ intensity and its intensity rate. The experimental results show that neural network control for precision positioning can effectively improve positioning speed with high accuracy. It has the advantages of fast, stable response and good robustness. The device based on neural network can achieve the positioning accuracy of ± 0. 5μm.
针对精密定位装置存在非线性,精确数学模型难于建立的缺陷,提出了精密定位的神经网络控制方法.将BP神经网络应用于该控制系统中,系统以光栅常数100μm的光栅为定位标记,以激光衍射产生的莫尔光光强及光强的变化率为神经网络的输入变量,利用神经网络的自学习功能进行精密定位控制.建立了精密定位的神经网络控制模型,模型由输入层、隐层和输出层3层神经元组成,通过对光强及光强变化率的映射,得到电机驱动信号.实验结果表明,使用神经网络控制,控制响应快,稳定性好,鲁棒性强,可有效改善控制质量,提高定位速度,系统可获得±0·5μm的定位精度.
基金
The Natural Science Foundation of Higher EducationInstitutions of Jiangsu Province (No.04KJB510073).
