用神经网络方法标定激光平场扫描系统

Neural Network Approach to the Calibration of Laser Scanning System

在激光快速成型机中，激光扫描精度直接影响到成型精度．在基于检流计扫描器的后物镜平场扫描系统中，输出平面的笛卡尔坐标系与扫描器的球面坐标系之间进而与控制两路扫描镜的电压数字量之间存在着非线性映射关系．在理想情况下，这种映射关系是已知的，但在实际上，由于很多因素的影响，若按理想关系映射，会产生较大的非线性扫描误差．采用神经网络方法对激光扫描系统进行全场标定补偿，显著提高了扫描系统的扫描精度．

In rapid prototyping machines, the position accuracy of laser spot has a direct influence on the accuracy of parts. However, in post objective flat scanning system based in galvanometer scanner, complex nonlinear errors can be introduced by many uncertain factors. In this paper, a neural network method is proposed to calibrate and correct the scanning system. The results show a remarkable improvement of the position accuracy.