基于PSO-BP神经网络光电编码器误差补偿研究

摘要

High precision absolute photoelectric encoder is used widely,and the system precision is affected by the angle measuring precision of the photoelectric encoder.Due to the errors of the angle sensor generated in the production and the installation,the error occurs during the practical application of the photoelectric encoder,and the compensation effect of the traditional error compensation method is not good enough.Thus,an angle sensor error compensation algorithm employing BP neural network based on PSO is proposed by the paper.The experiment validation is performed,and the results show the good angle sensor error compensation effect is obtained by using the method.Compared with the uncompensated sensor,the standard deviation is elevated by 12.5 times,the maximum error and the average error is decreased to 9.6％and 8.5％,and the sensor measuring precision is promoted.The comparison test between the error compensation systems based on the traditional BP neural network,that based on the polynomial fitting method and that based on the method proposed by the paper is performed,the results show the compensation effect presented by the paper is better than other two methods.%考虑到高精度绝对式光电编码器应用广泛,其角度测量精度对整个系统精度影响较大,但由于角度传感器生产安装过程中产生的误差等原因,使得传感器在实际应用中存在一定的误差.而使用传统误差补偿方法难以得到较好的补偿效果,本文使用一种基于PSO的BP神经网络作为角度传感器误差补偿系统的算法.通过实验验证,该种算法能够对角度传感器误差进行较好的补偿,与补偿前相比,其标准偏差提高了12.5倍,最大误差和平均误差降低到9.6％和8.5％,提高了传感器检测精度.与使用了基于传统BP神经网络和基于多项式拟合算法的误差补偿系统进行对比实验,结果表明,其补偿效果亦优于这两种算法.