双通道多级旋转变压器的标定和基于BP神经网络的误差补偿方法  

Calibration for Twin-channel Multipole Resolvers and the Method of Error Compensation on BP Neural Network

作　　者：张飞[1] 史士财[1] 孙敬颋[1] 郭闯强[1] 陈泓[1] 刘宏[1] 

机构地区：[1]哈尔滨工业大学机器人技术与系统国家重点实验室,黑龙江哈尔滨150001

出　　处：《机械与电子》2011年第9期13-15,共3页Machinery & Electronics

摘　　要：通过高精度的编码器实现对旋转变压器的标定,为了提高旋转变压器的精度,提出了一种基于BP神经网络的误差补偿方法,用该网络对误差曲线进行训练,再将训练结果写入DSP程序中进行实时补偿。实验结果表明,通过该方法对误差进行补偿,可以将误差从±9′补偿到±0.3′,重复性好。The paper realizes calibration for resolvers by high - accuracy encoder. To enhance measure precision,a method of error compensation on BP neural network is presented and trains the error curve by this network, then writes the training result into DSP to realize the real - time compensation. The experiment shows that the resolver＇s error can be reduced from 9′ to 0. 3′ by the compensation and the repeatability is high.

关 键 词：旋转变压器 零位误差 幅值误差 BP神经网络 

分 类 号：TM383.2[电气工程—电机]