摘要
研究了H型直线电动机工作台的误差测量、建模及补偿技术。首先分析了定位平台的误差来源,采用激光矢量测量方法测量工作台的定位误差;然后用最小二乘法分别建立工作台的分段线性,用BP算法建立神经网络误差模型,利用误差模型构造了各电动机的一维或二维误差校正表;最后,根据误差校正表进行误差实时补偿实验。实验结果表明,经过样本训练的神经网络模型对工作台的误差具有较强的预测能力,将工作台两个方向的定位精度都提高到1μm。
The error measuring, modeling and compensation techniques for H - type positioning stage driven by linear motors were studied. Firstly, the error sources of the linear positioning stage were analyzed. And the positioning errors were obtained by the laser vector measurement technique. Secondly, based on the measured results, the subsection linear error regression models were set up by least square method, and the neural network error model was built by BP neural network. The 1D and 2D Error correction tables of the linear motors were derived from the error models. Finally, error compensation experiments were conducted. The experimental results show that the BP neural network error model trained by good samples had strong predictive ability, the positioning accuracy in both directions of the stage were 1 μm through error compensation.
出处
《制造技术与机床》
CSCD
北大核心
2008年第5期22-25,共4页
Manufacturing Technology & Machine Tool
基金
国家自然科学基金资助项目(No.50390063)
关键词
直线电动机
定位平台
误差补偿
误差建模
神经网络
Linear Motor
Positioning Stage
Error Compensation
Error Modeling
Neural Network
