基于GA优化BP神经网络的机床转台定位误差分析
Localization Error Analysis of Machine Tool Turntable Based on GA Optimized BP Neural Network
路忠响 1肖元昭2
作者信息
- 1. 恒达智控科技股份有限公司,河南 郑州 450064
- 2. 郑州轻工业大学工程训练中心,河南 郑州 450001
- 折叠
摘要
为了提高数控机床转台误差补偿效果,设计了一种通过遗传算法(GA)来完成BP神经网络优化过程,并加入坐标参数、运动速度指标建立了转台定位误差模型.通过Matlab软件构建GA优化BP模型,得到优化权值与阈值后,将结果移植至DSP内开展建模与预测,促进预测速率的大幅提升.研究结果表明:机床转台处于各空间位置与速率下形成了不同定位误差,总体呈现较为复杂的变化特征采用GA优化BP神经网络构建的模型预测时达到了更高精度,误差残差范围在-0.5～0.5 μm,满足实际需求.该研究有助于提高机床转台定位精度,增强加工精度.
Abstract
In order to improve the error compensation effect of CNC machine tool turntable,a BP neural network optimization process is designed by genetic algorithm(GA),and the positioning error model of turntable is established by adding coordinate parameters and moving speed index.The GA optimization BP model was constructed by Matlab software,and after the optimization weights and thresholds were obtained,the results were transplanted into DSP to carry out modeling and prediction,which greatly improved the prediction rate.The results show that the machine tool turntable has different positioning errors at different spatial positions and rates,and the overall change characteristics are more complex.The model built by GA optimization BP neural network can achieve higher accuracy,and the error residual range is-0.5～0.5 μ m,which meets the actual demand.The research is helpful to improve the positioning accuracy of the machine tool turntable and enhance the machining accuracy.
关键词
在机测量系统/定位误差/实时预测/数字信号处理器(DSP)/遗传算法-反向传播(GA优化BP)网络Key words
in-machine measurement system/positioning error/real-time prediction/digital signal processor(DSP)/genetic algorithm-back propagation network引用本文复制引用
出版年
2024
NETL
NSTL
万方数据