基于GA-ACO-BP网络的机床主轴热误差预测  

Thermal error prediction of machine tool spindle based on GA-ACO-BP network

作　　者：田春苗 季泽平 阿勇嘎 张学炜[1] 唐术锋[1] 郭世杰[1] TIAN Chunmiao;JI Zeping;A Yongga;ZHANG Xuewei;TANG Shufeng;GUO Shijie(School of Mechanical Engineering,Inner Mongolia University of Technology,Huhhot 010051,CHN)

机构地区：[1]内蒙古工业大学机械工程学院,内蒙古呼和浩特010051

出　　处：《制造技术与机床》2022年第9期153-161,共9页Manufacturing Technology & Machine Tool

基　　金：国家自然基金(52065053);中央引导地方科技发展计划(2020ZY0002);内蒙古自然科学基金(2019BS050008);内蒙古关键技术攻关计划(2021GG0255);内蒙古自治区高等学校创新团队发展计划支持(NMGIRT2213);内蒙古自然基金(2020BS05003)。

摘　　要：为解决反向传播(BP)神经网络建立的主轴热误差预测模型精度低、收敛速度慢和易陷入局部最优解的缺点,利用K-means++算法和相关性分析对温度测点进行优化并提取热敏感点,并利用遗传算法(GA)对蚁群进行交叉变异处理,构建GA-ACO网络来确定最优的隐含层节点数、权值、阈值,实现对BP神经网络拓扑结构的优化。分别建立基于BP和GA-ACO-BP网络的主轴热误差预测模型,以双转台五轴加工中心为研究对象,采用五点法对主轴热误差进行测量。热误差实验结果表明:K-means++算法与Person、Sperman和Kendall相关分析相结合可有效降低温度变量间的多重共线性;GA-ACO-BP模型可实现对主轴热误差的预测具有更高的预测精度。In order to solve the shortcomings of the spindle thermal error prediction model established by back propagation(BP)neural network,such as low accuracy,slow convergence speed and easy to fall into the local optimal solution,the K-means++algorithm and correlation analysis were used to optimize the temperature measurement points and extract the thermal sensitive points,and the genetic algorithm(GA)was used to cross-variation the ant colony.GA-ACO network was constructed to determine the optimal number of hidden layer nodes,weights and thresholds,so as to realize the optimization of BP neural network topology.The thermal error prediction models of spindle based on BP and GA-ACO-BP networks were established,and the thermal error of spindle was measured by five-point method with the five-axis machining center of double turntable as the research object.The thermal error experiment results show that the K-means++algorithm combined with Person,Sperman and Kendall correlation analysis can effectively reduce the multicollinearity between temperature variables.Ga-aco-bp model can predict the thermal error of spindle with higher accuracy.

关 键 词：数控机床 主轴热误差 K-means++算法 热误差预测 GA-ACO-BP神经网络 

分 类 号：TH161[机械工程—机械制造及自动化]