摘要
单一工况条件下数控机床主轴热误差模型无法准确预测其它工况下的热误差。通过研究分析支持向量机回归的算法和参数的关系,提出一种经过遗传算法(GA)在多工况条件下优化的支持向量机(SVM)的建模方法。以一台数控车床为研究对象,进行热误差测量实验,利用电涡流位移传感器和温度传感器同步测量机床主轴两个方向热误差和温度变化数值,获取两种工况的建模数据。运用遗传算法对SVM的惩罚函数、核函数参数和不敏感损失函数进行多工况条件下的优化选择,建立机床主轴热误差补偿模型。通过热误差建模实验验证,该方法在工况一的残差为0.838μm,工况二的残差为0.653μm,在保持较高预测精度的同时,能在两种工况下进行有效的热误差预测,使热误差补偿更适合实际加工环境。
In a single working condition, the error model of NC machine tool can not accurately predict the thermal error of other work conditions. Analysis of the relationship between support vector machine regres- sion algorithm and the parameters, put forward a modeling method with support vector machine （SVM） op- timized by genetic algorithm （GA） in multiple working conditions. To do the thermal error measurement experiment on the CNC lathe, using the eddy current displacement sensor and temperature sensor to measure the thermal error and temperature on machine tool spindle direction, collecting the data of 2 different work conditions. Selection and optimization using genetic algorithm for SVM kernel function parameter, penalty function and insensitive loss function under multiple working conditions, to establish the model of machine tool spindle thermal error. Through the thermal error mode experimental validation, The method of the re- sidual in work condition one is 0. 838μm, work condition two is 0. 653μm, this method can keep high predic- tion accuracy, while forecast the effective thermal errors in the 2 working conditions, The thermal error com- pensation is more suitable for the actual processing environment.
出处
《组合机床与自动化加工技术》
北大核心
2017年第7期27-31,共5页
Modular Machine Tool & Automatic Manufacturing Technique
基金
国家自然科学基金资助项目(51365009)
广西硕士研究生科研创新资助项目(YCSZ2014134)
2015年度桂林电子科技大学北海校区青年教师基础能力提升资助项目(UB16005Y)
关键词
数控机床
热误差
多工况
支持向量机
CNC machine tool
thermal error
multiple working conditions
support vector machine
