摘要
为使得数控机床热误差实时补偿更有效,提出一种基于粒子群算法(Particle Swarm Optimization,PSO)的温度测点优选方法和基于极限学习机(extreme learning machine,ELM)神经网络的机床热误差补偿模型。利用PSO优化K均值聚类方法,实现了机床上温度测点的优化筛选。利用ELM人工神经网络建立机床热误差补偿模型,通过合理选取隐层神经元数,从而实现更精确、更有效地对数控机床热误差进行实时补偿控制。通过与传统BP(Back Propagation)、RBF(Radial Basis Function)神经网络进行对比分析,该补偿模型具有计算简便、预测精度高、结构简单等优点,可有效应用于数控机床热误差实时补偿模型。
In order to improve the precision of real time compensation for thermal error on NC machine tool,this paper proposed a method which is based on the PSO temperature measuring point clustering and the modeling of ELM neural network for thermal error compensation in NC machine tools. K-means clustering is optimized by PSO algorithm so as to decrease the number of the temperature sensors. And then the ELM neural network is established the thermal error model based on the main temperature points so that a NC machine tool is online compensated more effectively. Compared with BP,RBF methods,ELM neural network has an advantage of calculation speed,structure and precision which could be used to the real compensation for NC machine tools.
出处
《组合机床与自动化加工技术》
北大核心
2015年第7期69-73,共5页
Modular Machine Tool & Automatic Manufacturing Technique
基金
国家自然科学基金项目(51275305)
2013年度上海市引进技术的吸收与创新技术项目(13XI-07)
辽宁省科技创新重大专项(201301001)
国家科技重大专项课题(2011ZX04015-31)
