基于PSO算法优化BP神经网络的数控机床热补偿  

Thermal Error Compensation for CNC Machine Tools Based on PSO-BP Neural Network

作　　者：任兵[1] 任小洪[1] 李国志[1] 

机构地区：[1]四川理工学院自动化与电子信息工程学院,四川自贡643000

出　　处：《机床与液压》2013年第3期59-61,25,共4页Machine Tool & Hydraulics

基　　金：人工智能四川省重点实验室重点项目(2010RZ002);四川理工学院2011年研究生创新基金项目(y2011013)

摘　　要：热变形误差是影响数控机床加工精度的主要因素之一。为了减小机床热误差,提高加工精度,以GMC4000H/2五坐标横梁移动龙门加工中心为研究对象,运用粒子群优化算法优化BP神经网络,建立机床的温度变化和热变形误差之间的补偿模型,并介绍了热变形误差的检测方法以及运用基于嵌入式的热补偿系统进行误差补偿的方法。优化后的BP神经网络模型的补偿效果优于普通BP神经网络,且具有很好的预测补偿能力和拟合性能。仿真实验结果表明,该方法具有可行性和有效性。Thermal error is one of the most primary factors that affect the machining precision of NC machine tool. In order to minish the thermal error and to improve the machining precision, taking GMC4000H/2 five coordinate moving beam machining center as study object, compensation model between temperature change and thermal error was established using PSO-BP neural networks. The detection approach of thermal error and error compensation based on embedded systems were introduced. PSO-BP neural networks model with a better compensation capability and fitting performance is more excellent in compensation purpose than ordinary BP neural network. The simulation results show that the method is feasible and valid.

关 键 词：PSO算法 数控机床 热误差补偿 BP神经网络 

分 类 号：TH161[机械工程—机械制造及自动化]