基于概率神经网络的蜗杆砂轮磨齿机径向热误差建模  

Radial thermal error modeling of CNC worm wheel gear grinding machine based on probabilistic neural network

作　　者：钟金童[1] 李国龙[1] 刘达斌[1] 刘鹏祥[1] 廖琳[1] 

机构地区：[1]重庆大学机械传动国家重点实验室,重庆400044

出　　处：《计算机集成制造系统》2017年第3期534-541,共8页Computer Integrated Manufacturing Systems

基　　金：国家科技支撑计划资助项目(2014BAF08B02);国家自然科学基金资助项目(51375512)~~

摘　　要：径向热误差直接影响蜗杆砂轮所磨削的齿轮M值,是热误差中的关键因素。为此提出了一种基于概率神经网络的热误差补偿方法。利用模糊C均值聚类选取温度变量,采用概率神经网络建立了齿轮M值与温度关系的补偿模型,在数控蜗杆砂轮磨齿机上进行了补偿实验验证。结果显示,补偿后齿轮M值误差从0.06mm降到0.01mm以内,表明该模型用于热误差补偿具有较高的精度。通过与其他模型进行比较,验证了该模型的稳定性和可行性。The radial thermal error was the key factors of thermal error which directly affected M value of gear in CNC worm wheel gear grinding machine.Therefore,a method for the radial thermal error compensation of CNC worm wheel gear grinding machine based on Probabilistic Neural Network（PNN）was put forward.The temperature variables was optimized and selected by adopting the methods of fuzzy clustering,and the thermal error compensation model were set up by PNN.The compensation model was tested with CNC worm wheel gear grinding machine.The result showed that the error of the gears M Value was reduced from 0.06 mm to 0.01 mm or less.The above research indicated that the PNN model had high accuracy for thermal error modeling of CNC worm wheel gear grinding machine.Through comparing with other model,the stability and the feasibility of this model were verified.

关 键 词：蜗杆砂轮磨齿机 概率神经网络 热误差 建模 

分 类 号：TH161[机械工程—机械制造及自动化]