基于时间序列预报的丝杠磨床误差补偿算法与仿真分析  

An Error Compensation Algorithm and Simulation Analysis of Screw Grinding Machines Based on Time Series Predictions

作　　者：洪江力[1] 殷爱华[1] 冯虎田[1] 金鑫[1] 

机构地区：[1]南京理工大学机械工程学院,江苏南京210094

出　　处：《机床与液压》2011年第5期10-14,共5页Machine Tool & Hydraulics

基　　金：江苏省创新基金(学者攀登计划项目;BK2008050);国家科技重大专项(2009ZX04001-171-02)

摘　　要：分析现有的几种用于丝杠磨床的误差补偿方法,提出一种基于时间序列的预报补偿方法,给出时间序列建模公式和模型参数的辨识方法。通过大量的仿真实验分析时间序列模型的建立过程中起始采样数据点和模型参数辨识过程中加权系数对预报结果的影响。结果表明,通过使用合理的加权系数和模型参数可使预报偏差控制在误差大小的1/5以内。Some existing error compensation methods used in screw grinding machines were analyzed, and a method based on time series predictions was proposed. The formula of time series modeling and model parameter identification method were given. The influences of two important aspects, namely the origination of sampling data points in process of time series modeling and weighting co- efficients in process of model parameter identification, on predication results were analyzed through a large number of simulation experi- ments. The results show that by using a reasonable weighting coefficients and model parameters can control the error of prediction bias the size of 1/5.

关 键 词：丝杠磨削 在线补偿 时间序列法 参数辨识 

分 类 号：TH161[机械工程—机械制造及自动化]