点击次数：

论文类型：期刊论文

发表时间：2012-11-01

发表刊物：INTERNATIONAL JOURNAL OF ADVANCED MANUFACTURING TECHNOLOGY

收录刊物：SCIE、EI、Scopus

卷号：63

期号：1-4

页面范围：129-136

ISSN号：0268-3768

关键字：Bearing roller; Electrochemical mechanical polishing; Least squares support vector machines; Prediction

摘要：The surface quality of bearing rollers has great influence on the service performance of bearings. Electrochemical mechanical polishing (ECMP) is used to polish bearing rollers because it does not suffer from the disadvantages inherent in traditional bearing roller machining. However, predicting surface quality and determining processing parameters are difficult to accomplish because ECMP results are influenced by many factors. To overcome these problems, we develop an ECMP prediction model on the basis of least squares support vector machines with radial basis function. An orthogonal experiment is conducted to assess the effect of polishing parameters on surface roughness. Experiment results and predicted values show that ECMP is suitable for the machining of bearing rollers, with noticeable improvement in surface qualification. The mean absolute percent error (e (MAPE)) between the predicted and experimental values of surface roughness is 5.4%, and the root mean square error (e (RMSE)) is 6.5%. In addition, the e (MAPE) between the predicted and experimental values of current density is 4.8%, with an e (RMSE) of 6.6%.