关键词: 滚珠丝杠, 支持向量机, 串行优化算法, 振动信号

Abstract:

Aiming at the difficulty of on - line monitoring of the grinding process, the support vector machine was used to build the intelligent model. Based on the surface waviness and roughness theory of ball screw, extracted the vibration signal in the grinding process combining with machining parameter are taken as the model input. The support vector machine and support vector regression machine based on serial optimization algorithm are selected for the different problems of flutter classification and roughness prediction. In addition the model parameters are optimized by cross validation. Judge the flutter at first, predicted surface quality in judging the case of non-flutter. The experimental results show that the model can better monitor the flutter classification and roughness prediction.

Key words: ball screw;support vector machine;serial optimization algorithm;vibration signal