基于连续小波和支持向量机的铣削颤振预报.docVIP

基于连续小波和支持向量机的铳削颤振预报 （1哈尔滨理工大学机械动力工程学院 哈尔滨，150080 2黑龙江大学电子工程系 哈尔滨，150080） 摘要：木文研究了一种应用连续小波特征和多类球形支持向量机进行铳削颤振预报方法，该方法棊?丁?连续小波变换分别提取 高、低频段铳削振动信号的特征，利用多类球形支持向量机进行平稳铳削状态、饶削濒振孕育状态、饶削颤振状态识别。为 了简化二分类支持向量机进行多类分类时所带來的计算父杂性,该算法使每一类样本都获得一个球形支持向量机，在特征空 间中以测试样本与球形支持向量机中心距离作为决策函数，來进行识别。实验表明，在切削颤振识别系统中多类球支持向量 机与连续小波特征向量相结合貝有良好的识别效果，颤振孕育预报正确率达97. 3%o 关键词：硕振预报；连续小波；球形支持向量机；多类支持向量机 中图分类号：TG506.1 TP391.42 文献标识码：A The Milling Chatter Forecast Based on Continuous Waveletand Support Vector Machine Method Wu Shi, Lin Liandong, Xiao Fei, Qu Da (School of Mechanical and Power Engineering, Harbin University of Science and Technology, Harbin, 150080, China School of Electrical Engineering, Heilongjiang University, Harbin, 150080, China) Abstract: A cutting chatter forecast method based on continuous wavelet feature and multi-class spherical Support Vector Machines is studied in this paper, this method based on the continuous wavelet transtorm to extract the characteristics of the milling vibration signal in high and low frequency band. Spherical support vector machine was used to recog nize the smooth cut, chatter gestate and outbreak states .In order to simplify the computati onal complexity whe n using binary classification support vector machines to do multi-class classification, the algorithm to make every kind of samples have a spherical support vector machine, in the feature space the center distance between test sample and spherical support vector machine as a decision-making function to recognize. Experimental results show that in the cutting vibrati on chatter recog nition system spherical support vector machine combined the continu ous wavelet feature has a better recognition effect, the accuracy of chatter gestate forecast can reach at 97.3%? Keywords: chatter prediction; continuous wavelet; spherical support vector machine; multiclass support vector machines 1引言 颤振是发牛在铳削过程小一种强烈的白激振 动，颤振会严重制约铳削效率、降低零件的加工精 度和表面质量、损坏刀具羡至机床本身。如果在铳 削过程屮存在潜在颤振，一般会通过刀具、工件上 的振动表现出來。如果在颤振孕育过程屮就能发现 潜在颤振，就能通