摘要
为了实现刀具磨损状态监测的自动化与高精度,文章提出了一种基于希尔伯特-黄变换(hilbert-huang transform,HHT)和等距特征映射(isometric feature mapping,Isomap)的刀具磨损状态监测方法。首先采用经验模态分解算法对采集的信号进行降噪处理;然后对降噪后的信号进行Hilbert变换得到Hilbert时频谱,同时计算Hilbert边际谱及其统计特征量;最后利用Isomap算法进行特征融合及优化,将优化后特征向量送入支持向量机(support vector machine,SVM)中,并通过网格搜索法优化SVM的相关输入参数来建立最优分类模型。研究结果表明:Isomap算法具有较好的特征融合及降维效果,且Isomap-SVM分类模型对测试集的识别准确率为95%,文章所提方法可以有效地识别刀具磨损状态。
Time-frequency analysis can excavate the tool wear information contained in the tool wear sound wave signal in the time-frequency domain. This paper presents a tool wear monitoring method based on the Hilbert-Huang transform and isometric feature mapping. First,the empirical mode decomposition algorithm is used to denoise the collected signal. Then the Hilbert transform is performed on the noise reduction signal to obtain the Hilbert time spectrum. The Hilbert marginal spectrum and its statistical feature are also calculated. Finally,the feature fusion is carried out by using the isometric feature mapping algorithm. The optimized feature vector is sent into the support vector machine,and the optimal input model of the support vector machine is optimized by the grid search method. The results showthat the isometric feature mapping algorithm has better feature fusion and dimension reduction,and the accuracy of the Isomap-SVMclassification model is 95%. The method mentioned in this paper can effectively identify the tool wear state.
作者
宋伟杰
关山
庞弘阳
SONG Wei-jie;GUAN Shan;PANG Hong-yang(School of Mechanical Engineering,Northeast Electric Power University,Jilin Jilin 132012,China)
出处
《组合机床与自动化加工技术》
北大核心
2018年第6期114-118,共5页
Modular Machine Tool & Automatic Manufacturing Technique
基金
吉林省科技厅科技公关计划(20140204004SF)
吉林省省教育厅"十二五"科学技术研究项目
关键词
刀具磨损
希尔伯特-黄变换
边际谱
等距特征映射
支持向量机
tool wear
hilbert-huang transform
marginal spectrum
isometric feature mapping
supportvector machine
