基于互信息的小波特征提取方法及其在机械故障诊断中的应用  

Feature Extraction with Wavelet Based on Mutual Information and Its Applications to Fault Diagnosis of Machines

作　　者：焦卫东[1] 

机构地区：[1]嘉兴学院机电与建筑工程学院

出　　处：《中国机械工程》2004年第21期1946-1949,共4页China Mechanical Engineering

基　　金：国家自然科学基金资助项目 ( 5 0 2 0 5 0 2 5 ) ;浙江省自然科学基金资助项目 ( 5 0 0 10 0 4)

摘　　要：通过将小波方法与互信息———一个基于信号所有各阶统计量构建的统计测度相结合 ,形成了一种基于互信息分析的小波特征提取新方法。该方法克服了传统小波方法的不足 ,可由时域信号直接获取小波分解的低维量化特征 ,用于后续故障模式分类器的训练与设计 ,从而为构建自动化的智能机器故障诊断系统奠定了基础。实验结果表明 ,该方法简单有效 。By combining wavelet with mutual information——a measure constructed by all-order statistics, a new method for feature extraction with wavelet based on mutual information was proposed. This new method can overcome some drawbacks, and capture low-dimensional and quantitative features directly from wavelet decomposition of data in time domain, which were used for training classifier for fault patterns subsequently. Thus, a foundation for automatic and smart system for fault diagnosis was set up by use of the new method. The experimental results imply that the new method is not only simple and effective, but also of great potential in fault diagnosis of machines.

关 键 词：故障诊断 小波变换 互信息 智能机器故障诊断系统 

分 类 号：TN912.3[电子电信—通信与信息系统]