论文部分内容阅读
针对基于EMD、WPT的特征提取方法各自存在的问题,采用将EMD与WPT结合的信号分解方法用于包络特征的提取,保证信号的分解不仅具有较高的分辨率,并且能够避免虚假IMF分量带来的错误信息。在包络特征的提取及结合之后,采用了基于PCA的特征选择方法对缺陷信号的特征数据集进行最有效特征的分类识别。根据以上特征提取及识别方法,使用Matlab以及Labview的混合编程进行了面向S700K-C型电动转辙机的故障诊断系统操作,并通过应用试验证明了该系统能够准确、快速地提取出故障信号。
Aiming at the existing problems of the EMD and WPT based feature extraction methods, the signal decomposition method combined EMD and WPT is used to extract the envelope features, which not only ensures the resolution of the signal but also avoids the false IMF components Bring the wrong message. After extracting and combining the envelope features, a PCA-based feature selection method is adopted to classify the feature data sets of the defect signals most effectively. According to the above feature extraction and recognition methods, the hybrid fault diagnosis system for S700K-C electric point machine is implemented by hybrid programming of Matlab and Labview, and the application test shows that the system can accurately and quickly extract the fault signal.