摘要
为了从发动机缸盖振动信号中快速提取反应故障状态的特征参数,给出了提升小波包变换算法及其移频算法,依据信号时频特性,定义了标准化相对能量,并给出了特征向量提取方案。方案同时选取多个完整工作循环数据进行提升小波包分解,将各频带内系数进行阈值量化,利用各频带系数直接求取标准化特征向量,同时做到了消除波动影响、降噪、使故障特征更加敏感。通过实例验证了方法的有效性与优越性。
In order to obtain the feature parameters reflecting engine fault state from cylinder head vibration signal,the lifting wavelet package transformation and the frequency-shift arithmetic are introduced.The standard relative energy is defined based on the signal's time-frequency characteristic,and a feature extraction scheme is proposed.The data of several work circles is selected and decomposed through the lifting wavelet package transformation.Coefficients of each frequency band are processed by the threshold.A standard feature vector is obtained from coefficients directly.The fluctuation is eliminated,the signal-to-noise ratio is improved,and the fault features is more sensitive.The validity and superiority of the scheme are verified by an example.
出处
《振动与冲击》
EI
CSCD
北大核心
2008年第2期34-37,共4页
Journal of Vibration and Shock
基金
河北省自然科学基金资助项目(编号:E2007001048)
关键词
缸盖振动信号
短时傅里叶变换
提升小波包变换
特征提取
故障诊断
cylinder head vibration signal,short time Fourier transformation,lifting wavelet package transform,feature extraction,fault diagnosis
作者简介
曹建军 男，博士生，1975年生
