基于滤波器组的时频分布及其在故障诊断中的应用

Time Frequency Distribution Based on Filter Banks and Its Application to Diagnose Faults

采用滤波器组将信号分解成一系列子带信号 ,依据信息熵最小原则 ,对各个子带信号的伪 Winger分布 ( PWVD)的参数进行优化 ,得到各个子带信号的优化 PWVD,求这些优化 PWVD的加权和 ,得到原始信号的一种新的时频分布。当计算时频信息熵的分组数取得合适时 ,该分布具有一定的自适应性 ,且与其它自适应时频分布相比 ,具有计算量小的特点。将该时频分布应用于转子故障的诊断中 ,诊断效果十分理想。

A new time-frequency distribution is put forward in this paper. Firstly, the original signals are divided into a serial of subband signals based on the use of filter banks. Secondly, deciding the window length of the PWVD for each subband signal is based on the principle of Least Information Entropy (LIE). At last, the linear combination of every optimized PWVD is computed and the time frequency distribution of the input signals is obtained. When the numbers of the group for the value of PWVD are appropriate, the distribution is of some adaptive and the computation quantity is less than that by other adaptive time frequency. The results of application in the field of fault diagnosis for rotor are perfect.