为了提取有效的故障特征,提出了基于频率切片小波变换时频分解的故障特征分离提取方法。先对信号进行频率切片小波变换获取其时频分布,然后根据信号的能量分布特点选择时频区域,再以较高的时频分辨率对选择的时频区域进一步细化分析,以突出隐含在信号中的时频特征,在此基础上分割出含有故障特征时频区域,再通过滤波和逆变换重构分离出有效的故障特征。仿真实验和工程应用表明,这种方法可从噪声信号中分离出有效的特征分量,在发电机组故障特征提取时取得了较好的效果。
Vibration signals from a monitored machine are usually nonstationary, so to extract efficient features we should process the signals both in time domain and frequency domain. The frequency slice wavelet transform (FSWT) is a new method for time-frequency analysis, which has the following properties: the center of time-frequency window is the observing center,