通过典型信号的MATLAB仿真讨论了小波在检测信号突变点时的选取原则。针对滚动轴承故障振动信号,先进行小波消噪,再进行小波分解与重构,对重构后的细节信号作Hilbert包络并进行谱分析,从功率谱中可清晰地识别出滚动轴承故障特征频率。 Through the MATLAB simulation of typical signals, the principle of selecting wavelet when detecting signal abrupt point is discussed. Aiming at the fault vibration signal of rolling bearing, the wavelet denoising is firstly carried out, and then the wavelet decomposition and reconstruction are performed. The reconstructed detail signal is Hilbert-enveloped and analyzed. The frequency characteristic of rolling bearing fault can be clearly identified from the power spectrum.