基于小波和统计检验的瞬态成分检测的方法及应用

Detection of Signal Transients Based on Wavelets and Statistics Method and Its Application

将连续小波变换技术和统计检验结合用于检测信号中具有一定时频分布的瞬态成分,指出连续小波变换具有分离信号中瞬态成分和噪声的作用。提出一种基于连续小波变换和统计检验的瞬态成分检测的“逐步去除法”。将“逐步去除法”用于齿轮箱振动中的瞬态成分的检测与提取,能比较有效地检测出瞬态成分;基于连续小波变换反演的瞬态成分估计能比较有效地表示设备的状态。

Continuous wavelet transform (CWT) and statistics are studied for the extraction of signal transients. It is pointed out that CWT separates the signal transients and the noise in the signal. The 'step-by-step detection' method for combining CWT with statistical testing is proposed for the detection and representation of transients, and then utilized in gearbox vibration signals for fault diagnosis. The detected transients in time-scale plane show that the condition of the gearbox is more clear than the original signal, and the reconstructed time domain transients from the transient coefficients are effective for the feature of gearbox fault diagnosis.