Anomaly Detection of Vibration Signals based on Wavelet Modulus Maximal Multifractal Analysis

A method for detecting abnormality of telemetry vibration signal based on multi-fractal analysis of wavelet modulus maxima was proposed to solve the complexity, non-stationary nonlinearly of frequency domain analysis. First, zero drift correction and trend item elimination are performed on the vibration signal. Then the wavelet modulus maximum multifractal analysis method is used to decompose and multifractal analysis of vibration signals. Finally, the characteristic parameters are input to the SVM classifier, and the anomaly detection of the telemetry vibration signal is performed according to the classification result. The measured data verifies the effectiveness of the method.