Study on Adaptive Wavelet De-Noising for Measurement Signals and Its Application

In order to reduce the negative influence of noise on the extracting fault feature (correlation dimensions) in fault diagnosis, an adaptive wavelet de-noising method is presented in this paper. Based on the constructive theory of orthogonal binary wavelet basis, a parameter expression equation of orthogonal wavelet basis is constructed and a adaptive goal function of de-noised effect is defined. By applying genetic optimization method, the best wavelet basis was obtained, and the correlative arithmetic is presented. Applying the optimal wavelet basis to eliminate noises from signals, and computed the correlation dimension of the de-noised signals as fault eigenvalue. Simulation and experiments show that the adaptive wavelet de-noising makes the mechanical fault feature extraction more reliable