Damage detection based on multi-wavelet basis and multi-scale feature fusion

Multi-scale decomposition of wavelet packet transform can obtain more detailed time-frequency local characteristics of signals, which has unique advantages in the processing and analysis of both stationary and non-stationary. Therefore, wavelet transform is more and more applied in the field of damage identification, and the selection of medium and small wave base and scale of wavelet packet transform will greatly affect the final results. A damage identification method based on the fusion of multi-wavelet basis and multi-scale features is proposed. By selecting different wavelet basis for the transformation of different scales, the relative energy distributions of noisy signals in different frequency bands are obtained. These relative energy distributions are fused as the feature vectors of the classification algorithm to identify the damage. The results show that this method can effectively improve the accuracy of the classification algorithm.