Filtering Method of Transformer Noise Acoustic Signal Based on WOA-VMD-FASTICA and Improved Mutual Fuzzy Entropy

Since the traditional blind source separation method can not effectively filter out the instantaneous interference and the noise of the transformer cooling fan, the disadvantage of causing uncertainly in the correspondence between the separated signal and the source signal. This essay proposes a new whale optimization algorithm to optimize variational modal parameters combined with fast independent component analysis and an improved fuzzy entropy transformer noise sound signal filtering method. First and foremost, decompose the fault acoustic signal of the fault transformer decomposed into multiple sub-model components through whale-optimized variational modal parameters, to satisfy the positive definite condition of blind source separation for mixed signals. And besides, use the fast independent component analysis method to construct the source signal on the modal components. Finally, combined with the improved fuzzy entropy, the noise signal in the source signal is set to zero through the threshold criterion, filters out the instantaneous interference and fan noise, and the transformer acoustic signal is reconstructed, and the reconstructed sound signal is better-preserved transformer body acoustic signal Combined with the analysis of the acoustic signal data collected in the field, the feasibility and effectiveness of the method are verified.