Investigating Electromagnetic Acoustic Emission Signal Processing Based on Optimized Variational Mode Decomposition

Abstract

Electromagnetic acoustic emission (EMAE) signal is a kind of faint signal and susceptible to noise interference. In order to obtain pure and representative impairment signal in metal devices, which is crucial to the subsequent impairment identification and early warning, an optimized VMD method is implemented. Moreover, to realize the denoising of EMAE signal properly, we compare common denoising ways in the simulation and real collected signal. Aiming at the serious influence of the decomposed level K and penalty factor α, and uncertainties of parameter selection in the variational mode decomposition (VMD), we utilize genetic algorithm (GA) for optimizing VMD, namely GAVMD. Firstly, this approach views sample entropy as fitness function, and global minimum fitness value is used as optimization objective to screen out the best VMD parameter combination. Then, a series of intrinsic mode functions (IMFs) are obtained, and the denoising signals are reconstructed according to the correlation coefficients. The results show that the improved GAVMD algorithm has the better effect in the suppression of non-stationary signal noise, with SNR of 18.4684 and RMSE of 0.0017. In addition, we also develop the reservoir computing modeling approach to predict the noise-reduced data to further validate the superiority of GAVMD.