Research on CPO-ICEEMDAN-WTD Signal Processing Method for Enhanced Magnetic Memory Detection

Enhanced magnetic memory detection signals often contain numerous environmental interferences. Conventional signal processing methods, such as improved complete ensemble empirical modal decomposition with adaptive noise (ICEEMDAN) and wavelet threshold de-noising (WTD), face significant challenges in determining parameters such as noise standard deviation (Nstd), number of realizations (Nr), wavelet basis function, and decomposition level. ICEEMDAN suffers from some feature signal loss, complicating the recognition of stress concentration, plastic deformation, and microdefects. This study addresses these issues by proposing a joint ICEEMDAN-WTD method based on the crested porcupine optimizer (CPO) algorithm called CPO-ICEEMDAN-WTD. The proposed approach leverages CPO to optimally select Nstd and Nr using the minimum entropy value method, performs ICEEMDAN decomposition to obtain the signal, and classifies it through the mutual information entropy (MIE) method. The noise components are then processed with WTD, and the noise-reduced signal is reconstructed. This method reduces noise in enhanced magnetic memory detection signals for Φ2× 0.5mm indentation defects, using signal-to-noise ratio (SNR) and information entropy (IE) as evaluation indicators. Experimental results demonstrate that the CPO-ICEEMDAN-WTD method significantly enhances noise reduction, with the SNR of the horizontal component Bx, normal component By, tangential component Bz, and their gradient values in the magnetic field yielding a maximum improvement of 312.49%, and IE minimized by 43.13%.

Duke Scholars

Author Lihong Wang Psychiatry & Behavioral Sciences, Adult Psychiatry & Psychol ...