Equilibrium Optimizer-Based Variational Mode Decomposition Method for Partial Discharge Denoising

Partial discharge (PD) of polymide (PI) is an important cause of insulation deterioration in high-frequency power transformers (HFPT), while the acquisition of partial discharge signals under high-frequency voltage is prone to noise interference. Therefore, this work proposes a variational mode decomposition (VMD) method based on equilibrium optimizer (EO) optimization, which adopts minimum envelope entropy to optimize the parameters, including the number of modal decomposition K and the penalty factor a. Based on the obtained optimal parameters, decompose the original signals into a series of band-limited intrinsic mode functions (BIMF). According to kurtosis criterion, effective components are selected for signal reconstruction. The residual white noise is further removed by wavelet threshold denoising. This work obtained the denoising analysis of polyimide PD signals under the measured voltage of 40 kHz and 2.5 kV at 100 degrees C. Compared with wavelet threshold method and empirical mode decomposition method, the results show that the denoising method combining EO-VMD and wavelet threshold has better denoising effect. It can retain the characteristics of partial discharge, and be applied to the actual partial discharge test.