递归式变分模态分解在电力变压器声音信号降噪中的应用

Recursive Variational Modal Decomposition for Noise Reduction of Power Transformer Sound Signals

针对电力变压器声音信号易受噪声干扰问题,提出了一种基于递归式变分模态分解(Recursive Variational Mode Decomposition,RVMD)的声音信号处理方法.首先,对三种常见的模态分解方法进行仿真分析,对比三种方法在信号自动分解和抗模态混叠方面的特点;采样高斯信号对以上三种方法进行等效滤波特性分析,阐述其滤波特性;最后,选用RVMD方法对实测电力变压器声音信号进行降噪处理.仿真与实测数据分析表明:RVMD具有类似变分模态分解(Variational Mode Decomposition,VMD)的带通滤波特性,具有抗模态混叠能力;RVMD可有效去除电力变压器声音信号中的噪声和干扰,具有良好的工程应用价值.

A recursive variational modal decomposition(RVMD)based sound signal processing method is proposed for power transformer sound signals that are susceptible to noise interference.First,three decomposition methods,RVMD,VMD and EMD,are simulated and analyzed to compare their characteristics in terms of automatic signal decomposition and anti-modal aliasing;the Gaussian signal is sampled and the equivalent filtering characteristics of the above three methods are analyzed to elaborate their filtering characteristics;finally,the RVMD method is selected to perform noise reduction on the measured power transformer sound signal.Simulation and analysis of the measured data show that RVMD has band-pass filtering characteristics similar to VMD and has anti-modal aliasing capability;RVMD can effectively remove noise and interference from power transformer sound signals and has good engineering application value.