As an important carrier of equipment operation status, electromagnetic signals are of great significance to power system monitoring and equipment fault diagnosis. This paper proposes an electromagnetic signal feature processing method that integrates dynamic mode decomposition (DMD) and multi-dimensional correlation analysis, and constructs a complete technical framework of “feature extraction correlation analysis and redundancy elimination”. This method first extracts the time domain, frequency domain and energy domain characteristics of electromagnetic signals through multi-resolution dynamic mode decomposition (MRDMD), and captures the difference in electric waveforms with dynamic time warping (DTW). Secondly, the intrinsic relationship between the characteristics and the device state is explored through linear and nonlinear correlation analysis; finally, the key features are screened based on the Variance Inflation Factor (VIF) and correlation intensity. Experimental results show that this method can effectively retain features that are strongly related to the equipment metrological characteristics and provide accurate feature support for the status evaluation of power equipment.