Microgrid Protection using Empirical Mode Decomposition and MRMR Feature Ranking Approach

Microgrid deployment has increased significantly in recent years due to the advantages of supplying load locally with increased reliability. However, the intermittent nature of renewables and presence of bi-directional power, and various modes of operation constitute different set of protection problems. This article presents a novel approach to microgrid protection using empirical mode decomposition (EMD) signal processing and minimum redundancy maximum relevance (MRMR) feature ranking coupled with k-nearest neighbor (KNN) classification. The proposed methodology leverages EMD to decompose the voltage and current signals into intrinsic mode functions (IMFs), which are then ranked using MRMR algorithm to select the most relevant features. KNN is then employed for fault detection and classification. The proposed method has been tested on a modified IEEE 13-bus distribution microgrid and performance indices validate the effectiveness of the proposed methodology in detecting and classifying faults.