EvoK-Means: A Differential Evolution Optimized K-Means Algorithm for Clustering Electric Power Consumption Patterns

This paper introduces EvoK-Means, a Differential Evolution-optimized K-Means algorithm for improved clustering of electricity consumption data. Traditional methods like K-Means and DBSCAN often struggle with complex datasets containing noise and outliers, leading to poor clustering results. EvoK-Means integrates Differential Evolution with K-Means to optimize initial cluster center selection, enhancing clustering accuracy. Evaluated on a large-scale electricity consumption dataset using Silhouette and Davies-Bouldin indices, EvoK-Means outperformed traditional methods, achieving better intra-cluster cohesion and inter-cluster separation. The method identifies diverse consumption patterns, offering valuable insights for load management, anomaly detection, and energy-saving strategies.