An Efficient Aco-Based Refinement Strategy For Enhancing The K-Means Algorithm

The K-means algorithm remains one of the most widely used techniques for unsupervised clustering; however, its performance is often limited by sensitivity to centroid initialization and premature convergence to local optima, particularly in highdimensional and heterogeneous energy consumption datasets. This study proposes an ACO-Based Refinement Strategy that integrates an incremental mechanism within the Ant Colony Optimization phase to enhance clustering quality without significantly increasing computational complexity. Experimental findings indicate that the proposed approach outperforms both conventional K-means and the standard K-means/ACO hybrid, achieving superior structural quality and improved computational efficiency. The method establishes a robust balance between metaheuristic exploratory capability and controlled computational cost in large-scale clustering scenarios.