A hybrid metaheuristics-based algorithm for electricity load curves profiling

Abstract

Clustering-based load profiling utilizes unsupervised machine learning algorithms to form homogenous clusters composed by electricity load curves with similar characteristics. Due to the importance of load profiling in modern power systems, a variety of clustering algorithms of different type and complexity has been proposed in the technical literature. Nevertheless, no attention has been placed in metaheuristics-based clustering. The present paper proposes a novel hybrid algorithm that is composed by two clustering algorithms, namely a hierarchical algorithm and a meta-heuristic one. A comparison takes place with the most common algorithms of the literature. Experimental results indicate the robustness of the proposed approach to the electricity load curves profiling research problem. © 2021 IEEE.