Exploring the Potential of Metaheuristics-Based Clustering in Load Profiling Applications

The clustering tool is ideal for scientific and engineering problems where no information about the data structure is available. Usually, in load profiling applications no sufficient information is a priori given for the consumers' classes, their consumption patterns and other attributes that will aid on building a descriptive model for the consumption trends. Many clustering algorithms of different type have been proposed. However, no attention has been placed so far in the utilization of metaheuristics for building robust clustering tools. The scope of this study is to develop a hybrid algorithm that combines partitional clustering and Particle Swarm Optimization (PSO). Simulation results indicate the superiority of the proposed algorithm over conventional clustering approaches.