Comparison of several clustering methods in the case of electrical load curves classification

In the electricity market, it is highly desirable for suppliers to know the electricity consumption behavior of their customers, in order to provide them with satisfactory services with the minimum cost. Information on customers´ consumption pattern in the deregulated power system is becoming critical for distribution companies. One of the suitable tools for extracting characteristics of customers is the clustering technique. Selection of better methods among several existing clustering methods should be considered. Therefore, in this paper, we evaluate the performance of Classical K-Means, Weighted Fuzzy Average K-Means, Modified Follow the Leader, Self-Organizing Maps and Hierarchical algorithms that are more applicable in clustering load curves. The performances were compared by using two adequacy measures named Clustering Dispersion Indicator and Mean Index Adequacy.