A generic load profiling technique using fuzzy classification

This paper proposes a new method for the automatic classification of large sets of electric demand profiles into predetermined categories. Such a method is essential for tariff study/setting and charging purposes. The proposed approach requires the use of a fuzzy membership function (FMF) that is modelled through statistical considerations. Although recent research into the theory of fuzzy sets has been abundant, its application to power systems has been slow. Classifying electric load profiles have been tackled by different researchers using a variety of standard known approaches for pattern recognition. However, recently the fuzzy mathematics approach has been one of the most popular pattern recognition and clustering methods used in other areas of engineering applications. In this paper, the authors argue that application of the fuzzy membership functions to classifying the load profiles is both natural and appropriate as the patterns have vague boundaries. Using this approach it has been possible to recognise similarities and classify patterns as load profiles of different customers. In the paper, the authors describe a straight forward method, which uses a spreadsheet model, that has been developed to examine the shape of a variety of different load patterns. Test results of the algorithm, based on actual customer load information are used to highlight the ease of use, efficiency, accuracy and finally the robustness of the proposed methodology