Fuzzy based clustering of smart meter data using real power and THD patterns

In most of the electric utilities of the world, the classification of customer is based on the amount and purpose of the load sanctioned. Classification of customer based on the duration of energy usage is possible by smart meters. With the smart meter data, time of energy consumption, daily usage pattern and actual volume of energy usage offers better insight of customer consumption pattern. Clustering is the process forming groups (or cluster) based on the similarity among the data. By applying clustering methods various groups of customer can be formed based on their consumption pattern. So far clustering is based on amount of real power sanctioned for the customers but due to increased addition of renewable energy and non linear loads to the utility grid raises issues like voltage fluctuation, harmonic resonance and Total Harmonic Distortion(THD). In this paper, K-means method is used for the clustering the real power and THD load data of smart meter independently. Fuzzy logic technique is used to illustrate the combined effectiveness of real power and THD data for clustering customers and demonstrated with the real time smart grid project. (C) 2017 The Authors. Published by Elsevier Ltd.