A three-stage scheme for consumers' partitioning using hierarchical clustering algorithm

The clustering of any type of consumers (residential, commercial, industrial) is of great importance in the operation of Smart Grids. In this paper, we propose a three-stage hierarchical scheme for residential consumers' partitioning using the Hierarchical clustering algorithm. The aim of this study is to cluster the consumers in well-separated and compact clusters using information from the near past (almost real time). The usage of electricity from a resident to another varies and this information can be used from the system operator for improving the efficiency of the distribution network. The first stage corresponds to the consumers' clustering of the distribution grid using data driven every three minutes (simulation time) from the meter of each residency. The procedure of the second stage takes part every a specific number of hours, called h, that is defined by the user. The average value of each of the k*20 clusters formed the last h hours is used as input for the hierarchical algorithm. In the third stage and in the end of each h hours, the average value of the data in each cluster is calculated and each consumer is reassigned to the cluster where the a distance metric is minimized. The results of the second stage provide deeper information about the load patterns existing each hour in the distribution grid. This information can be used from suppliers to design the energy tariffs for suiting better to the consumers' needs. This approach is tested using the IEEE-13 test feeder. The data are driven from 56 residencies.