Electrical Load Clustering Using K-Means Algorithm for Identification of Village Clusters to Draw Optimal Power from Distributed Solar Generating Plants

Abstract

In recent days, renewable energy Distributed Generating (DG) plants are becoming reality in village electrification projects, where village clusters are identified to draw electricity from installed solar power plants around group of villages. Clustering results can be exploited by various energy providers to identify load groups among the villages and tailor-make more attractive time-varying tariffs and power generation schemes for their customer needs. In this article, we explored a clustering algorithm to compare the typical load profiles of different villages are different days of the week. We found that better results are obtained if the clustering is not performed on the entire data, but on some subset of the extracted data. These clusters were thus identified based on their Vicinity and Total Power Load requirements of the villages. In particular, despite the relevant differences among the several compared countries, we obtained the interesting result of identifying a single feature, the Village Level Load, which is alone able to identify how much power is needed in a village. On identifying the clusters, if total load requirements are more than the generating capacities of the plants, the villages are re-clustered until the load requirements are met for a cluster. The task is accomplished based on unsupervised machine learning technique called K-Means Clustering, and has been implemented using Python programming language where positive results were observed.