Solar Customer Detection based on Power Consumption Patterns

Nowadays, solar photovoltaic (PV) systems are rapidly growing worldwide. The utility needs to grasp the changing trends for power system planning, and penalize an illegal installation solar system in order to prevent impacts on the grid. Therefore, this paper aims to detect a solar customer from weekly consumption patterns by three classification algorithms: Logistic regression, cosine similarity and K-nearest neighbors. Furthermore, the clustering methods, K-means and Density-based spatial clustering of applications with noise (DBSCAN), are utilized for similarity grouping, and computational cost reduction with the two stage clustering technique. The study has been conducted with the non-resident customers in Thailand, the classification results are discussed.