Comparisons among Bat algorithms with various objective functions on grouping photovoltaic power patterns

Photovoltaic power pattern clustering is fundamental in providing enhanced knowledge on the impacts of integrating photovoltaic systems into the electrical grid without extensive analysis and simulations. This paper investigates a set of clustering algorithms and validity indices to find the most efficient ones in grouping photovoltaic power patterns data. Furthermore, the introduction of the recently-developed bio-inspired optimization method, Bat, with various objective functions in clustering photovoltaic power patterns is presented. In order to evaluate the clustering results in a comprehensive manner, six internal validity indices are employed and a method to determine the optimum number of clusters is introduced. The clustering results on two datasets show that bio-inspired clustering algorithm Bat based on within-cluster-sum-of-squares to between-cluster variation (Bat WCBCR) as an objective function produces significantly high separated and well compact clusters.