CF-PSO based loss sensitivity clustering technique to identify optimal DG allocation nodes for energy efficient smart grid operation

Abstract

Recently there has been increasing interest in improving smart grid energy efficiency using computational intelligence. In a smart grid, Distributed Generation (DG) has gained much attention due to numerous advantages. However, inappropriate selection of DG allocation nodes may increase the total power loss of the distribution system. Therefore, it is important to identify similar type of nodes where energy efficient DG allocation is possible. In this paper, Constriction Factor Particle Swarm Optimization (CF-PSO), which is a major variant of Swarm Intelligence (SI), has been used with traditional well studied k-means algorithm to enhance the clustering performance. Experiments are performed considering test data from UCI repository of machine learning databases which shows that the CF-PSO based hybrid clustering outperforms the traditional k-means algorithm. This improved clustering algorithm is then employed to identify the potential nodes for DG allocation using loss sensitivity indices. Extensive experiments have been carried out considering IEEE benchmark 123 node test distribution system to justify the clustering output. Results show that the clustering algorithm provides an insight to select the appropriate DG integration nodes for power loss reduction.