Network Clustering for Distribution System with Photovoltaic System and Electric Vehicles
Primary author: Lusha Wang
Faculty sponsor: Noel Schulz
Primary college/unit: Voiland College of Engineering and Architecture
The penetration of both Photovoltaic (PV) system and electric vehicles (EVs) are increasing rapidly in distribution systems, which brings challenges to system operation. The distribution system should meet some requirement to operate safely, the most important one being that the voltage magnitude should be within the desired range. Compared with the traditional centralized voltage control where all the information of the system is obtained and an optimization problem is solved in the control center, the decentralized voltage control is more flexible and consumes less computation time, making it suitable for large-size distribution system. To realize the decentralized voltage control, proper division of the system and choice of regional agents should be well determined. The rapid change of system configuration and DG output as well as EV movement brings the need of rapid and frequent determination of network clusters. We proposed a new algorithm to cluster a distribution system with PVs and EVs. The modularity index is used to evaluate the clustering result. The original modularity index is purely based on system structure, so we add the information of PV generation and EV driving distance into the index to accommodate power system properties. The Louvain algorithm with the aim of maximizing the modified modularity index is used to cluster the distribution system, which shows great computation speed and reasonable results. An IEEE 123 node system is used to demonstrate the clustering result, with comparison of network clustering based on solely structure, structure with DG output and the three together.