Entropy Based Spectral Clustering For Distribution Network With High Penetration of DGs

Abstract

High penetration of distributed generation (DG) into the distribution network (DN) causes voltage limit violation at several buses. With the evolution in inverter technology, DGs offer an optimistic solution to maintain the voltage profile by controlling its active and reactive power. Centralized voltage control (CVC) using the voltage controlling capability of available DGs in the distribution network is not a good solution due to the size of the DN. Additionally, perturbation at any DGs does not impact the voltage level at all buses equally. For maintaining the voltage profile at a few buses, power curtailment from all the DGs is not an economically fair solution for the utility grid. Therefore, decentralized voltage control is considered as a more practical and economical approach to maintaining the voltage profile at all buses. In this paper, novel entropy-based spectral clustering has been proposed to partition the DN. For implementing the proposed method, the impact of DGs on voltage deviation has been measured in terms of entropy. Further using entropy, spectral clustering has been performed. The proposed clustering decouples the distribution network in such a way that available DGs in clusters are capable of maintaining voltage profiles in the clusters. Further effectiveness of the proposed method has been done on IEEE 33 test system.