Fair hierarchical clustering of substations based on Gini coefficient

Dajun Si,Wenyue Hu,Zilin Deng,Yanhui Xu

DOI: https://doi.org/10.1016/j.gloei.2022.01.009

2021-12-01

Global Energy Interconnection

Abstract:For the load modeling of a large power grid, the large number of substations covered by it must be segregated into several categories and, thereafter, a load model built for each type. To address the problem of skewed clustering tree in the classical hierarchical clustering method used for categorizing substations, a fair hierarchical clustering method is proposed in this paper. First, the fairness index is defined based on the Gini coefficient. Thereafter, a hierarchical clustering method is proposed based on the fairness index. Finally, the clustering results are evaluated using the contour coefficient and the t-SNE two-dimensional plane map. The substations clustering example of a real large power grid considered in this paper illustrates that the proposed fair hierarchical clustering method can effectively address the problem of the skewed clustering tree with high accuracy.