Spatial distribution assessment of power outage under typhoon disasters. (November 2021)

• Record Type: Journal Article
• Authors: Hou, Hui; Zhu, Shaohua; Geng, Hao; Li, Min; Xie, Yufeng; Zhu, Ling; Huang, Yong
• Is Part Of: International journal of electrical power & energy systems. Volume 132(2021)
• Publication Date: 2021-11
• Subjects: Typhoon disaster -- Power outage -- Multi factors -- Data-driven -- Prediction and evaluation -- Feature grid; Electrical engineering -- Periodicals; Electric power systems -- Periodicals; Électrotechnique -- Périodiques; Réseaux électriques (Énergie) -- Périodiques; Electric power systems; Electrical engineering; Periodicals; 621.3
• DOI: 10.1016/j.ijepes.2021.107169
• Languages: English
• ISSNs: 0142-0615

Abstract

Highlights: Using Smote algorithm to deal with the data, avoid the imbalance of data. A 1km*1km cell is used to predict outage area. It can ensure prediction accuracy. A two-stage data-driven prediction method is proposed to improve the prediction accuracy. ArcGIS is used to visualize prediction results, so as to facilitate the formulation. Abstract: Pre-disaster power outage prediction plays an important role in the safe operation of the distribution network and its restoration after disasters. Accurate outage prediction can provide guidance for the power production departments. However, power outage prediction is a challenging task due to massive volumes of heterogeneous data and complex causes and impacts on the grid of various factors. To facilitate an effective prediction, this paper develops a prediction algorithm and evaluation method with a focus on the spatial distribution of power outages. Our prediction algorithm uses multi-sources of information including meteorological, geographical, power grid data, and we consider 14 features. To integrate feature data, a 1 km*1km cell is established to build the spatial model for the distribution grids. We process the historical sample data associated with each cell and predict the power outage area based on the random forest algorithm. To further improve the prediction accuracy, the prediction of outage area is combined with the prediction of outage probability to correct the predicted evaluation result. The evaluation level is used to determine the order of emergency repair. The proposed method is validated through a numerical case study under typhoon 'Mujiage' that happened in 2015, the accuracy can reach 92.44%.