A Method to Forecast Storm-Caused Distribution Grid Damages Using Cost-Sensitive Regression Algorithm

The forecasting of the grid damages caused by an extreme weather event, such as storm, has become an important to improve grid resilience. It can be used to prepare the mitigation actions or to plan a rapid restoration during the storm event. In this manner, this paper presents a methodology to forecast the damages caused by the storm events on the distribution grid. First, it integrates three different types of historical data such as the storm data, local weather, and distribution system damages. Then, this integrated data is used to develop the forecasting model by employing the Pearson Correlation Coefficient (PCC) and regression analysis. The accuracy of the model is improved by applying a cost-sensitive learning algorithm on the regression analysis. The proposed methodology is verified by using the actual data of the storm event occurs in South Korea.