Natural Disaster Management Using Machine Learning for Resilient Electrical Grids

The purpose of this paper is to support the electrical grid to make it resilient using new-age artificial intelligent technology. The availability of daily weather data and past records of disasters and incidents assist in identifying certain natural disaster phenomena. Firstly, the general background, and crucial concepts in disaster identification to help electrical grid are summarized. Secondly, data modeling of interpolation techniques is discussed. We have incorporated machine learning (ML) models using an integration of two databases based on temporal and spatial resolution. We proposed a quantitative assessment approach for disaster classification analysis solutions using weather and disaster datasets. The proposed model improves the resiliency of grid from low-probability and high-risk events. Lastly, we discussed the evaluation of results based on the performance parameter precision, recall, F-score, and accuracy. The quantitative results recommend that the proposed methodology is capable of measuring the impact of disasters. ML classification analysis is incorporated for the severity calculation. This comprehensive statistical analysis suggests that proposed techniques may contribute to disaster-prone areas to improve power shortage and the effectiveness of a resilient grid for better safety. ML algorithms such as Artificial Neural Network (ANN), Support Vector Machine (SVM), and Random Forest (RF) are used for decision support systems that improve grid resiliency.