A Data-driven Framework of Resilience Evaluation for Power Systems under Typhoon Disasters

Abstract

This paper proposes a data-driven framework of resilience evaluation for power systems under typhoon disasters. A typhoon scenario generation model based on the recurrent neural networks (RNNs) and long-short term memory unit (LSTM) using historical typhoon data are presented. Under generated typhoon scenarios, the resilience of different components of power systems and the entire network are evaluated. We apply the proposed framework to an instance of IEEE-13 bus system to demonstrate its feasibility, and the results prove that our method outperforms in accuracy of resilience assessment than the traditional methods that based on hypothetical typhoon scenarios or single historical typhoon scenarios. Our proposed data-driven framework and the resilience evaluation results can provide system managers with guidance on power system planning and resilience enhancement.