Data-Driven Fragility Analysis of Transmission Lines Against Typhoon Disasters: A Bo-GRU-Based Approach

Fragility analysis of power system components can provide useful guidance for system disaster preparedness against extreme disasters. In practical applications, the reliability of components is always influenced by multiple factors. However, most studies adopt one-dimensional fragility curves to describe the relationships between disaster intensity and component status. In this context, a data-driven fragility assessment framework for transmission lines under typhoon disasters is proposed based on massive heterogeneous data and deep learning methods. Firstly, data preprocessing is conducted using data normalization methods. Then, a Bayesian-optimized Gated Recurrent Unit (Bo-GRU) model is constructed to estimate the accumulated failure probability of transmission lines during the typhoon period. Finally, the effectiveness of the proposed method is validated by numerical experiments on the IEEE 9-bus test system. The results show that the proposed Bo-GRU model can effectively capture the intricate relationships between multiple influence factors and component fragility, realizing more reliable fragility assessment.