The resilience of critical infrastructure is paramount for power systems prone to cyclonic activities characterized by varying wind speeds. This study conducts a thorough examination of the impact of cyclone lasting for a week on transmission lines for each passing day. Piecewise Cubic Hermite Interpolating Polynomial (PCHIP) is applied to extract hourly wind speed data for the cyclone. Sequential Monte Carlo Simulation (SMCS) is used to model the day-wise hourly transmission line fragility and determination of line vulnerabilities across high, medium, and low wind speed zones. Fragility curves are developed to depict the correlation between wind speeds and line failure probabilities in each zone. The line outage scenarios are generated based on the critical wind speeds and collapse wind speeds for each zone. Line Vulnerability Index (LVI) is proposed to determine the vulnerability and the resilience status of the transmission lines. Machine Learning (ML) based models, FFNN (Feed Forward Neural Network) and RFC (Random Forest Classifier), are proposed for assessment of infrastructure resilience by vulnerability ranking of the lines. Hourly wind speed and hourly line failure probability are the inputs to train the ML models and, LVI and ranking are predicted for the IEEE 30-bus system.
Read full abstract