Multi‐hazard typhoon and earthquake collapse fragility models for transmission towers: An active learning reliability approach using gradient boosting classifiers

Natural hazards such as earthquakes and typhoons pose a considerable threat to the power transmission grid. In regions prone both hazards, overhead infrastructure may experience multi-hazard effects where two consecutive can occur within an interval shorter than time that would be needed repair system for damages occurred in first event. This consequently increases susceptibility damage during subsequent hazard. this context, paper presents set of fragility models earthquakes. The computational cost numerical simulations analyses is high due complex nonlinear behavior towers, numerous sources material geometric uncertainty, multifaceted events. To tackle challenge, we introduce novel active learning approach probabilistic classification physics-based collapse data. An initial pool labeled data generated by performing limited number history on random realizations high-fidelity finite element model common double circuit tower. A machine learning-based extreme gradient boosting (XGBoost) trained via query committee (QBC) task predicting probability belonging class (i.e., collapse). Active established leveraging disagreement information between learners points are sequentially added collapse/survival Using approach, typhoon earthquake surfaces towers different directions sequences developed. Results point significance structures. They also reveal efficacy introduced reducing complexity assessments. developed study constitute key step toward resilience assessment optimal management