BPNN-Based Demand Models and Multihazard Performance Assessments for Transmission Towers Subject to Winds and Earthquakes

In light of the growing emphasis among researchers on extreme hazards, the approach to assessing the performance of engineering structures has transitioned from a singular focus to a more comprehensive, multifaceted methodology. The engineering demand model (EDM), which was previously employed for single-hazard performance evaluations, is now inadequate for the demands of multihazard assessments. To advance the multihazard design framework for engineering structures, this paper presents an EDM that integrates the backpropagation neural network (BPNN) algorithm within a machine learning context to analyze the potential correlations between multihazard intensity and structural response. The fitting accuracy of the BPNN-based model is rigorously compared and analyzed against that of traditional models, utilizing the results from the nonlinear dynamic response of a transmission tower subjected to the combined effects of wind and earthquake. Furthermore, a fragility analysis is performed using the BPNN-based model to ascertain the damage probability of transmission towers under the combined effects of the winds and earthquakes. The findings of this study are as follows: (1) under multihazard conditions, the BPNN-based model demonstrates superior capability in capturing the nonlinear characteristics of hazard-induced responses, thereby offering a more precise characterization than conventional models; (2) as the intensity of wind and earthquake hazards escalates, both the displacement response of the transmission towers and their probability of failure across various limit states increase; and (3) the increase in the damage probability of transmission towers due to heightened wind load intensity is more significant than the increase associated with ground motion effects, indicating a pronounced sensitivity of the transmission towers to wind. The BPNN-based EDM proposed herein is applicable to the performance assessment of diverse engineering structures under multihazard scenarios, providing a valuable reference for the investigation of multihazard resilience in engineering design.