Wind-induced fragility analysis of a transmission tower based on multi-source monitoring data and deep learning methods

Abstract

Structural health monitoring (SHM) technology can provide useful data for the assessment of the wind-resistant capacity of a transmission tower. However, most studies on wind-induced fragility assessment are based on a significant number of simulations. In this context, a wind-induced fragility assessment framework for a transmission tower is proposed based on multi-source monitoring data and deep learning methods. The framework consists of three main steps. First, methods for processing missing data and denoising the monitoring data are developed. Subsequently, a surrogate model of structural dynamic response under wind field data input is established using long short-term memory (LSTM) networks, and the optimal model hyperparameters are obtained by Bayesian optimization. Finally, wind field data with a uniform distribution of wind speed intensities are generated, and the structural dynamic responses are supplemented by surrogate model prediction. Fragility curves are generated under a variety of wind directions. The proposed framework was validated, and its applicability and efficiency were demonstrated using monitoring data from a real transmission tower. The results indicated that wind direction has a significant influence on fragility curves. The proposed framework is capable of efficiently expanding the database of wind-induced dynamic responses and realizing more reliable and rapid fragility assessments.