摘要
This research project focuses on enhancing civil infrastructure adaptation using machine learning models. We trained and validated models like artificial neural networks and LSTMs with experimental and numerical data. These models, combined with weather and vulnerability models, helped evaluate retrofit actions from engineering and economic perspectives. LSTM models effectively handled time-dependent data due to their gated memory units. We developed strategies for coastal infrastructure adaptation, focusing on bridge vulnerability, multi-criteria optimization, and robustness against climate uncertainties. Sensitivity analysis addressed future economic and climate uncertainties, offering decision-makers insights into the feasibility of various adaptation actions.
