Wu, Ting-YanTing-YanWuRIH-TENG WUWang, Ping-HsiungPing-HsiungWangLin, Tzu-KangTzu-KangLinChang, Kuo-ChunKuo-ChunChang2026-03-122026-03-122024https://www.scopus.com/record/display.uri?eid=2-s2.0-105027865226&origin=resultslisthttps://scholars.lib.ntu.edu.tw/handle/123456789/736245Visual surface damage patterns of reinforced concrete (RC) bridge columns are critical indicators to evaluate the performance level of the bridges. To assess potential failure subjected to earthquakes, laboratory experiments or in-situ tests are often required to investigate the damage status under seismic demands. However, such experimental works are quite costly, time-consuming, and labor-intensive. In this study, a novel failure prediction framework is proposed based on conditional generative adversarial networks (CGAN) to forecast the high-fidelity damage patterns of concrete bridge columns given the column design parameters and a user-desirable performance level, i.e., Damage Index. Trained with merely 110 samples collected from the cyclic loading test conducted at National Center for Research on Earthquake Engineering (NCREE) in Taiwan. Also, the proposed network eliminates the error between simulated and experimental hysteresis curves and extracts comprehensive features from the hysteresis loops for enhancing the model performance. Extensive experiments have demonstrated that the proposed framework synthesizes damage patterns with superior fidelity, providing bridge engineers with a platform to evaluate the potential failure patterns during seismic analysis and design.falseconference paper2-s2.0-105027865226