재킷 구조물에 대한 장기 변형률 계측 및 신경망 기반 예측모델 구성

Abstract

Structural strain responses of a jacket-type offshore structure are analyzed and the prediction model is constructed based on the neural networks technique using long-term measurement data. Uldolmok tidal current power plant structure under severe tidal environments is utilized as an example structure. From the measurement data during normal operation, it is observed that strain responses are obviously fluctuated with M2 and M4 tidal constituent periods and also with relatively short period of about 11 min due to the peculiar tidal characteristics in the Uldolmok strait. The neural networks based prediction model is also constructed for the signal-based structural health monitoring system, and the predicted strain responses are well coincident with the measured data. severe tidal environments is utilized as an example structure. From the measurement data during normal operation, it is observed that strain responses are obviously fluctuated with M2 and M4 tidal constituent periods and also with relatively short period of about 11 min due to the peculiar tidal characteristics in the Uldolmok strait. The neural networks based prediction model is also constructed for the signal-based structural health monitoring system, and the predicted strain responses are well coincident with the measured data.