주파수응답기반 인공신경망을 이용한 항만케이슨 지반-구조 경계부의 손상식별

Abstract

In this paper, damage identification for foundation-structure interface of harbor caisson structure using frequency response-basedartificial neural network (ANN) is presented. To achieve the objective, the following approaches are implemented. Firstly, frequency-response-basedANN model is designed. Power spectral densities (PSDs) are employed as the input parameter. Secondly, caisson-structure FE model isconstructed to evaluate the performance of ANN-based damage identification procedure, numerically. Thirdly, training scenarios are designedand changes of frequency domain responses are analyzed. Finally, neural networks are trained for the training scenario andfoundation-structure damage of harbor caisson is identified.lemented. Firstly, frequency-response-basedANN model is designed. Power spectral densities (PSDs) are employed as the input parameter. Secondly, caisson-structure FE model isconstructed to evaluate the performance of ANN-based damage identification procedure, numerically. Thirdly, training scenarios are designedand changes of frequency domain responses are analyzed. Finally, neural networks are trained for the training scenario andfoundation-structure damage of harbor caisson is identified.