A combined thermographic analysis-Neural network methodology for eroded caves in a seawall

An application of an artificial neural network (ANN) combined with thermographic analysis for estimating the depth of eroded caves in a seawall is presented in this paper. A model experiment was first conducted in a sandbox using a thermographic device to detect the interior conditions of a structure from its temperature changes measured on the surface. The temperature difference calculated from the air temperature and the measured concrete surface point on a thermographic image was obtained for the neural network. Based on the laboratory data, an optimum ANN model for the estimation of the depth of eroded caves in a seawall was established by using four input factors: the site temperature, humidity, thermographic area, and the temperature difference. The model was verified using data from a seawall in Tainan City, Taiwan. From the results, it was found that the present ANN model efficiently estimates the depth of eroded caves in a seawall.