LAUSR

In this research, the moisture diffusion model for concrete was inversely estimated using artificial neural network (ANN) and the data collected from virtual experiments. In addition, the moisture distribution was predicted using the ANN model in numerical analysis. For inverse estimation, virtual experimental data were used. The virtual experimental data were generated by adding noise to the moisture distribution obtained by a numerical simulation using a known moisture diffusion model. ANNs of two architectures were used in the inverse estimation. For performance test, the inversely estimated ANN model and the known moisture diffusion model were compared. The predicted humidity distribution using the ANN and virtual experiment data were also compared. The inversely estimated ANN model was in a good agreement with the known moisture diffusion model used for the virtual experiment.