General regression neural network (GRNN) for the first crack analysis prediction of strengthened RC one-way slab by CFRP

摘要

In this study, six strengthened RC one-way slabs with different lengths and thicknesses of CFRP were tested and compared with a similar RC slab without CFRP. The dimensions of the slabs were1800 x 400 x 120 mm and the lengths of CFRP used were 700, 1100, and 1500 mm, with different thicknesses of 1.2 and 1.8 mm. The results of the experimental operation for the first crack were used to generate general regression neural networks (GRNNs). Concerning the limited data for training and testing, the different data were extracted seven times for use as training and testing data. In this case, the optimum run was evaluated and compared with the experimental results. The results indicate that the amount of MSE and RMSE was acceptable and the correlation coefficient was close to 1.