In this project, 6 RC slabs with the same steel percent and different length and width of CFRP were tested and compared with the similar sample without CFRP. The dimension of slab was 2800×400×120 mm and the used length of CFRP was 700, 1100, and 1500 mm in two type of S512(width=50mm , thickness=12mm) and S812(width=80mm , thickness=12mm). The samples were tested and 5 parameters before first crack and 4 parameters after cracking were measured. The General Regression Neural Network (GRNN) was the first analytical method that has applied for prediction of data. The Feed Forward Backprop (FFB) was the second method with a pre regression method for data gathering to increase the number of data for training, verifying, and testing. The two used method had minimum error and maximum correlation coefficient. The amount of MSE & RMSE was in the range of 6E-6~ 0.0011 & 0.0024~0.03 in GRNN method and 9.91E-7 ~ 0.00158 and 0.00099~0.0398 in FFB method. The correlation coefficient for output data was closed to 1.
Keywords: CFRP, S512, S518, GRNN, FFB, MSE, RMSE