Prediction of effective strain for near-surface mounted FRP strips in strengthened RC beams using a neuro-fuzzy network

Experimental and analytical research have shown that the near-surface mounted (NSM) FRP technique is efficient and important in strengthening reinforced concrete (RC) beams. However, determining the effective strain for FRP strips remains a major challenge. In this paper, we investigated the possibility of using intelligent systems to determine the effective strain of FRP strips. For this purpose, we used the adaptive neuro-fuzzy inference system (ANFIS) to predict the effective strain of FRP strips. The proposed model employed variable compression strength of concrete, the bond length of the strengthened strip, and equivalent reinforcement ratio as inputs while the effective strain of FRP strips was the output. We compared the results of the ANFIS model with the experimental results that we collected from published literature using the coefficient of determination (R2) and root-mean-square error (RMSE), the results indicated that the considered neuro-fuzzy model was able to accurately and rapidly predict the effective strain of FRP strips in comparison to the artificial neural network and fuzzy approaches, where R2 = 0.971, RMSE = 0.04757 were achieved with ANFIS while this values did not exceed the (R2 = 0.923, RMSE = 0.07305) for the rest of approaches.