Prediction of shear strength of reinforced concrete beams using ANFIS and SVR Algorithms

The shear strength of reinforced concrete beams changes depending on the mechanical and geometrical parameters of the beam. Accurate estimation of shear strength in reinforced concrete beams is a fundamental issue in engineering design. However, the prediction of shear strength in these beams does not have very high accuracy. One of the strategies proposed in recent years to provide a suitable model for predicting shear strength of reinforced concrete beams is the use of artificial intelligence (AI) algorithms. In this investigation, the application of adaptive neural fuzzy inference system (ANFIS) and support vector regression (SVR) algorithms for predicting shear strength of reinforced concrete beams was studied and the results were compared with existing regulation. For this purpose, shear span, effective beam length, effective depth, cross-section width, 28-day compressive strength of concrete, yield stress of longitudinal reinforcements, yield stress of transverse reinforcements, percentage of longitudinal reinforcement and shear reinforcement percentage were selected as input parameters and shear strength of concrete beam as output. Using the k fold validation method, educational and test data were defined and based on these data, predictions were made. The results obtained from the prediction show that the mean square root error (RMSE) for ANFIS and SVR methods is 0.1514 and 0.0994, respectively. In general, it can be seen that both ANFIS and SVR algorithms predict the shear strength of reinforced concrete beams with great accuracy. Therefore, they can be a good alternative to time-consuming algorithms such as ANN and expensive laboratory methods.