Investigating the Effect of Parameters on Confinement Coefficient of Reinforced Concrete Using Development of Learning Machine Models

The current research aims to investigate the parameters' effect on the confinement coefficient, K-s, forecast using machine learning. Because various parameters affect the K-s, a new computational model has been developed to investigate this issue. Six parameters are among the effective parameters based on previous research. Therefore, according to the dimensions of the variables in the problem, a supply-demand-based optimization (SDO) model was developed. The performance of this model is directly dependent on its main parameters, such as market size and iteration. Then, to compare the performance of the SDO model, classical models, including particle swarm size (PSO), imperialism competitive algorithm (ICA), and genetic algorithm (GA), were used. Finally, the best-developed model used different parameters to check the uncertainty obtained. For the test results, the new SDO-ANFIS model was able to obtain values of 0.9449 and 0.134 for the coefficient of determination (R2), and root mean square error (RMSE), which performed better than other models. Due to the different relationships between the parameters, different designed conditions were considered and developed based on the hybrid model and, finally, the number of longitudinal bars and diameter of lateral ties were obtained as the strongest and weakest parameters based on the developed model for this study.

Automated summary

The current research aims to investigate the effect of parameters on the confinement coefficient, Ks, using machine learning and develop a new computational model to address this issue. Based on previous research, six effective parameters were identified and a supply-demand-based optimization model was developed. The performance of the model depends on main parameters such as market size and iteration. Comparisons with classical models showed that the new SDO-ANFIS model outperformed others. Different relationships between parameters were considered to identify the strongest and weakest parameters for this study.