Oxygen aeration efficiency of gabion spillway by soft computing models

Abstract The current paper deals with the performance evaluation of application three soft computing algorithms such as adaptive neuro-fuzzy inference system (ANFIS), backpropagation neural network (BPNN), and deep (DNN) in predicting oxygen aeration efficiency (OAE20) gabion spillways. Besides, classical equations, namely multivariate linear nonlinear regressions (MVLR MVNLR), including previous studies, were also employed OAE20 analysis results showed that DNN demonstrated relatively lower error values (root mean square error, RMSE = 0.03465; MSE 0.00121; absolute MAE 0.02721) highest value correlation coefficient, CC 0.9757, performed best spillways; however, other applied models, ANFIS, BPNN, MVLR, MVNLR, giving comparable evaluated to statistical appraisal metrics relative significance input parameters based on sensitivity investigation, porosity (n) materials was observed be most critical parameter, height (P) had least impact over