Investigation of effective Hydraulic Parameters on Scouring at downstream of Control Structures Using Gaussian Process Regression Method

Scour depth prediction downstream of control structures such as spillways and grade-control structures is one of the essential issues in the design of these structures. In the present study, the efficiency of the Gaussian Process Regression (GPR) method was evaluated to estimate the scour depth downstream of control structures. Various models were developed and the impacts of hydraulic parameters were evaluated. The results proved the high efficiency of the applied method in the research in estimating the scour depth compared to the semi-empirical equations. The best result for test series was obtained in the state of sharp crested weir with the values of CC=0.93, DC=0.90, RMSE=0.112 and MAE=0.079and in the inclined slope controlled with the values of CC=0.86, DC=0.81, RMSE=0.194 and MAE=.167. As well as, the sensitivity analysis results show that Fr1 and b/z parameters and Fr1 and Dp/h is the most important parameter affecting the scour depth simulation downstream of sharp crested weir and inclined slope controlled respectively. In fact, variations in the scour depth are influenced by geometry and flow energy in the upstream of these structures. The comparison of semi-empirical and classical relations with the results of the applied method in the research showed that the results of the intelligent GPR method are more accurate and reliable than the classical relations. So that the root mean square error decreased more than 80%.