Document Type : Original Article
Department of Construction and Construction Materials, М. Auezov South Kazakhstan State University, Shymkent, Kazakhstan.
Scouring is caused by riverbed erosion by water flows and water-borne materials. In the present work, the depth of scouring is estimated using the autonomous neural network (SOM). The results have been compared to the results of other models. Regarding the obtained results, the auto structure of the neural network (SOM) was found to have a higher coefficient of correlation (0.98) than other processes. The mean squared error was also lower than other methods (RMSE = 0.112). Estimated scavenger depth using the SOM approach revealed that this method provides good results. So that the correlation coefficient in the program’s execution is with subsequent data compared to the non-secondary data in the program. Moreover, the root mean square mistake (RSME = 0.09) was unclear in the subsequent data mode. In this study, sensitivity analysis has shown that the SOM program will be more sensitive to the average particle diameter when executed with subsequent data.