Wednesday 1 july 2015
12:36 - 12:39h at Oceania (level 0)
Themes: (T) Sediment management and morphodynamics, (ST) Sediment transport mechanisms and modelling, Poster pitches
Parallel session: Poster pitches: 8A. Sediment - Erosion
Correct prediction of sediment transport to prevent of deposition in sewer pipes is very important for designing of sewer network. The application of two different data-driven approaches, gene expression programming (GEP), which is an extension of genetic programming (GP) and group method of data handling (GMDH) for bed load sediment transport estimation, is compared in this paper. Also GEP and GMDH were compared with the other existing sediment transport equations, which were obtained using nonlinear regression analysis. Using non-dimensional parameters affecting sediment transport at sewer pipe, different models to determine minimum velocity have been provided. The non-deposition sediment transport data for bed load sediment transport from two different references are used. The root mean square error (RMSE), mean average percentage error (MAPE) and determine of coefficient (R2) are used for evaluating of the accuracy of the models. As the comparisons demonstrated, the GEP (RMSE = 0.14, MAPE = 2.82 R2 = 0.99) and GMDH (RMSE = 0.35, MAPE = 5.1 R2 = 0.95) models are more accurate than existing equations and could be successfully employed in forecasting minimum velocity. However, GEP is superior to GMDH in giving explicit expressions for the problem.