Application of SVM, ANN, GRNN, RF, GP and RT models for predicting discharge coefficients of oblique sluice gates using experimental data
Gates are commonly used to adjust water flow in open channels. By using an oblique/inclined gate, the water transferring capacity of open irrigation canals can be increased. Investigation of free and submerged discharge coefficients for inclined sluice gates is the focus of the present study. First...
Guardado en:
| Autores principales: | , , , |
|---|---|
| Formato: | article |
| Lenguaje: | EN |
| Publicado: |
IWA Publishing
2021
|
| Materias: | |
| Acceso en línea: | https://doaj.org/article/4b50368146fd4c0a80f5f5414644b215 |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| Sumario: | Gates are commonly used to adjust water flow in open channels. By using an oblique/inclined gate, the water transferring capacity of open irrigation canals can be increased. Investigation of free and submerged discharge coefficients for inclined sluice gates is the focus of the present study. First an experimental apparatus incorporating an inclined gate was created. The inclined angle (β) and gate opening (a) were experiment variables, and the five inclination angles include: 0° (vertical gate), 15°, 30°, 45° and 60°. Experimental results showed a greater convergence of flow lines under the gate and increasing the gate angle causes the discharge coefficient to increase. Also experiments showed that increasing the submergence rate (yt/a), decreases the inclined gate discharge coefficient. Performance metrics were created for the experimental results. The metrics utilized Gaussian process (GP) regression, support vector machine (SVM), artificial neural networks (ANN), generalized regression neural network (GRNN), random forest (RF) regression and random tree (RT) based models which were used to predict discharge coefficients (Cd) in both submerged and free flow conditions. The model input parameters were the ratio of the upstream water depth to gate opening (y/a) and the inclined angle (β) for free flow and also the submergence rate (yt/a) for submerged flow. The prediction models show that the ANN model in free flow conditions has the following performance metrics: Coefficient of determination, R2 = 0.9957, Root Mean Square Error (RMSE) = 0.0044, and Mean Absolute Error (MAE) = 0.0017. The performance metrics for submerged flow conditions were R2 = 0.9922, RMSE = 0.0079 and MAE = 0.0054. The ANN approach is the most accurate model compared to the others. |
|---|