Gholami, A., Bonakdari, H., Akhtari, A., Ebtehaj, I. (2019). A COMBINATION OF COMPUTATIONAL FLUID DYNAMICS, ARTIFICIAL NEURAL NETWORK AND SUPPORT VECTORS MACHINES MODEL TO PREDICT FLOW VARIABLES IN CURVED CHANNEL. Scientia Iranica, 26(2), 726-741. doi: 10.24200/sci.2017.4520
Azadeh Gholami; Hossein Bonakdari; Ali Akbar Akhtari; Isa Ebtehaj. "A COMBINATION OF COMPUTATIONAL FLUID DYNAMICS, ARTIFICIAL NEURAL NETWORK AND SUPPORT VECTORS MACHINES MODEL TO PREDICT FLOW VARIABLES IN CURVED CHANNEL". Scientia Iranica, 26, 2, 2019, 726-741. doi: 10.24200/sci.2017.4520
Gholami, A., Bonakdari, H., Akhtari, A., Ebtehaj, I. (2019). 'A COMBINATION OF COMPUTATIONAL FLUID DYNAMICS, ARTIFICIAL NEURAL NETWORK AND SUPPORT VECTORS MACHINES MODEL TO PREDICT FLOW VARIABLES IN CURVED CHANNEL', Scientia Iranica, 26(2), pp. 726-741. doi: 10.24200/sci.2017.4520
Gholami, A., Bonakdari, H., Akhtari, A., Ebtehaj, I. A COMBINATION OF COMPUTATIONAL FLUID DYNAMICS, ARTIFICIAL NEURAL NETWORK AND SUPPORT VECTORS MACHINES MODEL TO PREDICT FLOW VARIABLES IN CURVED CHANNEL. Scientia Iranica, 2019; 26(2): 726-741. doi: 10.24200/sci.2017.4520
A COMBINATION OF COMPUTATIONAL FLUID DYNAMICS, ARTIFICIAL NEURAL NETWORK AND SUPPORT VECTORS MACHINES MODEL TO PREDICT FLOW VARIABLES IN CURVED CHANNEL
Department of Civil Engineering, Razi University, Kermanshah, Iran
Abstract
This study show the combination of computational fluid dynamics (CFD) and soft computing techniques to make viewpoint for two-phase flow modelling and accuracy evaluation of soft computing methods in the three-dimensional flow variables prediction in curved channels. Therefore, artificial neural network (ANN) and support vectors machines (SVM) models with CFD is designed to estimate velocity and flow depth variable in 60° sharp bend. Experimental results in 6 different flow discharges of 5, 7.8, 13.6, 19.1, 25.3 and 30.8 l/s to train and test, ANN and SVM models is used. The results of numerical models with experimental values are compared and the models accuracy is confirmed. The results evaluation show that all three models ANN, SVM and CFD perform well in flow velocity prediction, with correlation coefficient (R) of 0.952, o.806, and 0.680, and flow depth (R) of 0.999, 0.696, and 0.614 respectively. ANN model to predict both velocity and flow depth variables with mean absolute relative error (MARE) of 0.055 and 0.004 is the best model. Then SVM and CFD models with MARE of 0.069 and 0.089 in velocity prediction and in flow depth prediction CFD and SVM models with MARE of 0.007 and 0.011 are the best models, respectively.
Top