Support Vector Machine to predict the discharge coefficient of Sharp crested w-planform weirs

In this paper, the discharge coefficient (Cd) of triangular labyrinth weir was predicted using Multilayer Perceptron Neural Network (MLPNN), Radial Basis Neural Network (RBFNN) and support vector machine (SVM). To this end, 223 data sets related to the effective parameters on Cd were collected. Using dimensional analysis techniques, the involved dimensionless parameters on Cd were derived. To find out the most effective parameters on Cd, the Gamma test (GT) was analyzed. Results of GT demonstrated that H/P, Lw/Lc, and Lw/Wm are the most effective parameters on Cd. To develop ANN and SVM, different types of transfer and kernel functions were tested. During the testing of transfer and kernel functions for developing the ANN and SVM models, respectively, it was found that tensing and RBFNN have the best performance for predicting the Cd. Overall evaluation of the results of developed models indicated that both models have a suitable accuracy in predicting the Cd; however, the SVM is a bit more accurate. Comparing the outcomes of the applied models in terms of DDR index shows that the data dispersivity of SVM is less than the others; therefore, this model is more reliable.