Improving the performance of learning machines in estimating and predicting Discharge coefficient

Piano key weirs are flow control structure that have more discharge coefficient than classic weir. In the present, study the support vector machine, hybrid of support vector machine and bat algorithm (SVR-BA) and M5 algorithm are used for predicting the discharge coefficient. Overall, 162 expremental data for seven pianokey weir model are extracted from an expremental study. Also, the discharge coefficient is prediction by employing the parameters included ratio of uppstream water hed to high of weir, width of inlet key, width of outlet key, high of weir, shape factor of fillet and shape factor of crest as input data and discharge coefficient (Cd) as output model. The results based assessment cretria shows that all tree used intelegent model can predicted the discharge coefficient of piano key weir. Nevertheless, in test period SVR-BA model has more accurate with value of 0.992, 0.007 and 0.01 respectively for R, MAE and RMSE.