Application of Gaussian Process Regression Model to Predict Discharge Coefficient of Gated Piano Key Weir

The Piano Key (PK) weir is a new type of long crested weirs. This study was involved the addition of a gate to PK weir inlet keys. It was conducted by the Department of Water Engineering, University of Tabriz, Iran to determine if the gate increased hydraulic performance. A Gated Piano Key (GPK) weir was constructed and tested for discharge ranges of between 10 and 130 l per second. To this end, 156 experimental tests were performed and the effective parameters on the GPK weir discharge coefficient (C-d), such as gate dimensions (b and d), gate insertion depth in the inlet key (H-gate), the ratio of the inlet key width to the outlet key width (W-i/W-o) and the head over the GPK weir crest (H) were investigated. In addition, application of soft computing to estimate of C-d was carried out using MLP, GPR, SVM, GRNN, multiple linear and non-linear regressions methods using MATLAB 2018 software. This study suggests the relation for C-d with non-dimension parameters. The results of this study showed that H, W-i/W-o, H-gate and b and d, had the greatest effect on the GPK weir discharge coefficient, respectively. The GPR method was introduced as a new effective method for predicting discharge coefficient of weirs with RMSE = 0.011, R-2 = 0.992 and MAPE = 1.167% and provided the best results when compared with other methods.