Weirs change the flow velocity in the open channels in different ways. This change in velocity can cause a change in turbulence.
Therefore, by changing the shape of the weirs, it is possible to change the flow pattern in the weirs. For this purpose,
a suitable weir called vertically cosine shape weir was introduced for the mentioned purposes. Four optimization metaheuristic
methods including a genetic algorithm (GA), imperialist competitive algorithm (ICA), election algorithm (EA),
and gray wolf algorithm (GWO), based on the support vector regression method (SVR), were used to estimate the discharge
coefficient (Cd) of vertically cosine shape weirs. To evaluate the performance of these models, four statistics including correlation
coefficient (R2), root means square error (RMSE), mean absolute percentage error (MAPE), and standard deviation
(δ), were used. To test and validate the proposed models, the experimental dataset of Salehi et al. was utilized. Four different
combined input parameters were selected as the most effective input parameters in predicting the Cd. The results showed the
high accuracy of the GWO-SVR model compared to the other models with values R2 = 0.93, RMSE = 0.012, MAPE = 0.665,
and δ = 0.847. In general, a comparison of the results obtained from the models used in this study showed the high ability
and accuracy of the GWO-SVR model in estimating the Cd of vertically cosine shape weirs compared with EA-SVR, ICASVR,
and GA-SVR models.