Determination of Labyrinth and Arced Labyrinth Weirs Discharge Coefficients Using Support Vector Regression

Labyrinth weirs are widely used for controlling and regulating flow rate in rivers and canals as well as for conveying flow from upstream to downstream of dams. In this study, the discharge coefficient of the different kinds of labyrinth weirs was predicted by using support vector regression (SVR) technique. Totally 527 laboratory test data were used for predicting discharge coefficient of four different types of weirs including: normal and inverted labyrinth weirs in flume and arced labyrinth weirs with and without the nappe breakers in reservoir. The root mean squared error (RMSE), coefficient of determination (E or DC), and squared correlation coefficient (R2) statistics were used to evaluate the models’ performance. The obtained results showed that the support vector regression technique had a high capability of predicting discharge coefficient of labyrinth weirs .Statistical evaluation criteria of the best model for normal labyrinth weirs were R2=0.990, DC=0.988, RMSE=0.0077 in validation stage. The results of sensitivity analysis showed that (Fr, HT/p) and (HT/p, α/θ) were the most influential parameters for the labyrinth and arced labyrinth weirs, respectively.