Prediction and parameter quantitative analysis of side orifice discharge coefficient based on machine learning

Abstract In this paper, the sparrow search algorithm is used to predict discharge coefficient (Cd) of triangular side orifice for first time. Dimensionless parameters influencing size Cd orifices are obtained as input values and output model. The results show that determination R2 0.973, root means square error RMSE 0.0122, average absolute percentage 0.010% in testing phase. model has high forecast accuracy, strong generalization ability higher accuracy than other models traditional empirical formulas. Quantitative analysis by Sobol's method shows ratio W/H top height height, Fr upstream Froude number, B/L channel width a bottom edge length main factors capacity orifice. first-order sensitivity global 0.23, 0.11, 0.17 0.41, 0.39, 0.35 respectively.