The broad-crested weir is convenient to construct and has a small amount of ex-cavation,widely used in practical engineering.Discharge computing has been the focus of research on this structure,thus developing general...The broad-crested weir is convenient to construct and has a small amount of ex-cavation,widely used in practical engineering.Discharge computing has been the focus of research on this structure,thus developing generalized regression neural network(GRNN),genetic programming(GP),and extreme learning machine(ELM)are used to predict the discharge coefficient(Cd)of the triangular broad-crested weir.The comprehensive analysis shows that the ELM model has high stability,predictive ability,and computational speed.The coefficient of determination(R^2)is 0.99982,the mean absolute percentage error(MAPE)is 0.000261,the Nash-Sutcliffe coefficient(NSE)is 0.99977,and the root means square error(RMSE)is 4.17E-05 in the testing phase.The apex angleθis the most critical parameter affecting the Cd,and the contribution to the Cd is 52.45%.A new computational formula is proposed and compared with the accuracy of empirical formulas,showing that the intelligent method has higher accuracy and efficiency.展开更多
The flow over broad-crested weirs was simulated by computational fluid dynamic model. The water surface profile over broad crested weir was measured in a laboratory model and validated using two and three dimensional ...The flow over broad-crested weirs was simulated by computational fluid dynamic model. The water surface profile over broad crested weir was measured in a laboratory model and validated using two and three dimensional Fluent programs. The Reynolds Averaged Navier-Stokes equations coupled with the turbulent standard （k-ε） model and volume of fluid method were applied to estimate the water surface profile. The results of numerical model were compared with experimental results to evaluate the ability of model in describing the behaviour of water surface profile over the weir. The results indicated that the 3D required more time in comparison with 2D results and the flow over weir changed from subcritical flow at the upstream （U/S） face of weir to critical flow over the crest and to supercritical flow at downstream （D/S）. A reasonable agreement was noticed between numerical results and experimental observations with mean error less than 2 %.展开更多
基金Xi'an University of Technology Excellent Master Seed Fund,Grant/Award Number:310/252082213。
文摘The broad-crested weir is convenient to construct and has a small amount of ex-cavation,widely used in practical engineering.Discharge computing has been the focus of research on this structure,thus developing generalized regression neural network(GRNN),genetic programming(GP),and extreme learning machine(ELM)are used to predict the discharge coefficient(Cd)of the triangular broad-crested weir.The comprehensive analysis shows that the ELM model has high stability,predictive ability,and computational speed.The coefficient of determination(R^2)is 0.99982,the mean absolute percentage error(MAPE)is 0.000261,the Nash-Sutcliffe coefficient(NSE)is 0.99977,and the root means square error(RMSE)is 4.17E-05 in the testing phase.The apex angleθis the most critical parameter affecting the Cd,and the contribution to the Cd is 52.45%.A new computational formula is proposed and compared with the accuracy of empirical formulas,showing that the intelligent method has higher accuracy and efficiency.
文摘The flow over broad-crested weirs was simulated by computational fluid dynamic model. The water surface profile over broad crested weir was measured in a laboratory model and validated using two and three dimensional Fluent programs. The Reynolds Averaged Navier-Stokes equations coupled with the turbulent standard （k-ε） model and volume of fluid method were applied to estimate the water surface profile. The results of numerical model were compared with experimental results to evaluate the ability of model in describing the behaviour of water surface profile over the weir. The results indicated that the 3D required more time in comparison with 2D results and the flow over weir changed from subcritical flow at the upstream （U/S） face of weir to critical flow over the crest and to supercritical flow at downstream （D/S）. A reasonable agreement was noticed between numerical results and experimental observations with mean error less than 2 %.
