Enhancing accuracy of discharge capacity prediction of a sharp-crested curved plan-form weirs under free flow conditions using evolutionary algorithms and artificial neural networks

Weirs are one of the common structures for discharge and flow measurement. Therefore, these types of hydraulic structures depending on the purpose of usage, have different shapes. Weirs have been widely used for the purpose of flow measurement and flow control in open channels. Generally they are used as normal weirs. For the purpose of flow diversion, they can also be used as side weirs or skew weirs. Various weirs of modified plan form have been suggested in the past to enhance their discharging capacity with minimum head over the weirs and to restrict the afflux. The aim of this study is to apply different methods to investigate the discharging capacity of a sharp-crested curved plan-form weirs under free flow conditions using original experimental dataset through the Artificial Neural Networks (ANNs) and Genetic Expression Programming (GEP) techniques. Subsequently, for training and testing of the proposed equation, experimental data of Kumar et al. have been used. A preliminary investigation on various GEP operators is also carried out for selecting the proper operators. The obtained results indicate that applied machine learning techniques have reliable performance in predicting discharging capacity of a sharp-crested curved plan-form weirs. Comparison of results obtained from this equation with the experimental data reveals high accuracy of the new equation of genetic programming and result of the ANNs. Determination coefficient of the proposed equation for discharge coefficient have been calculated as 0.956 and 0.924 for the model with best functions (F2 and F4), also this parameter calculated for ANNs as 0.962 for testing phase.