چكيده به لاتين
Due to the ever-increasing growth of the population, the need for water increases, the need to build dams and then overflows that have the ability to pass large flows is felt more and more. Potential problems such as Cavitation on chutes in different discharges, especially under flood conditions, have raised the study of flow characteristics on these structures as an unavoidable necessity. One of the solutions to prevent the phenomenon of cavitation is the use of aerators in the direction of Chutes. To design an aerator, the estimation of three parameters aeration coefficient (b), jet length (〖L/h〗_0) and jet impact angle on chute (tang) is essential. So far, to estimate the amount of air entering the flow, different experimental relationships have been presented, which were obtained based on laboratory studies in specific geometric and hydraulic conditions. In this study the applicability of Data-Driven Methods to estimate the parameters required to design an aerator was investigated. The data-driven methods used in this study are group method of data handling (GMDH), support vector machine regression (SVR), Decision tree (DT) and Adaptive Neuro-Fuzzy Interface System (ANFIS). Furthermore The ANFIS model combined with four metaheuristic algorithms, including Differential Evolution (DE), Ant Colony Optimization (ACO), Genetic Algorithm (GA), and Particle Swarm Optimization (PSO), to determine the ANFIS parameters. To do so, employed experimental data which resulted existing empirical relations were used. In Aerated flow estimation used data from chute aerator model at the Laboratory of Hydraulics, Hydrology and Glaciology _VAW_ of ETH Zurich. In this study, three statistical indicators, such as Root Mean Square Error (RMSE), Mean Average Error (MAE), and coefficient of determination (R2), were used to compare the proposed methods with empirical methods. According to statistical indicators, it concludes that data-driven methods are better predicted than empirical methods, So that, compared to the empirical methods, the ANFIS-DE method has the best prediction in estimating b (RMSE = 0.018), 〖L/h〗_0 (RMSE = 1.293) and the tang (RMSE = 0.009).
