Evaluation of hybrid metaheuristic models in simulation of dissolved oxygen in river water

Water is one of the most essential elements in nature that forms the basis of human life and contributes to the economic growth and development of societies. Healthy water is closely related to environmental health and activities. The life of all animals on Earth depends on water and oxygen. In addition, adequate dissolved oxygen (DO) is essential for the survival of aquatic animals. Therefore, in this study, to simulate the dissolved oxygen of the Cumberland River in the United States from the combined artificial neural network (ANN) model with wavelet and meta-heuristic algorithms of gray wolf (GWO) and bat (BA) on a monthly time scale during the statistical period. Used 2020-2010. The criteria of correlation coefficient (R2), squared mean square error (RMSE), absolute mean error (MAE) and Nash-Sutcliffe productivity coefficient (NSE) were used to evaluate and compare the performance of the models. The results showed that all three hybrid models have better results in hybrid models than the other designated models. Also, according to the evaluation criteria, it was found that among the models used in the simulation of dissolved oxygen in river water, the model of artificial neural network-wavelet with coefficient of determination (R2 = 0.958), the root mean square error (RMSE = 0.651), The mean absolute value of error (MAE = 0.334) and Nash Sutcliffe coefficient (NS = 0.962) in the validation stage showed better performance than other models.