Evaluation of combined Models with Optimization Approach of PSO and GA in ANFIS for Predicting of Dispersion Coefficient in Rivers

Recently, water pollutions in rivers and canals have become the main issue for researchers. In addition, water pollutants have different effects on human and aquatic health. So, the prediction of pollution in water in different water resources like rivers has been the main topic for researching. The longitude dispersion coefficient which is experimental and theoretical method that is the best way for describing longitude dispersion. In this study, a new method has been used for predicting the longitude dispersion by ANFIS developing with PSO and GA optimization. For this purpose, the programs run with 116 normalizing data by writing of code in MATLAB software. The river wide, water depth, velocity and Shear velocity were used for input parameter and Dispersion coefficient was used for the porpuse parameter. Results showed that the ANFIS-PSO model predicts dispersion coefficient with MSE=0.0037, RMSE=0.061 and R=0.9622 and ANFIS-GA model predicts dispersion coefficient with MSE=0.012, RMSE=0.11 and R=0.739 that have better accurate than ANFIS with MSE=0.040m, RMSE=0.200 and R=0.698. By evaluating the two models, it was found that the PSO algorithm has better performance than GA algorithm in ANFIS model. The ANFIS-PSO model was the most accurate among the three studied models. Finally, it was concluded that the ANFIS-PSO model is more appropriate model to estimate in RMSE, MSE and R for dispersion coefficient