Prediction of Peak Velocity of Pollutant Transport in Rivers Using Group Method Data Handling (GMDH) and Intelligent Hybrid Method GMDH-HS

Estimating of the peak velocity of pollutant in a flow using hydraulic and geometrical parameters is very important in predicting the pollution transport in rivers. Suitable empirical equations are developed, none of which is reliable enough in estimating the peak velocity of pollutant before a thorough calibration. So, in this research artificial intelligence methods are used for this purpose. The objective of this research was to predict the peak velocity of pollutant by Group Method Data Handling (GMDH) methods and an intelligent hybrid method (GMDH-HS). The result of these methods were compared to the best regression equation. The dimensionless relative discharge (Q'a), dimensionless drainage area, the ratio of discharge at the section at the time of measurement to drainage area (Q/Da) and the reached slope (S) were taken as input parameters to these models. These data were collected from several different rivers in the United States. Mackey-Glass standard function was used to evaluate the performance of the GMDH and GMDH-HS models. The results indicated that the proposed models predicted the peak velocity of pollutant precisely (CE GMDH =0.9328, CE GMDH-HS =0.9038 & CE Vp Equation=0.3802) and these models are more accurate compared to the best nonlinear regression equation..