Performance of landfill fresh leachate pretreatment by circulation on primitive waste deposits predicted by artificial neural network (ANN) mathematical
modeling. Landfill simulation reactor (LSR) for laboratory scale modeling was applied to prepare the old waste and evaluation of pretreatment performance. Concentration of BOD5 and COD in leachate measured as organic compound indicators and NH3 and NH4 concentrations as nitrogen compound. The results indicated that organic matters removal (64%) was higher compare to nitrogen compounds removal (42%). Optimized neural network was constructed with a mean absolute error (MSE) of 30.85 and 6-7-4 structure. The input of model was leachate characteristics and time, while the output removal performance for organic and nitrogen compounds. The R value of 0.979 from regression analyses indicated there was very good agreement in the trends between forecasted and measured data.