A modeling approach for iron concentration in sand filtration effluent using adaptive neuro-fuzzy model

Effluent iron concentration is an important water quality criterion used for the assessment of the performance of rapid sand filters, in addition to other criteria. This study deals with the prediction of effluent iron concentrations by adaptive neuro-fuzzy (ANFIS) model with input parameters including filter hydraulic loading rate, influent iron concentration, bed porosity and operation time. With trying various types of membership functions, two rule base generation methods, namely subtractive clustering and grid partition were used for a first order Sugeno type inference system. Models were evaluated using root mean squared error (RMSE), index of agreement (IA) and R2 as statistical performance parameters. The fit between experimental results and model outputs showed good agreement for tap water and deionized water; testing RMSE values were 36.33 and 7.66 μg/L, the IA values were 0.996 and 0.971, and R2 values were 0.99 and 0.89, respectively. It was concluded that neuro-fuzzy modeling may be successfully used to predict effluent iron concentration in sand filtration.