A new statistical framework for parameter subset selection and optimal parameter estimation in the activated sludge model

A new model-calibration method has been proposed to solve the problems associated with parameter subset selection and parameter estimation of the activated sludge model (ASM). We propose the use of a statistical methodology for reasonable parameter selection and parameter estimation that consists of sensitivity analysis, similarity measures, hierarchical clustering and response surface methods (RSM). The introduction of effluent quality index (EQI) can reduce all of the outputs of the ASM model into one factor. The EQI was used to calculate a sensitivity matrix. Then, the hierarchical clustering algorithm was used for parameter subset selection. This selection was based on a similarity measure using the sensitivity matrix and was used to reduce the number of model parameters by selecting only one parameter per cluster group (parameter subset selection step). Lastly, a RSM analysis was conducted in order to determine the optimal parameter values. This study was conducted in order to develop a new statistical framework that can greatly reduce the computational effort required to find the optimal solution by reducing the number of parameters. The experimental results indicated that the calibrated model can improve the prediction quality of the ASM model and the efficiency of the modeling.