Hybrid Dynamic Model of Anoxic-Aeration Biological Wastewater Treatment Plant

Abstract

Due to high complex of the biological wastewater treatment, it is difficult to develop an accurate mathematics model. A hybrid dynamic modeling in both parallel and serial configuration was used in an anoxic-aeration activated sludge process. In the hybrid model, case-based reasoning (CBR) system is placed in series with a mechanistic model, which was the Activated Sludge Model No. 1 (ASM1) by the International Water Association (IWA); whereas the CBR system was used to identify the key kinetic and stoichiometric parameter at the different conditions. A neural network in parallel configuration compensates for the known difference between the results of a purely mechanistic model and the process data. The proposed method was applied in the shenyang wastewater treatment plant. Simulation results demonstrate the predictions of the hybrid model were significantly improved.