Reduced Order Models for On-Line Parameter Identification of the Activated Sludge Process

Abstract

A reduced order dynamic model for an activated sludge process performing carbonaceous removal, nitrification, and denitrification is proposed. Based on directly measurable real time data - by methods available today - all model parameters are identified on-line. A simplified extended Kalman filter is used for the actual identification. Verification of the results is based on computer simulations of the IAWQ Activated Sludge Model No. 1. The reduced order model presented herein may serve as a tool for predicting the dynamic behaviour of a biological wastewater treatment plant since the parameters under varying operating conditions can be tracked on-line. The model parameters are effectively estimated even when the measurements are affected by a significant level of noise. The model is aimed for operation and control purposes as an integral part of a hierarchical control structure.