Simulation and Activated Sludge Model-Based Iterative Learning Control for a Sequencing Batch Reactor

Abstract

In this paper, a state-space model based on an activated sludge model (ASM) was developed and an iterative learning control (ILC) algorithm for a sequencing batch reactor (SBR) are proposed for optimal nitrogen removal. A reduced-order state-space model of an ASM was derived by lumping some components of an ASM in the aerobic and anoxic phase, and then the ILC algorithm was suggested as a batch tracking control that worked in a repetitive mode of an SBR using the reduced state-space model. A new algorithm of ILC was suggested to control dissolved oxygen (DO) concentration in the aerobic phase and to find the optimal amount of external carbon, which was added to the anoxic phase in an SBR. By using information from previous batches, a suitable ILC control action was found iteratively. Simulations clearly represented that the ILC algorithm in the aerobic phase showed much better control performance of DO concentration than the conventional PID control, and could find a minimum amount of added external carbon for the anoxic phase in an SBR, which reduced the total cost by 35.6%. Results showed that the ILC model was derived from the microbiological concept of an ASM, and the learning ability of ILC could reduce the error of the controller.