An activated-sludge-process application of integrated design and predictive control with instantaneous linearization

摘要

In this work the integrated design of the activated sludge process is addressed. The objective is to determine the best plant parameters and working point that minimize the operation costs related to the Effluent Quality and the Aeration Energy, while imposing constraints on the plant condition number (γ) and the perturbation condition number (γp), ensuring open-loop controllability. A linear multivariable predictive control (MPC) is used for the closed loop design, and it is used also a very practical non-linear version of the MPC, based on the instantaneous linearization of non linear models of the plant (the phenomenological model as well as a neural network model obtained by identification). The results are analyzed based on two aspects: the improved performance of the controlled system when using the integrated design instead of a classical economic design, and the convenience of the instantaneous linearization to realize a non-linear MPC.