Model predictive control with on-line model identification for anaerobic digestion processes

This paper presents a nonlinear model predictive control approach for the anaerobic digestion process. A new model reduction strategy with estimation of the model parameters is proposed for the anaerobic digestion process. The reduced model is then used to predict future plant states in the nonlinear model predictive control. We develop a terminal feasible set to constrain terminal states in the prediction horizon, such that the controlled process beyond the horizon lies within a stable region and the predictive controller is recursively feasible. In addition, to make the predictive controller more practical, we design a predictive control algorithm that explicitly considers the influence of process disturbances and satisfies given constraints. Numerical simulations on the benchmark model ADM1 demonstrate the performance of the proposed method. (C) 2017 Elsevier B.V. All rights reserved.