• ottimizzazione
• nonlinear control
• controllo non lineare
• optimization
• Kalman filter
• filtro di Kalman
• predictive control
• controllo predittivo

Offset-free model predictive control refers to a class of control algorithms able to track asymptotically constant reference signals despite the presence of unmeasured, nonzero mean disturbances acting on the process and/or plant model mismatch. Generally, in these formulations the nominal model of the plant is augmented with integrating disturbances, i.e. with a properly designed disturbance model, and state and disturbance are estimated from output measurements. To date the vast majority of offset-free MPC applications are based on linear models, however, since process dynamics are generally inherently nonlinear, these may perform poorly or even fail in some situations. Better results can be achieved by making use of nonlinear formulations and hence of nonlinear model predictive control (NMPC) technology. However, the obstacles associated with implementing NMPC frameworks are nontrivial. In this work the offset-free tracking problem with nonlinear models is addressed. Firstly some basic concepts related to the observability of nonlinear systems and state estimation are reviewed, focusing on the digital filtering and putting a strong accent on the role of the disturbance model. Thus, a class of disturbance models in which the integrated term is added to model parameters is presented together with an efficient and practical strategy for its design and subsequent implementation in offset-free NMPC frameworks.