INTERNATIONAL JOURNAL OF ROBUST AND NONLINEAR CONTROL Int. J. Robust Nonlinear Control 2007; 17:1651–1667 Published online 9 May 2007 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/rnc.1214 Decentralized MPC of nonlinear systems: An input-to-state stability approach D. M. Raimondo 1 , L. Magni 1, * ,y and R. Scattolini 2 1 Dipartimento di Informatica e Sistemistica, Universita’ di Pavia, via Ferrata 1, Pavia 27100, Italy 2 Dipartimento di Elettronica e Informazione, Politecnico di Milano, Piazza Leonardo da Vinci 32, Milano 20133, Italy SUMMARY This paper presents stabilizing decentralized model predictive control (MPC) algorithms for discrete-time nonlinear systems. The overall system under control is composed by a number of subsystems, each one locally controlled with an MPC algorithm guaranteeing the input-to-state stability (ISS) property. Then, the main stability result is derived by considering the effect of interconnections as perturbation terms and by showing that also the overall system is ISS. Both open-loop and closed-loop min–max formulations of robust MPC are considered. Copyright # 2007 John Wiley & Sons, Ltd. Received 22 August 2006; Revised 21 March 2007; Accepted 21 March 2007 KEY WORDS: nonlinear model predictive control; input-to-state stability; decentralized control 1. INTRODUCTION Decentralized model predictive control (MPC) techniques are of paramount interest in the process industry; in fact a decentralized control structure is often the most appropriate one due to topological constraints and limited exchange of information between subsystems, while the MPC approach allows one to include in the problem formulation both performance requirements and state and control constraints. Moreover, a decentralized implementation of MPC often has the advantage to reduce an original, large size, optimization problem into a number of smaller and easily tractable ones. For these reasons, decentralized MPC has already been studied for discrete-time linear systems in, e.g. [1, 2] and in a number of papers quoted there. Recently, in [3] a decentralized MPC algorithm for nonlinear systems has been proposed, *Correspondence to: L. Magni, Dipartimento di Informatica e Sistemistica, Universita’ di Pavia, via Ferrata 1, Pavia 27100, Italy. y E-mail: lalo.magni@unipv.it Contract/grant sponsor: MIUR projects Advanced Methodologies for Control of Hybrid Systems Copyright # 2007 John Wiley & Sons, Ltd.