Plug-and-play predictive control of modular nonlinear systems with coupling delays

This paper presents a decentralized model predictive control strategy for nonlinear network systems having multiple coupling delay elements. Linear interconnections, whose topologies are defined by an unweighted Laplacian matrix, and persistent input disturbances are considered in this development. On the ground of the incrementally accumulative quadratic constraint presented previously, the time-domain robust stabilizability conditions for nonlinear interconnected systems are developed in this work. To deal with multiple coupling delays, the accumulative dissipativity criteria for subsystems are derived for use with the accumulatively asymptotic dissipativity constraint (aADC) in the stabilizability conditions. aADC is an asymptotic constraint on the accumulation of supply rates over the window of delay time interval. A compound output in the supply rate is deployed such that the dissipativity criterion can be rendered in linear matrix inequalities. A convex stability constraint for the local optimization of the model predictive control in a fully decentralized architecture is then derived for implementations. Numerical simulation for a petrochemical process is provided to illustrate the theoretical developments.