Economic model predictive control for robust optimal operation of sparse storage networks

Abstract

A tube-based robust economic model predictive control scheme is proposed for sparse, linear storage network systems under bounded disturbances. We present a modeling framework that reflects conservation principles, and for which a centralized controller optimizes a performance metric for the entire network over a finite time horizon. Our main result is a computationally efficient projection-based approach that exploits the sparse structure of the system dynamics for tightening state and input constraints, in the meantime ensuring that average performance, recursive feasibility, and asymptotic stability of the system are maintained. An inventory control problem under demand uncertainty is presented to illustrate the advantages of the proposed framework and benchmark against nominal economic model predictive control.