Survey of industrial applications of embedded model predictive control

Abstract

Summary form only given. The past two decades have witnessed enormous advances in solving model predictive control (MPC) problems on embedded controller hardware. Besides featuring limited computational resources, the main challenge in embedded MPC applications is to make the underlying numerical optimization run highly reliably without any user interaction. Today, theory, algorithms and implementations have reached a level of maturity that enables the use of embedded MPC technology in commercial products. This survey provides an overview of successful industrial applications of embedded MPC from various application domains. A couple of examples are discussed in detail in order to illustrate the use of different problem formulations and the associated solution approaches. The description of industrial applications is complemented by a concise overview of engineering tools for embedded MPC. Finally, we discuss challenges in embedded MPC that currently limit a wider spread of the technology in industry.