Model Predictive Control: Algorithmic Development and Applications

Model Predictive Control (MPC) is one of the few areas that has received on-going interest from researchers in both the industrial and academic communities. The key aspects of model predictive control that make the design methodology effective in engineering applications are its ability to handle both ‘soft’ and hard constraints in a multivariable control framework, and to perform process on-line optimization. This special issue brings together a group of experts in the field of model predictive control to present their recent results of algorithmic development and applications. It begins with the investigation of MPC theory and algorithms to support new applications for stochastic systems. Professor Braatz and Dr Kim propose a new approach to stochastic MPC problems in the presence of both parametric model uncertainty and exogenous disturbance. In their paper, the effects of uncertainties are quantified using generalized polynomial chaos expansions with an additive Gaussian random process as the exogenous disturbance, and it provides an explicit way to handle a stochastic system with parametric uncertainty and disturbance. The second paper in stochastic MPC is written by Professor Goodwin and Dr Medioli. In their paper, scenario based closed-loop model predictive control is investigated for the application to emergency vehicle scheduling. The next three papers in the special issue propose numerical solutions to the constrained control problem that exists in all model predictive control algorithms. Professor Boyd and his colleagues investigate fast and reliable solutions to nonconvex model predictive control, where a sequential convex optimisation method is used in the core algorithm that typically converges in fewer than five or so iterations and is more than fast enough to run in real time for many applications. Their algorithm has been demonstrated in model predictive control of a commercial multi-zone refrigeration system for optimization of energy consumption. To further address numerical issues in the solution of MPC problems, Drs Cairano, Brand and Bortoff propose projection-free parallel quadratic programming for linear model predictive control. In their paper, the original parallel quadratic programming algorithm proposed by Brand is analysed for its convergence and an acceleration technique based on a projection-free line search is used to speed-up the convergence to the optimum. Their parallel quadratic programming algorithm is simple in implementation and the new acceleration technique to convergence will see its wide applications in model predictive control. Professor Morari and Dr Almer propose an efficient online solution to multi-parametric mixed integer quadratic problems arising from model predictive control of power electronics. The results presented in their paper are a generalization and simplification of an algorithm previously proposed by the authors. Drs Rossiter and Khan investigate a more efficient approach to model predictive control from the angle of parameterisation of control trajectory using orthonormal basis functions. Professors Manzie, Good and Dr Lam propose a new approach to multi-axis model predictive contouring control with discussions on the implementation of their algorithm for industrial machine control. The final two papers of this special issue discuss the combination of classical adaptive control techniques with model predictive control for controlling nonlinear systems. Professor Guay and Dr Adelola investigate the design of economic MPC systems using adaptive control techniques for a class of nonlinear systems subject to parametric uncertainties and exogenous variables. Their approach is applicable to both dynamic and steady-state economic optimization. Professor Wang and Mr Lu present the design and implementation of a gain scheduled model predictive controller. Their paper is written in a tutorial style to show the detailed design and implementation steps. Their results are verified using the experimental results obtained from controlling an industrial sized induction motor. In conclusion, this special issue has presented recent developments of algorithms and applications in model predictive control with contributing authors from both academic and industrial communities. We hope, as guest editors for this special issue, that the papers are interesting to researchers, students and engineers in the control engineering community at large.