11Generalization of Predictive Control Principles

11.1 Interpretation of Methodological Principles of Predictive Control

The application of predictive control was originally only in industrial processes, but today it is extended to many application fields. In addition to the technological reason that it might be the only control technique capable of explicitly incorporating constraints into the optimization problem and effectively solving it, more important is the universality of its methodological principles. As Richalet et al. early pointed out in [1], the basic ideas underlying this approach are related to the “scenario technique,” which to some extent is similar to what the human operator is assumed to do with his internal model of the external world. The powerful methodological principles implied in predictive control should naturally have wider applicability.

In Section 1.2, the basic principles of predictive control, i.e. prediction model, rolling optimization, and feedback correction, have been explained in detail, which concretely embody the concepts of model, optimization, and feedback in cybernetics, respectively. For the prediction model, since the most emphasized issue is its function rather than structure, it can be flexibly established using known information in a most reasonable way according to the plant characteristics and the control requirements. Due to the use of error prediction of unmodeled information and the nonclassical rolling style optimization, ...

Get Predictive Control now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.