Algorithms for industrial model predictive control

This article is concerned with control methods that have been embedded in a n industrial model predictive control software package and that have been a pplied to a wide variety of industrial processes. Three methods are describ ed and the various features are evaluated by considering a constrained mult ivariable simulation. One method has been in use since 1988 and is widely e xploited in industry. The latest methods employ quadratic programming, whic h has become realistic to employ because of the advances in computing. The relative attributes are contrasted by assessing the ability of the controll ers to recover effectively from the impact of a large unmeasured disturbanc e.