Aspects in Selecting Predictive Adaptive Control Algorithms

In the late seventies the first approaches to long-range predictive control methods, e. g. the nonparametric model based DMC, MAC and IDCOM algorithm, have been introduced in literature. Following the main idea of long-range predictive control, several control algorithms have been proposed since then. Some of them - EHAC, EPSAC, GPC, OCS, MURHAC, DMC, MAC and MUSMAR - will be discussed here. The enhancement of all these algorithms with adaptive parameter adjusting techniques leads to their main advantage in dealing with variable or unknown dead-time processes. As in a sense, all these algorithms share the same definition based on using, partly or completly, a common optimization criterion, this joint approach may be used for a common discussion. Within this discussion, a redefrinition of certain methods is proposed and, using this, conclusions of possible aspects of usability for distinct process demands are found in theoretical and simulative examinations. As the GPC algorithm seems to be a favourite one, the disadvantage of computational expense is reduced by proposing a new algorithm.