Selection of Deterministic Models for Adaptive Regulation of Industrial Time-Delay Processes

Abstract

Performance of adaptive regulation systems has been analysed for different forms of linear deterministic model of a single-input/single-output time-delay process. In particular, the ordinary difference equation, weighting sequence, Wittenmark and, to some extent, state-space models have been investigated. It is possible for open-loop stable minimum phase systems only to be minimum variance controlled, no matter what kind of input-output model quoted is initially used. On the contrary, it has been shown that there are quite different conditions for different input-output models to optimum controlling open-loop unstable and/or nonminimum phase systems. Specifically, different performance Indices must be used for different models to gain equivalent, favourable control performance and, as a result, different control strategies are obtained, with different numbers of weighting coefficients to be Together with some computational aspects of both identification and control stages, this forms a basis for a new aptive-control-oriented classification of linear deterministic to be hopefully utilized especially in computer control engineering problems. Surprisingly, a common difference equation model has appeared to be most advisable.