Multi-scale models of processing systems offer an attractive alternative to
models defined in the time- or frequency-domain. They are defined on dyadi
c or higher-order trees, whose nodes are used to index the values of any va
riable, localised in both time and scale (range of frequencies). This dual
localisation is particularly attractive in solving estimation and control p
roblems, in this paper, multiscale models are used to design model-predicti
ve controllers (MPC), resulting in design techniques with several important
advantages, such as; (a) natural depiction of performance characteristics
and treatment of output constraints. (b) fast algorithms for establishing t
he constrained control policies over long prediction/control horizons, (c)
rich depiction of feedback errors at several scales, and (d) optimal fusion
of multi-rate measurements and control actions. (C) 2000 IFAC. Published b
y Elsevier Science Ltd. All rights reserved.