Parametric bounding for identification and adaptive control

Two parameter-bound-based adaptive control schemes are outlined. They are tested by comparison with the generalised predictive control and generalized pole-placement self-tuning control schemes. They are robust, and are reasonably easy to tune by choice of their design parameters (such as δ, ρ and the control and output horizons). Compared with more established schemes, they are computationally heavy, although sampling intervals of well under a second on a 386 PC have been attained. As with other self-tuning schemes, many variations on the details, e.g. of handling time variation or control constraints, are possible