On implicit modelling theory: Basic concepts and application to adaptive control (Q581326)

The paper is concerned with modelling aspects in adaptive/self-tuning systems and their use in control objectives based on pole-assignment and quadratic optimisation. Assuming that the actual system is described by an input-output (linear) model with coloured noise, the paper discusses when the system can be characterized by an input-output model with white noise. The existence of the latter type of model normally requires the system to be in cloed-loop operation under a control law. However, it is shown that a certain subclass of these models can still be used even if the inputs (over some specified time instants) had been chosen arbitrarily. These models are useful when the objective of quadratic optimization is pursued.