Adaptive pole-placement control using parameter bounds

Adaptive control using robust pole-placement is developed on the basis of identified bounds on model parameters. The parameter values of the linear output-feedback controller are bounded by polytopes to guarantee that the closed-loop poles lie in a prescribed region. Given free choice from this set of feedback laws, pole assignment may be complemented with online experiment design to improve excitation and identifiability. To make the approach practicable in online applications, a new ellipsoidal parameter bounding is introduced. Using this type of parameter bounding the amount of controller calculation can be kept under control. The overall approach can find practical application where large abrupt changes are possible in system dynamics which stays steady for a while after each change, a typical situation in some electromechanical systems