Parameter identification for systems with bounded disturbances

This thesis contains three parts. In the first part, a set-membership identification algorithm is developed for systems with bounded disturbances. Except an upper bound of the disturbances, the disturbances are totally unknown. The algorithm contains a variable non-negative weighting factor which is selected according to whether the new observed data contains sufficient information. The proposed approach not only ensures that the estimation error is bounded and non-increasing, but also possesses some good inherent features for any bounded system input. Furthermore, it will be shown that, with persistent exciting input, the parameter estimate provided by the algorithm converges to a residual region, and its upper bound is also given.Master of Engineerin