Adaptive control of nonlinear systems with applications

Modelling and parametric identification of mechanical systems withnonlinearities is in the scope of the research. Nonlinear system representationschemes for parametric system identification are discussed in detail. Algorithms foridentification of the models of concern are derived and simulated. Special attention ispaid on the Hammerstein system structure. Results are tested on a real nonlinearmechanical system composed of a DC motor with uncertain load.This thesis considers adaptive control of an unstable mechanical system withthe approximate input-output linearization technique. The nonlinear system ofconcern is the beam and ball mechanism with open-loop unstable, highly nonlinearand underactuated nature. The system has ill-defined relative degree, which is aserious handicap for input-output linearization application. The method of input-output linearization is reviewed, and adaptive input-output linearization controldesign is given in detail. The beam and ball system model is developed usingNewton dynamics and Euler-Lagrange equations. Using the beam and ball example,an approximate version of the adaptive input-output linearizing tracking control isdesigned. The beam and ball model is modified for suitability for adaptive tracking.The system with the proposed control structure is tested via computer simulationsand experiments. Results present the tracking performance of the designed control.Key words: Adaptive control, approximate input-output linearization, beam and ball,Hammerstein system identification.