Abstract

This thesis deals with the issue of controlling nonlinear systems by integrating available classical as well as modern tools such as fuzzy logic and neural networks. The proposed approaches throughout this thesis are based on the well known first-order Sugeno fuzzy system. To achieve a better understanding of the approximation and interpolation capabilities of Sugeno fuzzy system, the influence of the fuzzy set parameters and the reasoning method on the interpolation function of the fuzzy system is investigated. Control of nonlinear system based on known dynamic is considered first. A fuzzy gain scheduling approach is developed. The proposed approach is based on quasi-linear dynamic models of the plant. Classical optimal controllers for each set of operating conditions were developed. These controllers are used to construct a single fuzzy-logic gain scheduling-like controller. Adaptive-neuro-fuzzy inference system was used to construct the rules for the fuzzy gain schedule. This will guarantee the continuous change in the gains as the system parameters change in time or space. This procedure is systematic and can be used to design controllers for many nonlinear systems. Also, a modeling approach of some known types of static nonlinearities is proposed. Control of nonlinear system with unknown dynamic is also considered. An adaptive feedback control scheme for the tracking of a class of continuous-time plants is presented. A parameterized Sugeno fuzzy approximator is used to adaptively compensate for the plant nonlinearities. All parameters in the fuzzy approximator are tuned using a Luapunov-based design. In the fuzzy approximator, a first-order Sugeno consequent is used in the IF-THEN rules of the fuzzy system, which has a better approximation capability compared with that of a constant consequent. Global boundedness of the adaptive system is established. Finally, simulation and experimentation are used to demonstrate the effectiveness of the proposed controllers.