DESIGN AND ANALYSIS OF SELF-TUNING AND NEURO-FUZZY CONTROLLERS

Abstract

Most real systems, relevant from control perspective exhibit non-linear behavior and are characterized by poor models, high dimensionality of the decision space and high noise levels. Classical conventional techniques demand mathematical modeling of plants and prior tuning of controller parameters, which is often troublesome. This has resulted in motivation of intelligent control schemes, where fuzzy and neural controls have been recognized as attractive alternatives to the Classical control schemes for the low cost and facilitated design of control laws of partially known, non-linear and complex processes. This dissertation deals with the design of Fuzzy Logic based controller, Neural Network based controller and Neuro-Fuzzy controller. A simple model independent self-tuning scheme is implemented for fuzzy and neural controllers, where the output-scaling factor is adjusted online. NEFCON model where, fuzzy sets and rules are learned by Reinforcement learning algorithm is used for the creation of learning environment for neuro-fuzzy controller. Simulation results are included to demonstrate the influence of designed controllers on the performance of the closed loop control system with and without dead time in terms of peak overshoot, rise time, settling time and Integral of absolute error (IAE