COMPARATIVE ANALYSIS OF MIT RULE BASED ONLINE AND DIFFERENTIAL EVOLUTION BASED OFFLINE TUNING METHODS FOR PID CONTROLLERS

Abstract

In high accuracy applications such as designing autopilot system for aircrafts, missile guiding systems and in various fields for robotics and automation, one needs to design the control system with more powerful and advanced techniques so as to maintain the satisfactory performance of the overall system. This thesis presents the comparative analysis of MIT rule based control with Differential Evolution (DE) algorithm based control by implementing them on magnetic levitation system and ball & beam system. It also shows the development of adjustment mechanism with necessary mathematics using gradient algorithm based MIT rule along with the mathematical modeling of magnetic levitation system and ball & beam system. In the thesis, closed ioop controllers have been designed based on two techniques and the performance of these controllers has been evaluated in simulations as well as in real time for both systems. It also compares DE tuned PID controller with Z-N tuned PID controller on third order, fourth order and unstable second order linear systems. The thesis also proposes the hybridization of MIT rule and DE algorithm for magnetic levitation system, where the online as well as the offline tuning of PID controller parameters is carried out. The simulations have been performed using MATLAB, and a comparative study among two strategies has been carried out based on these simulations. The performance of DE based controller with appropriately chosen range for parameters is better than MIT based controller. Selection of this range is very critical in DE algorithm and is carried out carefully. The performance of MIT controller is very much depends upon the value of adaptation gain and selection of this gain decides the performance of the system as shown in the thesis. Results also depict that MIT rule control is very sensitive to parameter variations, whereas DE based control shows robust performance when system is subjected to parameter variations