Soft Computing Techniques for Control of Non Linear Systems

Abstract

newline Control of Inverted pendulum system is one of the most popular and challenging problem which has been continuously attracting researchers in the past few decades. The system comprises of rigid/flexible rod or pendulum pivoted on movable cart, therefore it is also called cart and pole system. A normal pendulum is stable when hanging downwards whereas an inverted pendulum is inherently unstable, and must be actively balanced in order to remain in upright position. The objective is to control pendulum at vertically upright position when cart is controlled stationary at particular location. The inverted pendulum system has its center of mass above its pivot point which makes the complete system highly unstable. These are highly non-linear, multi-variable, unstable and complex system which finds wide applications in various sectors including Aerospace, Robotics, Defence and Industrial etc. These systems act as testing bed for various controllers and algorithms. newline There exist various configurations of Inverted pendulum systems whose dynamics mimics behavior of various physical systems like Humanoid robot, aircraft motion in turbulence, ship anti-sway motion, launching of missile or satellite etc. The different types Inverted pendulum systems which are considered for control in this thesis includes Elastic inverted pendulum, Double inverted pendulum, Triple inverted pendulum, Rotary inverted pendulum, Gantry inverted pendulum, Inverted pendulum climbing on inclined plane, Single and Two wheel mobile robotic systems etc. The above mentioned nonlinear systems are controlled in Matlab-Simulink environment via various soft-computing control techniques like Proportional-integrative-derivative (PID), fuzzy logic reasoning, Adaptive neuro fuzzy inference system (ANFIS), and Neural networks. newline The study not only examines the control phenomenon of all these variants but also considers the effect of variation of physical attributes like mass of cart, mass of pendulum, length of pendulum, friction etc on their perf