Genetic algorithm optimized PID controllers for a solid oxide fuel cell system

Solid Oxide Fuel Cells (SOFC) is an emerging technology because of their ability to deliver power cleanly and efficiently. It is an electrochemical process involving reactants fuel and air, which are converted into electrical power/energy. In order to form a complete fuel cell system, we require a CPOX (Catalytic partial oxidation) fuel reformer to generate hydrogen (because hydrocarbons are the most commonly used fuel) and a DC/DC boost converter (because fuel cell voltages are low). Two most important parameters that need to be controlled are the fuel cell power and temperature to ensure efficient power delivery and safety of operation. The appropriate control/manipulated variables are the reactant flow rates. Currently available control strategies do not dedicate much focus to the optimization of controllers.;MATLAB/SIMULINK was used in this thesis for the development of a complete SOFC system as well as control design, implementation, optimization and simulation. Previous work on SOFC have established control but without focus on optimization of the control design. Also, controller design and implementation on standalone fuel cell models or approximated models may not portray or simulate their real time behavior accurately. This work uses a complete SOFC system and also focuses on optimization of the designed controller using the Genetic Algorithm optimization technique. Proportional, Integral and Derivative controllers are proposed in this thesis to achieve control of power and temperature of a SOFC system. Control of power uses a PI (Proportional Integral) controller. Control of temperature uses the combination of a feed forward controller and a PID (Proportional Integral Derivative) controller with a lead compensator which compensates any disturbance in the form of air inlet temperature changes. The controller has been optimized using a genetic algorithm technique to minimize the mean square error as a fitness function. The Matlab GA (Genetic Algorithm) toolbox is utilized to implement the optimization technique.;The results of implementing optimized controllers have been studied through simulation showing applicability to a wide range of load demand, by performing good disturbance rejection and satisfactory response time.