Model Order Reduction by Teaching Learning Based Optimization and FOPID Controller Designing

Reduced order systems, in comparison with original higher order systems make the topic of model reduction a subject of big importance as it provides a simpler system which is easy to be studied and used further for controller designing. To reduce the system Teaching learning based technique has been used in this paper. This method provides a stable reduced system for a stable original model. The technique is applied on the system such that the parameters of reduced system is optimized to obtain minimum error between the original and reduced order system. In this paper integral squared error (ISE) between the step response of original and reduced model is minimized. The performance of reduced order system has been analysed by step response and bode plot. A comparison has been provided based on ISE for different existing reduction methods. Robust and efficient performing systems are obtained by Fractional-order control (FOC) systems which is obtained by fine tuning of parameters. Fractional order Controllers are fractional-order differential equations. In this paper the reduced system's performance has been improved by fractional-order PID (FOPID) controller which is tuned by TLBO method. In expressions of percentage overshoot, rise time and settling time, FOPID controller gives best performance as compared with that of the PID controller.