Genetic Algorithm Based Self Tuning Regulator for ball and hoop system

In this paper, an optimal adaptive controller using Genetic Algorithm for a ball and hoop system is proposed. It is difficult to design an optimal PID controller for a ball and hoop system because of its continuously varying parameters. The proposed scheme employs a PID controller with an ability to adjust its parameters when the dynamic behavior of the system changes, thus maintaining the optimality in the response specifications. Genetic Algorithm is used for both parameter estimation and optimal PID tuning. The system response after applying disturbances shows the robustness of the scheme. Also, there is a remarkable improvement in the transient response specifications compared to the conventional Zeigler Nichols tuning.

I. Introduction

Almost all real time processes are stochastic in nature, i.e. a predetermined input-output relationship is not sufficient to determine the response of the system for any other input. The reason is the fact that all the physical systems are characterized by some degree of randomness in response because of both the internal nature of system and the unpredictable external disturbances. When the external parametric changes deviates the system from its normal characteristics, the identified model of the system is no longer valid. So it becomes necessary to model the process again by a new parameter estimation algorithm and update the controller in accordance with the new model identified. An adaptive algorithm which automatically identifies a suitable new model for the process and tune the controller accordingly as the plant dynamics changes will avoid the unnecessary process interruption. Moreover, it is important to ensure the optimality of the controller designed.