Genetic Algorithm Tuned IMC-PI Controller for Coupled Tank Based Systems

The proportional Integral and Derivative (PID) Controllers remains one the most versatile and widely adopted controller for industrial as well as educational applications. However, the efficacy of this controller lies in the ability to know how to tune them effectively and efficiently to suit operational needs. There exist numerous approaches to tuning the gains of the controller with varying degrees of complexity. Of all the existing approaches, the internal model control (IMC) stands out because it requires only the filter gain to determine the corresponding PID parameters. However, the ability to determine the appropriate filter gain is also a challenge as it is often than not selected arbitrarily using a trial by error approach. To this end, in this work, a genetic algorithm (GA) technique has been adopted in tuning this filter parameter to eliminate the associated problems of the trial by error approach. The results of the implementation on the double couple tank problem show the performance of the GA tuned IMC outweighing that of the conventional GA tuned PI controller approach.