Fuzzy rule-based auto-tuned internal model controller for real-time experimentation on temperature and level processes

Abstract

Recently, internal model control (IMC) technique has been widely employed for various industrial close-loop control applications. Rewarding feature of IMC controller is that we need to tune only one parameter λ (close-loop time constant) for achieving the desired close-loop response. But, finding an appropriate value of λ is not an easy task. From the basic behaviour of IMC-based close-loop responses, it is found that when the process variable is moving very fast towards the desired value, relatively larger value of λ (smooth control) is required to reduce the possible oscillations. In contrary, smaller value of λ (tight control) is preferred when the process response is quickly moving away from the set point to restrict its further deviation. Hence, to mitigate the limitation of conventional IMC tuning with a fixed λ, a fuzzy rule-based auto-tuning scheme is proposed here for IMC-proportional integral derivative (IMC-PID) controller and its performance is validated through real-time experimentation on temperature and level control loops.