Implementation of an adaptive fuzzy compensator for coupled tank liquid level control system

In this paper, an adaptive fuzzy control (AFC) system is proposed to realize level position control of two coupled water tanks, often encountered in practical process control. The fuzzy control system includes an adaptive model identifier and controller. The gains of AFC are obtained by using the fuzzy identifier model which is defined by real system outputs and control inputs. The parameters of fuzzy identifier model are adjusted online by using recursive least square algorithm. Because the controller has a recursive form it treats model uncertainties and external disturbances in an implicit way. Thus there is no need to specify uncertainty and disturbances for this controller design in advance. A well-tuned conventional proportional integral (PI) controller is also applied to the two coupled tank system for comparison with the AFC system. Experimentation of the coupled tank system is realized in two different configurations, namely configuration #1 and configuration #2 respectively. In configuration #1, the water level in the top tank is controlled by a pump. In configuration #2, the water level in the bottom tank is controlled by the water flow coming out of the top tank. Experimental results prove that the AFC shows better trajectory tracking performance than PI controller in that the plant transient responses to the desired output changes have shorter settling time and smaller magnitude overshot/undershoot. Robustness of the AFC with respect to water level variation and capability to eliminate external disturbances are also achieved. Experimental results show that AFC is a strong and a practical choice for liquid level control. (C) 2016 Elsevier Ltd. All rights reserved.