Automatic T-S fuzzy model with application to designing predictive controller

Zhi-gang Su, Pei-hong Wang, Yu-fei Zhang

A novel methodology is proposed to automatically extract T-S fuzzy model with enhanced performance using VABC-FCM algorithm, a novel Variable string length Artificial Bee Colony algorithm (VABC) based Fuzzy C-Mean clustering technique. Such automatic methodology not requires a priori specification of the rule number and has low approximation error and high prediction accuracy with appreciate rule number. Afterward, a new predictive controller is then proposed by using the automatic T-S fuzzy model as the dynamic predictive model and VABC as the rolling optimizer. Some experiments were conducted on the superheated steam temperature in power plant to validate the performance of the proposed predictive controller. It suggests that the proposed controller has powerful performance and outperforms some other popular controllers.