Development of an adaptive neurofuzzy controller

Abstract

Practical systems have successfully been controlled by adaptivestrategies which used the feedback error to estimate the requiredcontrol action. This thesis aims at improving the performance ofsuch controllers by including the feedback error and its historyin the learning rule. Emphasis is placed on developing a stabilitycriteria and studying an alternative method for commissioning theadaptive controller. Analysis of the performance of the adaptiveneurofuzzy controller is extended to non-linear plants, with aliquid level plant and a pH neutralization process being used astest beds. Experimental results are presented to demonstrate thefeasibility of the proposed controller.