Comparative Study of a Fuzzy Logic Based Controller and a Neuro-Fuzzy logic Based Controller for Computer Fan

Abstract

The impact of soft-computing in modern day engineering and technology cannot be overemphasized.  Fuzzy logic approach as proposed by Lofti Asker Zadeh, popularized by the Japanese, has found its way into the control of many domestic and industrial appliances/machines.  Unlike the popular PID controllers and the pulse width modulation based controllers, the performance of computer fan is investigated using the fuzzy logic approach with two inputs parameters, that is, the computer loads and the temperature and one output parameter which is the speed at which the computer fan operates.  For the fuzzy inference system, four membership functions are selected for the inputs as well as the output.  Relevant rules are set to determine the operating conditions and boundaries for the controller.  In order to make the controller adaptive, neurofuzzy logic appproach is used with parameters set as the case with fuzzy logic. Training of the controller is carried out and the performance of each controller is presented in three dimensional view and two dimensional surface view with neurofuzzy based controller, in performance, having an edge over the fuzzy logic based controller.

Keywords: Anfis, Fuzzy logic, Computer fan, Controller, Performance comparison