Automatic Turning Methods for Look-up-table Fuzzy Controllers

Abstract

We have developed a refrigerant flow controller of package airconditioners using fuzzy control technoligies. A look-up-table type fuzzy controller is employed for ease of implementation. An increasing demand is, however, to add adaptability to machines'setting environments into existing performances without affecting the look-up-table structure of the controller. In order to solve this problem, we developed new automatic tuning methods for look-up-table type fuzzy controllers using two coefficients of compensation, scaling factor(SF)and weighting factor(WF). The optimal SF is first decided by neural-networks, then repeated adjusting yields WF gradually. We evaluated these methods by simulations. The methods might find applications in many other electrical appliances, where loopk-up-table type controlleres are implementad.