Identification and analysis of fuzzy model for air pollution-an approach to self-learning control of CO concentration

The authors present identification and control for a fuzzy prediction model of CO (carbon monoxide) concentration. There are many uncertainty (imprecise) factors for predicting CO concentration. The basic approach proposed is to handle this imprecision by fuzzy-logic-based techniques. The fuzzy modeling technique proposed by G.T. Kang and M. Sugeno (see Fuzzy Sets and Systems, vol.18, no.3, p.329-46, 1986) is used for identifying a fuzzy prediction model. The model identified concerns the prediction of CO concentration in the air at a traffic intersection point of a large city of Japan. It is shown that the identified fuzzy model is very useful for predicting CO concentration. Furthermore an attempt is made to simulate a self-learning control of CO concentration by the Widrow-Hoff learning rule. Simulation results show that this self-learning controller is useful for CO concentration control