2011 Volume 19 Issue 5 Pages 41-49
In a previous parameter design study to improve the fuel efficiency of a car,guidelines for fuel-efficient driving were obtained, together with several findings concerning the fuel consumption behavior of cars. In practice, however, the optimal conditions obtained in the experiment were difficult to maintain in daily driving because of the effects of various other factors, including other traffic and the weather. An attempt was therefore made to predict actual mileage by bringing these varying factors into the experiment, by predicting mileage using the Both-sides Taguchi method with data obtained while the author drove to work. Predictive accuracy adequate for practical use was obtained. Effects on predictive accuracy were confirmed in cases such as when the unit space was defined using mean values of the signal data instead of being defined in the ordinary way. Through selection and diagnosis of specific items, it was objectively shown that frequent stops at traffic signals, wet road surfaces, and low temperatures lead to low fuel efficiency.
