Fuel Consumption Estimates Based on Driving Pattern Recognition

Abstract

Fuel consumption and air pollution caused by transportation is becoming more and more serious. Developing an effective meso model to analyze dynamically the temporal and spatial distribution of fuel consumption in the urban road network has become a research focus. However, parameters used by most meso models in the traffic field are not detailed enough, and the generalization ability and accuracy of these models are still to be improved. For discovering typical driving features and the relationship with fuel consumption, this paper collected 150 million records within two weeks by the intelligent terminal based on CAN (Controller Area Network) installed in 600 private vehicles. These records contained the information of latitude, longitude, speed, fuel consumption, etc. Based on these data, this paper discovered typical driving patterns that were closely related to fuel consumption, and then developed a fuel consumption calculation model. At last, this paper developed a fuel consumption monitoring system to show the distribution of fuel consumption of the road network in Beijing.