Time and Environment Dependency Aware Fuel Consumption Tracking Method for Improving Drivers and Trucks Management

Until the following two things would have become able to do while adapting to the changes of the time and environment, the level of saving fuel consumption must be unfortunately limited because the key factors affecting fuel consumption are varied hourly on a case by case: 1) driving controls such as gear/accel/idling can be adjusted by each driver and 2) each truck can be mechanically best maintained so as to improve the fuel efficiency. Thus, this paper proposes a time and environment condition dependency aware fuel consumption tracking method, which enables to give a timely feedback to the drivers and truck maintenance facility by making more clear what are the key factors affecting fuel consumption every period while considering their time and environment dependencies. Since the data obtained by embedded sensors (canbus&GPS) are complicated and its size is very huge, the LightGBM is used to timely abstract the key factors in this work, while comparing the prediction accuracy and available abstract information with the other models of SVM and Random Forest (RF). The results have shown that the best accuracy of 98.7% was achieved among the three models. SHAP value is used to explain the key influencing factors, which is used for feedback to each driver for his adjustment. It was found that overall fuel consumption can be reduced by 16% if some key factors could be carefully reduced by 50%.