Research On Fuel Consumption Analysis Based On Machine Learning

With the rapid development of China’s economy,the number of motor vehicles has increased year by year.The widespread use of motor vehicles has improved the life quality of people and brought serious environmental pollution.The evaluation of energy saving driving level has great significance to reduce the energy consumption in the transportation industry.At present,the research on energy saving evaluation about drivers ignores the influence of vehicle driving conditions on driving mode,and usually carries out research on selected sections of road.There is a lack in a single application scenario.In this thesis,a method based on machine learning for evaluating driver’s energy conservation is put forward.The vehicle driving condition is applied to the evaluation of driver’s energy saving,and the influence of road traffic condition on energy saving evaluation is eliminated.First of all,this thesis divides kinematics segments through travel analysis,which is based on principal component analysis and cluster analysis,and according to the driving characteristics of the vehicle from the overall classification to build the vehicle driving condition data set.Secondly,the main purpose is to analyze vehicle fuel economy.Based on the collected CAN bus data,the main factors affecting vehicle fuel consumption are studied,and the characteristic parameters of energy saving driving are summarized.Then,the kinematic segment data set is processed based on SMOTE algorithm to solve the problem of uneven distribution of data samples.Finally,based on the machine learning algorithm,a kinematic segment classification model is established.According to different classification scenarios,the classification model can adaptive select the best algorithm in random forest,support vector machine,nearest neighbor k and the XGBoost to analyze the energy saving degree of kinematics segment.The energy saving evaluation scores of drivers can be calculated through the formula of energy saving evaluation based on weighted scoring method.This thesis carries on the statistics and the mining to the vehicle traveling data.Through the comparison and analysis of the existing methods of energy conservation evaluation of drivers,it is shown that the evaluation method proposed in this thesis can eliminate the influence of road traffic conditions on energy conservation evaluation.The evaluation methodis universal and can be popularized in practical application.