基于神经网络的驾驶行为与油耗相关性分析

Correlation Analysis for Driving Behaviour and Fuel Consumption Based on Neural Network

随着我国汽车保有量的迅猛增长和环境保护的迫切需求,提高汽车节油性能和驾驶员的驾驶行为已成为汽车产业发展的一项重要任务。论文根据汽车主厂采集的大量实车运行数据,提取出汽车单日行驶数据中的相关驾驶行为与车辆行为数据,然后提取出其中跟汽车油耗相关性最强的特征值进行降维。并经由神经网络进行回归训练,寻找分组内车辆由于驾驶行为不同而导致油耗上的差异。根据训练的结果优化驾驶员的驾驶行为,从而达到降低油耗的目的。

With the dramatic increase of vehicle population and critical need for environment protection,improving automobile fuel economy and driving behavior have become a dominant task in development of automobile industry.In this paper,firstly abundant automobile running data is collected from automobile company,and then the data of driving behavior and vehicle behaviorare extracted from the database,next,the dimension of data is reduced and the highly relevant data is derived with fuel consumption,finally,a relationship between automobile fuel economy and driving behavior are searched based on neural network.In a conclusion,the results of neural network are utilized to optimize the driving behavior and decrease automobile fuel consumption.