基于实际行驶工况的纯电动汽车续驶里程在线估算方法研究  

Study on Method for Online Estimating Driving Range of Battery Electric Vehicle Based on Actual Driving Cycle

作　　者：魏恒 何超[1,2] 李加强 赵龙庆[1,2] WEI Heng;HE Chao;LI Jia-qiang;ZHAO Long-qing(School of Machinery and Transportation,Southwest Forestry University,Kunming Yunnan 650224,China;Key Lab of Vehicle Environmental Protection and Safety in Plateau and Mountainous Areas of Universities in Yunnan Province,Kunming Yunnan 650224,China)

机构地区：[1]西南林业大学机械与交通学院,云南昆明650224 [2]云南省高校高原山区机动车环保与安全重点实验室,云南昆明650224

出　　处：《公路交通科技》2020年第12期149-158,共10页Journal of Highway and Transportation Research and Development

基　　金：国家自然科学基金项目(51968065);云南省应用基础重点项目(2017FG001(-010));云南省"万人计划"项目(YNWR-QNBJ-2018-066)。

摘　　要：为了提高纯电动汽车续驶里程估计方法的准确度,在采用主成分分析和模糊聚类相结合的方法分析汽车实际行驶工况的基础上提出了一种在线续驶里程估算模型。以某型号纯电动汽车在实际道路上行驶采集的运行数据为研究对象,根据60 s为时间间隔将工况数据进行划分,一共得到了420个行驶工况片段,选取了平均速度、最高速度和速度的平方和等11个可以描述行驶工况的运动学特征参数。接着对行驶工况片段作了主成分分析,根据载荷因子矩阵,选取与主成分相关性较大的4个参数:最高速度、平均速度、驻车比例和平均加速度,作为模糊C均值聚类的特征参数。通过模糊C均值聚类分析后将行驶工况划分为了4类并且计算得到了聚类中心,根据纯电动汽车的整车主要技术参数建立了整车能耗模型,计算得出了4个聚类的平均能耗。最后利用另一辆同型号的纯电动汽车运行数据对该续驶里程估算方法进行了仿真验证。结果表明:该估算方法有较好的收敛性和鲁棒性,从试验车开始行驶,在大约经过100个行驶工况片段之后,剩余续驶里程的仿真估算值非常逼近于测试值,估算值与测试值之间的最大绝对误差为4.47 km,平均绝对误差为2.49 km,平均相对误差为3.76%。这说明此方法是可行的,所建立的模型具有较高的准确度。In order to improve the accuracy of the method for estimating the driving range of BEV,an online driving range estimation model is proposed on the basis of the method for analyzing actual driving cycle,which combines the principal component analysis with fuzzy clustering.Taking the driving cycle data of a certain type of BEV on actual road as the research object,the driving cycle data is divided with a time interval of 60 s,a total of 420 driving cycle segments are obtained,and the 11 kinematic characteristic parameters which can describe the driving cycle,such as average speed,the maximum speed,and sum of squares of speed,are selected.Then,the principal component analysis for driving cycle segments is conducted,according to the load factor matrix,4 parameters which have great correlation with the principal components(maximum speed,average speed,idling proportion and average acceleration)are selected as the characteristic parameters for fuzzy C-means clustering.The driving cycle is divided into 4 categories and the cluster center is obtained by fuzzy C-means clustering analysis,and an energy consumption model for the BEV is established by using its main technical parameters to calculate the average energy consumption of the 4 clusters.Finally,the estimation method of driving range is simulated and verified by the driving cycle data of another BEV of the same model.The result shows that the estimation method has good convergence and robustness.After the test vehicle has passed about 100 driving cycle segments from the start,the simulation estimation value of the remaining driving range is very close to the test value,the maximum absolute error between the estimation value and the test value is 4.47 km,the average absolute error is 2.49 km,and the average relative error is 3.76%,indicating that this method is feasible and the established model has high accuracy.

关 键 词：汽车工程 纯电动汽车 续驶里程 行驶工况 主成分分析 模糊聚类 

分 类 号：U469.72[机械工程—车辆工程]