Based on the definition of the kinematic sequence, large amounts of experimental data on driving cycles were classified and many kinematic sequences were obtained. Four principal components of every kinematic sequence were obtained through compression of 12 characteristic parameters representing features of road travel by principal component analysis ( PCA ) , including percent of time accelerating, percent of time decelerating, percent of time cruising, percent of time idling, average velocity, average driving velocity, maximum speed, average acceleration, maximum acceleration, minimum acceleration, standard deviation of velocity, standard deviation of acceleration. Then, the scores of the first and second principal components of all kinematic sequences were classified by fuzzy C-means clustering ( FCM). Proper sequences were selected to fit the representative driving cycle according to the correlation coefficient and the length ratio of time among all the categories. Through error analysis of the characteristic parameters and of the joint probability distribution of speed-acceleration, the construction of the proposed method was shown to be highly accurate, and the representative driving cycle could comprehensively reflect the real traffic conditions.%在定义了运动学片段的基础上,对典型道路上采集的大量工况试验数据进行划分,从而获得大量运动学片段.用主成分分析法对12个表征道路运行特征的参数(包括加速比例、减速比例、匀速比例、怠速比例、平均速度、平均运行速度、最大速度、平均加速度、最大加速度、最小加速度、速度标准偏差和加速度标准偏差)进行压缩,得到4个主成分.利用模糊C均值聚类技术对所有运动学片段的第一和第二主成分得分进行分类,根据相关系数的大小及各类别的时间长度比选取合适片段,最终拟合出代表性工况.通过对特征参数和速度、加速度联合概率分布的误差分析可知,所提出的构建方法精度较高,拟合工况能综合反映合肥市实际道路的交通状况.
展开▼
