摘要
基于通过实车试验采集的城市典型道路行驶工况数据,首先用主成分分析法对选取的12个表征道路运行特性的特征参数进行减缩,接着利用SOFM神经网络算法和K均值聚类法相结合的组合聚类技术对所有运动学片段的前3个主成分得分进行分类,再根据各类别的时间比例从各类别中选取合适片段,最终拟合出代表性工况。通过对各工况加速度分布的K-S检验和采用ADVISOR软件进行的发动机载荷谱和燃油消耗量仿真分析,表明和K均值聚类法相比,组合聚类法的行驶工况拟合精度更高,更能综合反映城市交通真实状况。
Based on the typical urban road driving data collected by real vehicle tests,the chosen twelve characteristic parameters representing driving feature are reduced by principal component analysis.Then the scores of first three principal components in all kinematic segments are classified by the combination of SOFM neural network algorithm and K-means clustering.And proper segments are selected from each category according to their duration percentage.Finally the representative driving cycle are fitted out.The results of the K-S test on the acceleration distribution of driving conditions and the simulation analysis on engine load spectra and fuel consumption indicate that compared with K-means clustering scheme,the combination clustering technique has higher accuracy in driving cycle fitting and can more comprehensively reflect the real situation of urban traffic.
出处
《汽车工程》
EI
CSCD
北大核心
2012年第2期164-169,158,共7页
Automotive Engineering
基金
国家自然科学基金(71071044)资助
