基于k-means++聚类的轻型汽车行驶工况构建

Construction of Light-duty Vehicles Driving Cycle Based on k-means++ Clustering

为了优化汽车使用性能,推进绿色交通发展,以福州市轻型汽车作为研究对象,对轻型汽车工况构建方法进行了研究.将采集的496467条数据进行预处理,从中提取了1340个短行程片段;利用主成分分析法和k-means++聚类算法对特征参数矩阵进行降维和分类处理,并以Silhouette指数作为衡量聚类结果的标准,构建具有代表性的福州市轻型汽车行驶工况并与实测数据及国外工况进行对比分析.结果表明:构建的工况与NEDC和FTP-15工况均存在显著差异,平均相对误差分别达到24.49%和30.57%;与实测数据的平均相对误差为3.12%,各个特征参数偏差均小于6.5%,能够满足开发需求.

In order to optimize the vehicle performance and promote the development of green transportation,the light-duty vehicles in Fuzhou were taken as the research object,and the construction method of light-duty vehicles was studied.496467 pieces of data were pre-processed,and 1340 short-run segments were extracted from them.Principal component analysis and k-means++ clustering algorithm were used to reduce the dimension and classify the characteristic parameter matrix.Taking the Silhouette index as the standard to measure the clustering results,the representative driving conditions of light vehicles in Fuzhou were constructed,and compared with the measured data and foreign driving conditions.The results show that there are significant differences between the constructed working conditions and NEDC and FTP-15 working conditions,with the average relative error reaching 24.49%and 30.57% respectively,and the average relative error with the measured data is 3.12%.The deviation of each characteristic parameter is less than 6.5%,which can meet the development needs.