论文部分内容阅读
为解决需采集大量汽车工况数据以保证试验数据的代表性以致试验成本显著增加的问题,研究了路网的等级特性,提出了一种基于K核和最优分割理论的城市道路等级划分方法。K核算法通过同时考虑道路流量和道路位置,使得Ks值相近的道路具有相似的速度和加速度概率分布;并可以根据不同道路的Ks值,采用最优分割理论将整个道路网络分为若干等级;伪F统计量被用来确定路网划分的合理等级个数。根据长春2辆出租车1年的行驶数据和通化1辆出租车半年的行驶数据,将长春和通化的路网分为3个等级。结果表明:同一等级下道路的速度、加速度概率分布具有很高的相似度,尤其是占城市绝大部分流量的前两等级,其样本相似系数αab高达70%以上,因此在进行汽车运行工况数据采集的过程中,只需采集每种等级中很少的几条道路就可以代表整个路网的特征;利用路网的等级特性研究的K核分级方法,可以在保证数据代表性的前提下,减少汽车运行工况数据采集的时间和数据量。
In order to solve the problem of acquiring a large amount of automobile working condition data in order to ensure the representativeness of the test data so that the test cost is significantly increased, the grade characteristics of the road network are studied and a method for classifying the urban road based on K kernel and the optimal segmentation theory is proposed . K kernel method can make the road with similar Ks value have similar speed and acceleration probability distribution by considering both the road traffic and the road location at the same time. According to the Ks value of different roads, the optimal road segment theory can be used to divide the whole road network into several grades; Pseudo-F statistics are used to determine the number of reasonable levels of road network division. According to Changchun 2 taxi driving data for 1 year and Tonghua 1 taxi six months of driving data, the road network in Changchun and Tonghua is divided into three levels. The results show that the probability distributions of speed and acceleration of road under the same class have high similarity, especially the first two grades which account for most of the traffic in the city. The sample similarity coefficient αab is over 70%. Therefore, In the process of data acquisition, only a few roads in each level can be collected to represent the characteristics of the whole road network. Using the K-core classification method of grade characteristics of road network, under the premise of ensuring data representation , Reduce the time and data volume of vehicle operating conditions data acquisition.