基于K-means聚类算法的公交运营时段分析

发布时间：2018-06-02 16:01

本文选题：智能公交 + 时段划分　；参考：《交通运输系统工程与信息》2014年02期

【摘要】：公交车辆在高低峰等不同时段的运营时间差异较大,因此,只有对各时段的运营时间分别加以分析才能准确掌握运营时间规律,这对提高公交运营方案的准点率具有重要影响,是公交运营分析和优化调度等工作的不可或缺的重要基础.目前我国公交时段划分主要依据人工经验,简单且粗糙.本文基于大量GPS运营数据,创新性地将K-means聚类算法应用于运营时段划分,并结合公交样本数据特点,提出一种改进的K-means聚类算法,其中改进了传统的初始簇中心选择方法,并设计了利用三角形不等式减少不必要的距离计算和基于模糊聚类思想的簇中心更新算法.十堰市和海口市公交的案例分析表明,本文的K-means聚类方法可行,改进算法的计算效率更高,划分的时段与实际调研分析结果更加吻合.
[Abstract]:The operation time of public transport vehicles in different periods, such as high and low peaks, is quite different. Therefore, only by analyzing the operation time of each time period can the operation time law be accurately grasped, which has an important effect on improving the punctuality rate of bus operation scheme. It is an indispensable foundation for bus operation analysis and optimal dispatching. At present, the division of transit time in China is mainly based on artificial experience, simple and rough. Based on a large number of GPS operation data, this paper innovatively applies the K-means clustering algorithm to the operation time division, and combines the characteristics of bus sample data, proposes an improved K-means clustering algorithm, in which the traditional initial cluster center selection method is improved. A cluster center updating algorithm based on fuzzy clustering is designed to reduce the unnecessary distance by using triangle inequality. The case study of Shiyan city and Haikou city shows that the K-means clustering method is feasible, the improved algorithm is more efficient, and the division time is more consistent with the actual investigation and analysis results.
【作者单位】：华中科技大学自动化学院图像信息处理与智能控制教育部重点实验室;
【基金】：国家自然科学基金(71171087,70971044) 国家社会科学基金重点项目(13&ZD175)
【分类号】：U491.17;U492.22

【参考文献】