This is our recent paper published in the Journal of Transportation Engineering and Information Technology, and the code has also been open source:
-
https://github.com/RobinYaoWenbin/Research-on-vehicle-travel-behavior-portrait-analysis-relating-to-the-topological-characteristics
-
https://gitee.com/RobinYaoWenbin/Research-on-vehicle-travel-behavior-portrait-analysis-relating-to-the-topological-characteristics
Paper link: https://doi.org/10.19961/j.cnki.1672-4747.2022.09.011
overview
The thesis first extracts the travel behavior characteristics of vehicles based on the license plate recognition data, and then uses clustering algorithm for cluster analysis to divide the vehicles on the road network into several categories. The road network data of openstreetmap is crawled by web crawlers, and the crawled road network topology data and license plate recognition data are fused together. On this basis, complex network methods and clustering algorithms are combined to analyze the image of intersections. Combining the results of vehicle portraits and intersection portraits, the travel behavior and travel characteristics of different types of vehicles on the road network are analyzed.
data source
The data sources used are the license plate recognition data and osm data, among which the license plate recognition data is not convenient to open source, and the osm data has source code, so you can crawl it yourself.
in conclusion
- Travel vehicles in the city can be divided into five categories, including temporary business vehicles, frequent crossing vehicles, household infrequently used vehicles, business vehicles, commuting vehicles, online car-hailing vehicles, taxis, and corporate business vehicles. Taking Xiaoshan District of Hangzhou City as an example, the number of five types of vehicles is 1558 287, 430 840, 193 826, 202 701, 44 661 respectively.
- Road network intersections can be divided into four categories. The first category is actually moderately important but topologically unimportant. The second category is topologically and practically very important intersections. The third category is topologically important. The intersections that are more important but actually unimportant, the fourth type are the intersections that are actually and topologically unimportant.
- Commuting vehicles have the most obvious commuting phenomenon in the morning and evening peaks. Online taxis, company business vehicles, and frequent cross-border vehicles also have obvious morning and evening peak travel phenomena, while family seldom-used vehicles, office vehicles, and temporary office vehicles have no obvious morning and evening peak travel phenomena. Therefore, for commuter vehicles, online rental vehicles, office vehicles, and frequent crossing vehicles, control measures during morning and evening peak hours may have a better effect, while for household vehicles that are not commonly used, office vehicles, and temporary office vehicles, control measures during the day likely to be more effective.
The passing frequency and passing frequency ratio of various types of vehicles at various types of intersections are analyzed. According to the analysis results, it can provide support for policy formulation. For example, when formulating control measures for commuter vehicles, commuter vehicles can be considered Intersections with a large proportion of travel.
references
Yao Wenbin, Rong Donglei, Hu Youwei, Su Hongyang, Chen Nuo, Jin Sheng. Research on Vehicle Travel Behavior Profile Analysis Based on Topological Characteristics of Road Network[J/OL]. Journal of Traffic and Transportation Engineering and Information: 1-20[2022-11-03] .DOI: 10.19961/j.cnki.1672-4747.2022.09.011.
Yao W.,Rong D.,Hu Y., et al.,Research on vehicle travel behavior portrait analysis relating to the topological characteristics of road network. Journal of Transportation Engineering and Information, 2022, doi:https://doi.org/10.19961/j.cnki.1672-4747.2022.09.011.