快速城镇化背景下的驾驶风格多样性分析

Driving styles during rapid urbanization

快速城镇化造成的交通拥堵、交通安全、交通环境等问题日益突出,从交通参与者个体的特性出发,分析驾驶风格多样性,从而建立"以人为本"的智能交通系统,是充分开发利用现有道路交通资源的有效途径。然而,驾驶风格难以检测和量化,这导致个体特征与系统性的计算难以进一步融合。该文通过实验车采集了16位驾驶员在实际道路上的驾驶行为数据,利用主题模型挖掘驾驶行为中的隐含主题,将数据结构由"驾驶风格—驾驶行为数据"转化为"驾驶风格—驾驶状态—驾驶行为数据"结构,发现了驾驶风格结构化信息,能够为建立更为有效的智能交通系统提供科学依据与理论支持。通过分析相关性,证实了模型重构数据与原数据有较好的一致性,验证了模型进行驾驶风格多样性分析的可行性。

Traffic congestion, traffic safety, and traffic related environmental problems caused by rapid urbanization have become increasingly important with ＂human-centered＂ intelligent transportation systems （ITS） needed that are based on individual driving characteristics. Thus, common driving styles were analyzed to more effectively utilize the existing road transport resources. However, driving styles are difficult to identify and quantify, which leads to difficulties modelling the individual driving characteristics. In this paper, the driving characteristics of 16 drivers on actual roads were analyzed using a topical model. The hidden driving characteristics were extracted and used to convert the data from ＂driving style-driving behaviour data＂ into ＂driving style-driving state-driving behaviour data＂. The discovered driving style structural information provides theoretical support for more efficient intelligent transportation systems. The reconstructed model data is shown to correlate well with the original data which verifies themodel for analyzing driving style diversity.