基于车联网数据挖掘的营运车辆 驾驶速度行为聚类研究

为了充分利用交通运输企业积累的海量车联网数据，挖掘营运车辆驾驶行为 特征的潜在规律.根据车联网数据属性提取涉及驾驶行为特征的参数，基于因子分析把8 个驾驶行为特征参数化为少数几个蕴含明确驾驶行为信息的综合变量，以此为指标通过 系统聚类，将选取的江苏范围内营运车辆驾驶行为特征进行聚类分析.结果表明，营运车 辆驾驶行为特征可有效聚为变速行为、超速行为、减速行为、加速行为，其中变速驾驶行 为程度较重的驾驶人其他3 种驾驶行为程度也较大.这类驾驶人具有较高驾驶风险，交通 运输企业需要对其重点监控.研究结果对我国营运车辆驾驶人的监管与培训具有一定参 考作用.

Driving Speed Behavior Clustering for Commercial Vehicle Based on Connected Vehicle Data Mining

In order to take full advantage of mass connected vehicle data from transportation enterprise and find potential laws of driving behavior characteristic, the parameter of driving behavior characteristic are extracted based on the data. Then eight parameters of driving behavior characteristic transform into several aggregate variables of specific driving behavior information based on factor analysis, and driving behavior for commercial vehicle is analyzed by hierarchical clustering in Jiangsu province. Results show that driving behavior for commercial vehicle can be divided into four classes reasonably, such as speed changing, speeding, deceleration, acceleration. Particularly, drivers of speed changing have other driving behaviors and the levels are high too, and they are high-risk drivers, as an important factor of influence road traffic safety, so transportation enterprises could monitor them specially. Research results are of positive significance to improve the monitoring capability of drivers for commercial vehicle.