基于数据挖掘的危险货物运输风险驾驶行为聚类分析

不良驾驶行为(车速过高，车速变化过快等)影响危险货物及车辆的稳定性，是导致危险货物运输事故频发且后果严重的重要原因. 基于对营运车辆联网联控系统海量数据的分析处理，根据其属性提取8 个驾驶行为量化指标，采用因子分析和模糊C均值聚类相结合的方法，实现对危险货物运输车辆驾驶员风险驾驶行为的科学聚类. 结果表明，危险货物运输车辆驾驶行为特征可有效聚为急加减速、超速驾驶和变速驾驶3 种行为，且在每种驾驶行为下实现对驾驶员安全等级的分类. 由此可以识别风险较高的驾驶员，这对危险货物运输企业和行业管理部门有重要参考意义.

Dangerous Driving Behavior Clustering Analysis for Hazardous Materials Transportation Based on Data Mining

In the process of hazardous materials transportation, bad driving behaviors such as too high speed and too fast speed change affect the stability of hazardous materials and vehicles, resulting in frequent accidents of hazardous materials transportation with serious consequences. In this paper 8 indicators for driving behavior evaluation were selected for quantitative analysis based on the massive data of the operating vehicle networked control system. Then by combining factor analysis and FCM algorithm, the risk driving behavior of the drivers of hazardous materials transport vehicles can be clustered scientifically. The results show that the driving behaviors of hazardous materials transport vehicles can be divided into three categories: acceleration & deceleration, speeding and variable speed driving，and the classification of the driver's safety level is achieved under each driving behavior. Therefore, it can identify the drivers with higher risks, which is of great reference significance to the transportation enterprises and industrial management departments of hazardous materials.