摘要
[目的]运输行业管理部门利用车联网系统获取了大量驾驶员的时空轨迹点数据,而对行车轨迹点数据进行挖掘分析可以评估驾驶员的安全驾驶行为习惯,管理部门可以据此有针对性地对驾驶员进行教育监管,有助于规避风险,提高交通安全。而原始的轨迹点数据由于GPS信号被遮挡或者干扰等原因,会包含大量噪声及一些错误信息,需要有效清洗才能使用。[方法]文中以运输车辆原始轨迹点数据为研究目标,分析总结出了其中常见的六类数据点异常现象,包括无效属性信息、时间信息错误、车速零点漂移、速度变化率异常、信息量过少的轨迹路段、经纬度漂移等问题,并针对这些具体的问题提出了相应的数据清洗方法;[结果]最后将该方法成功应用于某运输企业提供的车辆轨迹数据点的清洗。[结论]结果表明,文中提出的数据清洗方法能够有效去除异常数据,为后续驾驶员行为评估提供高质量数据。
[Objective]A large number of drivers'temporal and spatial track point data are obtained by the transportation industry man⁃agement department using the Internet of vehicles system.And the mining and analysis of the driving track point data can evaluate the driver's safe driving behavior.According to this,the management department can conduct targeted education and supervision of drivers,which is helpful to avoid risks and improve traffic safety.However,due to the blocking or interference of GPS signals,a large amount of noise and some error information exist in the original track point data,which need to be cleaned effectively before use.[Method]This pa⁃per took the original track point data of transportation vehicles as the research target,analyzed and summarized six kinds of common ab⁃normal point data phenomena.These anomalies include invalid attribute information,error of time information,speed zero drift,abnor⁃mal rate of speed change,information is too little of the track section,longitude and latitude drift,etc.According to these specific prob⁃lems,the corresponding data cleaning method was proposed.[Result]Finally,the method was successfully applied to the cleaning of ve⁃hicle track data points provided by a transportation enterprise.[Conclusion]The results showed that the proposed data cleaning method can effectively remove abnormal data and provide high quality data for subsequent driver behavior evaluation.
作者
高静文
蔡永香
甘艺垚
GAO Jing-wen;CAI Yong-xiang;GAN Yi-yao(School of Geosciences,Yangtze University,Wuhan 430100,China)
出处
《电脑知识与技术》
2019年第12X期189-192,194,共5页
Computer Knowledge and Technology
基金
地理信息工程国家重点实验室开放基金课题资助(SKLGIE2017-M-4-6)
关键词
数据清洗
驾驶行为
车辆轨迹
可视化表达
时空数据
data cleaning
driving behavior
vehicle trajectory
visualization representation
spatio-temporal data
