基于改进K-Means算法的交叉口影响路段行程速度估计
发布时间：2019-03-25 09:14
【摘要】：基于低频、低覆盖率、数据来源多样的GPS浮动车数据,在现有数据预处理方法的基础上,以交叉口影响路段数据点为研究对象,研究出更合理且准确获得交通参数的技术方案。GPS浮动车数据由于其具有全天候、多覆盖等特性,能够实时监测交通参数,估计交通状态。为克服数据本身缺陷,使数据能有效利用,精确得到交通参数,本研究获取短时内路段所有数据点代表整体状态。首先基于数据的特性和在路段分布的节律,利用曲线拟合及拉格朗日中值定理确定交叉口的影响范围;其次在该范围内利用改进K-Means聚类方法,确定初始聚类中心,并以有效性指数作为优化目标确定聚类数;在此基础上分配权重,结合交叉口影响范围外的数据点,对整个交叉口影响路段的行程速度进行估计。用杭州市局部路网中GPS数据进行案例分析,验证技术方案。通过实地调查获取实验真实值,分别讨论了在主、次干路路段本方案估计差异,并与传统模型进行了对比分析。分析表明,该方法得到的路段行程速度估计值与真实值较为接近,误差较小,在城市主干路和次干路中的误差分别为4.1%和9.5%,比传统模型误差更小更稳定,能较好地满足城市智能交通控制系统对于交通参数的精度要求。
[Abstract]:Based on the GPS floating car data with low frequency, low coverage and diverse data sources, based on the existing data preprocessing methods, the data points of intersection affected sections are studied. A more reasonable and accurate technical scheme for obtaining traffic parameters is proposed. GPS floating car data can monitor traffic parameters and estimate traffic status in real-time because of its all-weather and multi-coverage characteristics. In order to overcome the defect of the data, make the data can be used effectively, and get the traffic parameters accurately, all the data points in the short-term road section represent the whole state. Firstly, based on the characteristic of data and the rhythm of road section distribution, the influence range of intersection is determined by curve fitting and Lagrange mean value theorem. Secondly, the improved K-Means clustering method is used to determine the initial clustering center, and the efficiency index is used as the optimization objective to determine the clustering number. On the basis of this, the travel speed of the whole intersection is estimated by assigning weights and combining with the data points outside the influence range of the intersection. Using the GPS data of Hangzhou local road network to carry on the case analysis, validates the technical scheme. The actual value of the experiment was obtained through the field investigation, and the difference of the estimation of this scheme on the main and secondary trunk roads was discussed respectively, and the difference was compared with the traditional model. The analysis shows that the estimated value of road travel velocity obtained by this method is close to the real value and the error is small. The error in urban main road and secondary trunk road is 4.1% and 9.5% respectively, which is smaller and more stable than the traditional model. It can meet the precision requirement of urban intelligent traffic control system.
【作者单位】： 上海理工大学管理学院;
【基金】：教育部人文社会科学研究青年基金项目(17YJCZH225) 上海理工大学人文社会科学基金项目(SK17YB05)
【分类号】：U491
本文编号：2446837
[Abstract]:Based on the GPS floating car data with low frequency, low coverage and diverse data sources, based on the existing data preprocessing methods, the data points of intersection affected sections are studied. A more reasonable and accurate technical scheme for obtaining traffic parameters is proposed. GPS floating car data can monitor traffic parameters and estimate traffic status in real-time because of its all-weather and multi-coverage characteristics. In order to overcome the defect of the data, make the data can be used effectively, and get the traffic parameters accurately, all the data points in the short-term road section represent the whole state. Firstly, based on the characteristic of data and the rhythm of road section distribution, the influence range of intersection is determined by curve fitting and Lagrange mean value theorem. Secondly, the improved K-Means clustering method is used to determine the initial clustering center, and the efficiency index is used as the optimization objective to determine the clustering number. On the basis of this, the travel speed of the whole intersection is estimated by assigning weights and combining with the data points outside the influence range of the intersection. Using the GPS data of Hangzhou local road network to carry on the case analysis, validates the technical scheme. The actual value of the experiment was obtained through the field investigation, and the difference of the estimation of this scheme on the main and secondary trunk roads was discussed respectively, and the difference was compared with the traditional model. The analysis shows that the estimated value of road travel velocity obtained by this method is close to the real value and the error is small. The error in urban main road and secondary trunk road is 4.1% and 9.5% respectively, which is smaller and more stable than the traditional model. It can meet the precision requirement of urban intelligent traffic control system.
【作者单位】： 上海理工大学管理学院;
【基金】：教育部人文社会科学研究青年基金项目(17YJCZH225) 上海理工大学人文社会科学基金项目(SK17YB05)
【分类号】：U491
【相似文献】
相关期刊论文 前2条
1 游峰;张荣辉;李福j;冯琼莹;徐建闽;;基于K-means算法的行人检测方法研究[J];公路交通科技;2014年07期
2 陈建军;;浅析道路交叉口路面破损现象[J];科技与企业;2014年05期
相关会议论文 前1条
1 陈磊;胡佳敏;严华;;K-means算法在散货船代货运系统中的应用[A];全国第20届计算机技术与应用学术会议（CACIS·2009）暨全国第1届安全关键技术与应用学术会议论文集（上册）[C];2009年
相关硕士学位论文 前1条
1 郭鹏飞;城市交通信号自适应优化控制系统研究[D];南通大学;2016年
，本文编号：2446837
本文链接：https://www.wllwen.com/kejilunwen/daoluqiaoliang/2446837.html
