基于KNN-HA和KNN-RBF相融合的交通状态预测
发布时间：2018-10-12 09:14
【摘要】：随着机动车数量的不断增加,国内外大部分城市道路以及高速公路的交通拥堵状况日益严峻,这种现象严重的影响到了人们的日常工作和生活。为解决交通拥堵问题,智能交通系统被广泛采用并有效的缓解了拥堵的状况。由于交通数据的采集技术不断进步,使大量的历史数据作为交通状态的预测样本成为了可能。交通状态预测是智能交通系统进行交通管理中很重要的一部分,是交通诱导的前提。所以,交通状态预测的研究对交通规划及交通的优化控制有非常重要的作用。本文选择反映交通状态最直接的参数速度来作为状态预测的参数。针对现有速度预测方法的不足,提出了基于KNN-HA和KNN-RBF相融合的速度预测模型。首先使用KNN-HA方法和KNN-RBF方法对预测路段的速度做预测,分别得到周内和周末的预测结果。根据早晚高峰将一天分为5个时间段,比较每个时间段内两种方法的预测精度,得出了基于两种算法相融合的速度预测算法;其次将本文的方法与神经网络算法(NN)和支持向量回归算法(SVR)等经典方法进行比较,得出本文提出的预测模型优于其他预测模型,预测精度比支持向量回归算法提高了11%,比KNN-RBF算法提高了6%;最后根据速度阈值将交通状态划分为5个状态,以预测速度对道路的交通状态进行判断,并比较了预测值和实际交通状态的一致性,预测精度达到91.7%。
[Abstract]:With the increase of the number of motor vehicles, the traffic congestion of most urban roads and highways is becoming more and more serious at home and abroad, which seriously affects the daily work and life of people. In order to solve the problem of traffic congestion, Intelligent Transportation system (its) has been widely used and effectively alleviated the congestion. Because of the continuous progress of traffic data acquisition technology, a large number of historical data as traffic state prediction samples become possible. Traffic state prediction is an important part of intelligent transportation system in traffic management and the premise of traffic guidance. Therefore, the study of traffic state prediction plays an important role in traffic planning and traffic optimization control. In this paper, the most direct parameter speed which reflects the traffic state is chosen as the parameter of state prediction. A speed prediction model based on the fusion of KNN-HA and KNN-RBF is proposed to overcome the shortcomings of existing speed prediction methods. First, the KNN-HA method and the KNN-RBF method are used to predict the speed of the predicted section, and the results of the prediction are obtained at the end of the week and the weekend, respectively. According to the morning and evening peak, the day is divided into five time periods, and the prediction accuracy of the two methods in each time period is compared, and the speed prediction algorithm based on the fusion of the two algorithms is obtained. Secondly, compared with the classical methods such as neural network algorithm (NN) and support vector regression algorithm (SVR), the prediction model proposed in this paper is superior to other prediction models. The accuracy of prediction is 11% higher than that of support vector regression algorithm and 6% higher than that of KNN-RBF algorithm. Finally, the traffic state is divided into 5 states according to the speed threshold, and the traffic state is judged by forecasting speed. The consistency between the predicted value and the actual traffic state is compared, and the prediction accuracy is 91.7%.
【学位授予单位】：长安大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U491
[Abstract]:With the increase of the number of motor vehicles, the traffic congestion of most urban roads and highways is becoming more and more serious at home and abroad, which seriously affects the daily work and life of people. In order to solve the problem of traffic congestion, Intelligent Transportation system (its) has been widely used and effectively alleviated the congestion. Because of the continuous progress of traffic data acquisition technology, a large number of historical data as traffic state prediction samples become possible. Traffic state prediction is an important part of intelligent transportation system in traffic management and the premise of traffic guidance. Therefore, the study of traffic state prediction plays an important role in traffic planning and traffic optimization control. In this paper, the most direct parameter speed which reflects the traffic state is chosen as the parameter of state prediction. A speed prediction model based on the fusion of KNN-HA and KNN-RBF is proposed to overcome the shortcomings of existing speed prediction methods. First, the KNN-HA method and the KNN-RBF method are used to predict the speed of the predicted section, and the results of the prediction are obtained at the end of the week and the weekend, respectively. According to the morning and evening peak, the day is divided into five time periods, and the prediction accuracy of the two methods in each time period is compared, and the speed prediction algorithm based on the fusion of the two algorithms is obtained. Secondly, compared with the classical methods such as neural network algorithm (NN) and support vector regression algorithm (SVR), the prediction model proposed in this paper is superior to other prediction models. The accuracy of prediction is 11% higher than that of support vector regression algorithm and 6% higher than that of KNN-RBF algorithm. Finally, the traffic state is divided into 5 states according to the speed threshold, and the traffic state is judged by forecasting speed. The consistency between the predicted value and the actual traffic state is compared, and the prediction accuracy is 91.7%.
【学位授予单位】：长安大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U491
【参考文献】
相关期刊论文 前10条
1 乔春凯;赵佳文;;基于时序关联规则挖掘的交通拥堵预测研究[J];科技创新与应用;2017年01期
2 刘张;李坚;王超;蔡世民;唐明;黄琦;陈照辉;;基于复杂城市道路网络的交通拥堵预测模型[J];电子科技大学学报;2016年01期
3 姚磊;张开冉;;基于模糊逻辑的交通拥挤状态识别算法[J];西华大学学报(自然科学版);2014年03期
4 朱征宇;刘琳;崔明;;一种结合SVM与卡尔曼滤波的短时交通流预测模型[J];计算机科学;2013年10期
5 韦光波;韦维;胡尧;;一种交通拥堵流量范围界定及拥堵预测的方法[J];贵州大学学报(自然科学版);2012年05期
6 王吟松;王嘉文;张建跃;董秀军;杨晓光;;基于浮动车车速数据的交通状态实时判别系统和方法[J];交通信息与安全;2011年06期
7 潘志毅;;基于灰色神经网络的空中交通流量预测方法[J];微计算机信息;2011年09期
8 牛世峰;姜桂艳;李红伟;姜卉;;基于纵向时间序列的快速路交通事件检测算法[J];哈尔滨工业大学学报;2011年02期
9 聂佩林;余志;何兆成;;基于约束卡尔曼滤波的短时交通流量组合预测模型[J];交通运输工程学报;2008年05期
10 姜桂艳;Q，
本文编号：2265575
本文链接：https://www.wllwen.com/kejilunwen/daoluqiaoliang/2265575.html
