文章摘要
朱云霞,郭唐仪.基于GA-BP神经网络的交叉口短时交通流预测[J].交通运输研究,2019,5(2):45-51.
基于GA-BP神经网络的交叉口短时交通流预测
Intersection Short-Term Traffic Flow Forecast Based on GA-BP Neural Network
  
DOI:
中文关键词: 交叉口  短时预测  遗传算法  准实时  BP神经网络
英文关键词: intersection  short-term prediction  genetic algorithm  quasi-real time  BP neutral network
基金项目:
作者单位
朱云霞 南京理工大学 
郭唐仪 南京理工大学 
摘要点击次数: 45
全文下载次数: 83
中文摘要:
      交叉口车辆运行效率不高主要由两个原因造成:一是目前很多交叉口交通信号的配时与实际交通流不匹配;二是当前短时交通流预测的时长集中在5~15min,过长的时间间隔无法给信号控制提供准确的数据支撑。为了改善这种状况,提出了一种基于优化的BP(Back Propagation)神经网络的交叉口短时交通流预测方法。首先针对流量预测提出了准实时的概念,其次引入BP神经网络,再利用遗传算法对BP神经网络进行优化,最后在优化的BP神经网络的基础上建立交叉口短时交通流预测模型,且将传统的预测时长从5min 缩短到5s。利用南京市某一道路交叉口采集的数据对提出的预测方法进行验证,结果表明:与传统的BP神经网络以5min为预测时长相比,该预测方法以5s 为预测时长能够将预测精度提高77%。
英文摘要:
      The inefficiency of vehicle operation at intersections was caused by the following two reasons. One was that the signal timing of the traffic signal of many intersections didn′t match the actual traffic flow. The other was that the short-term traffic flow prediction time was focused on 5 to 15 minutes which was too long to provide effective data support for signal control. To improve this situation, a short-term traffic flow prediction method based on optimized BP (Back Propagation) neural network was proposed for intersections. Firstly, the quasi- time real concept was defined for traffic prediction. Secondly, the BP neural network model was introduced. Then the BP neural network was optimized by genetic algorithm. Finally, the intersection short-term traffic flow prediction model was established based on the optimized BP neural network. Also, the forecast duration was shortened from 5 minutes to 5 seconds. Using the data collected at a certain intersection in Nanjing, the prediction method was verified and compared with traditional BP Neural Network prediction method with 5 minutes duration. The result shows that the prediction method used in the research can increase the prediction accuracy by 77%.
查看全文   查看/发表评论  下载PDF阅读器