中科新闻网杭州华为公司:关于城市交通流量论文摘要的翻译
来源：百度文库 编辑：中科新闻网 时间：2024/05/07 14:38:02
城市道路交通流量研究
摘 要
准确的短时段交通流量预测在良好的道路交通管理中越来越成为至关紧要的一个步骤，对我国城市发展具有重要的意义。
本文中提到很多现在已经在应用和存在的预测模型，如ARIMA模型；基于卡尔曼滤波理论的交通流量预测模；针对不同的交通状况、集合现有的预测模型中的几种来对交通流量进行预测的一种综合模型。
本论文主要尝试了运用BP神经网络的方法进行交通流量的实时预测。通过一定的处理程序，就能够自动地根据之前的交通流量预测未来5分钟的车流量，从而安排交通的疏导、缓解道路拥挤状况。
所设计的神经元需要48个输入节点和48个输出神经元。网络取48-20-48的结构。网络的设计是为了根据相关交通流量值预测未来5分钟内的车流量。首先训练一个具有较低误差平方和的具有理想输入的网络。每一次训练后的权矢量作为下一组输入矢量训练的初始值。所有的设计思想均在程序中体现，其中的网络训练是用自适应学习速率及附加动量法的函数traingdx.m进行的。训练结果通过通流量预测的训练性能图表现。
该方法具有很强的学习能力和自适应性,因而具有很好的应用价值。
城市道路交通流量研究
Urban road traffic flow research
摘 要
准确的短时段交通流量预测在良好的道路交通管理中越来越成为至关紧要的一个步骤，对我国城市发展具有重要的意义。
The accurate short time interval traffic flow forecast more and more becomes an extremely important step in the good road traffic management, has the vital significance to our country urban development.
本文中~~~~
In this article mentioned very many now already in the application and the existence forecast model, like ARIMA model; Based on kalman filtering theory traffic flow forecast mold; In view of the different transportation condition, in the set existing forecast model several kinds come to the traffic flow to carry on the forecast one kind of unified model. The present paper mainly attempted has carried on the traffic flow using the BP nerve network method the real-time forecast. Before through the certain disposal procedure, will be able automatically to act according to the traffic flow forecast future 5 minutes traffic flow magnitudes, thus the arrangement transportation will unblock, alleviates the road congestion condition. Designs the neuron needs 48 inputs pitch points and 48 outputs neurons. The network takes 48-20-48 structure. The network design will be for according to the correlation transportation flow rate forecast future 5 minute in traffic flow magnitude. First trains has a lower error 平方和 has the ideal input the network. After each training power vector took the next group inputs the vector training the starting value. All designs thought manifests in the procedure, network training is carries on with the auto-adapted study speed and attachment momentum method function traingdx.m. The training result through passes the current capacity forecast the training performance chart performance.
该方法具有很强的学习能力和自适应性,因而具有很好的应用价值。
This method has the very strong learning capability and auto-adapted, thus has the very good application value.