公路交通技术
公路交通技術
공로교통기술
TECHNOLOGY OF HIGHWAY AND TRANSPORT
2014年
2期
103-106,110
,共5页
神经网络%交通流预测%遗传算法%转移函数
神經網絡%交通流預測%遺傳算法%轉移函數
신경망락%교통류예측%유전산법%전이함수
neural network%prediction of traffic flow%genetic algorithm%transfer function
利用通道交通流参数在通道相邻路段的传动机理，分析上下游交通流特征参数的耦合关系，提出通道交通流预测技术，并基于传统数理方法建立交通流预测模型的局限性，提出神经网络模型。利用遗传算法不断优化网络权值，并通过改变网络隐含层节点数以及网络各层之间的转移函数提高网络模型的预测精度，实现通道下游交通流的实时预测。
利用通道交通流參數在通道相鄰路段的傳動機理，分析上下遊交通流特徵參數的耦閤關繫，提齣通道交通流預測技術，併基于傳統數理方法建立交通流預測模型的跼限性，提齣神經網絡模型。利用遺傳算法不斷優化網絡權值，併通過改變網絡隱含層節點數以及網絡各層之間的轉移函數提高網絡模型的預測精度，實現通道下遊交通流的實時預測。
이용통도교통류삼수재통도상린로단적전동궤리，분석상하유교통류특정삼수적우합관계，제출통도교통류예측기술，병기우전통수리방법건입교통류예측모형적국한성，제출신경망락모형。이용유전산법불단우화망락권치，병통과개변망락은함층절점수이급망락각층지간적전이함수제고망락모형적예측정도，실현통도하유교통류적실시예측。
By means of transmission mechanism of traffic flow parameters of passageways in adjacent sections of passageways, this paper analyzes the coupling relationship between characteristic parameters of upstream and downstream traffic flows, proposes prediction technique for traffic flow of passageways, and puts forward neural network model based on limitation of prediction model of traffic flow established based on traditional mathematical method. The paper continuously optimizes network weights by means of genetic algorithm, and improves the prediction precision of network model by changing the number of nodes in hidden layers of network and transfer functions between all layers of networks to realize real-time prediction of traffic flow in downstream of passageways.
