摘要
本文提出一种结合小波分析和SOM(自组织特征映射)神经网络的交通事件检测算法。论文首先利用小波分析检测交通流初始信号的奇异性,然后将小波系数作为SOM网络的输入,对信号奇异点进行分类,再根据分类标准判断交通流状态,并运用M atlab进行了仿真分析。结果表明提出的交通事件检测算法利用较少样本数据即可快速实现交通事件检测,具有潜在的应用价值。
This paper proposes an algorithm for traffic incidents detection based on wavelet analysis and SOM （Self-organize feature map） network. First Wavelet analysis is applied to find out the odd points of original traffic flow signals. The wavelet coefficient is referred as the input of SOM network and the network classifies the odd points. Then the traffic flow status is judged according to the class parameters. Finally we stimulate with Matlab and find out the ,algorithm can detect incidents fleetly and give a few samples. The algorithm has potential applied value.
出处
《系统工程》
CSCD
北大核心
2006年第10期100-104,共5页
Systems Engineering
基金
交通部应用基础研究项目(200431982515)
国家自然科学基金资助项目(50348027)
