Short-term traffic flow detection based on wavelet
基于小波分析的交通流量异常数据检测*

For the bottleneck that traffic congestion had become a constraint to the development of urban economy and society, this paper proposed a solution to urban traffic congestion, with respect of the performance of traffic control system. By using the wavelet analysis to separate the high and low frequency components of traffic stream data, obtained the difference between the original signal and the reconstructed signal, found then outliers underlying in the data in combination with the least square method. It is confirmed from a number of experiments in actual roads that abnormal data in traffic flow can be detected precisely, which makes it possible to lower the rate of erroneous judgement and missing considerably and provide reliable data for practical traffic signal control system.