首页|基于多源数据融合的高速公路路网短时交通流参数实时预测

基于多源数据融合的高速公路路网短时交通流参数实时预测

扫码查看
原文链接
  • 万方数据
  • 维普
针对高速公路路网交通数据中存在大量噪声数据与缺失数据,数据完整度较低,导致预测精度下降的问题,提出一种基于多源数据融合的高速公路路网短时交通流参数实时预测方法.采用小波分析阈值法对高速公路路网交通数据进行去噪处理,在最小二乘支持向量机基础上采用组合阈值填补方法填补交通数据序列中存在的缺失数据,提高交通数据的完整度.结合小波神经网络和遗传算法建立短时交通流参数预测模型,采用遗传-小波神经网络处理多源检测器采集到的交通流参数,通过最小二乘动态加权融合算法融合多个检测器的交通流参数,将交通流参数输入预测模型中,得到高速公路路网短时交通流参数实时预测结果.实验结果表明,采用本文方法处理后的交通数据序列中不存在缺失数据,数据完整度较高,且所得预测结果与实际车流量变化曲线较为贴近,预测精度高,可以广泛应用在交通流预测领域.
High-speed highway road network short-term traffic flow parameters based on multi-source data fusion prediction
There are a lot of noise data and missing data in the traffic data of expressway network,and the data integrity is not high,leading to the decline of prediction accuracy.A real-time prediction method for short-term traffic flow parameters of expressway network based on multi-source data fusion is proposed.The wavelet analysis threshold method is used to denoise the traffic data of the expressway network.Based on the least squares support vector machine,the combined threshold filling method is used to fill in the missing data in the traffic data sequence to improve the integrity of the traffic data.The short-term traffic flow parameter prediction model is established by combining wavelet neural network and genetic algorithm.The traffic flow parameters collected by multi-source detectors are processed by genetic wavelet neural network.The traffic flow parameters of multiple detectors are fused by the least squares dynamic weighted fusion algorithm.The traffic flow parameters are input into the prediction model to obtain the real-time prediction results of short-term traffic flow parameters of expressway network.The experimental results show that there is no missing data in the traffic data series processed by the proposed method,the data integrity is high,and the predicted results are close to the actual vehicle flow change curve,with high prediction accuracy,which can be widely used in the field of traffic flow prediction.

multi-source data fusionexpressway networkshort term traffic flowparameter predictionwavelet analysis threshold methodleast squares support vector machinegenetic wavelet neural network

高海龙、徐一博、刘坤、李春阳、卢晓煜

展开 >

交通运输部公路科学研究院安全中心,北京 100088

港珠澳大桥管理局,广东珠海 519060

多源数据融合 高速公路路网 短时交通流 参数预测 小波分析阈值法 最小二乘支持向量机 遗传-小波神经网络

国家重点研发计划

2019YFB1600703

2024

10.13229/j.cnki.jdxbgxb.20221272

吉林大学学报(工学版)
吉林大学

吉林大学学报(工学版)

CSTPCD北大核心
影响因子：0.792
ISSN：1671-5497
年,卷(期)：2024.54(1)
  • 15