Abstract: Traffic data acquired by different types of traffic detectors contain different traffic information and traffic flow prediction plays an important role in traffic management and control. According to this, multi-source traffic data dynamic weighted fusion and short-term traffic flow forecasting were carried out. On the premise of synthetically analyzing the characteristics of multi-source data and the superiority of data fusion, combining with the advantages of global search of genetic algorithm and adaptive learning of wavelet neural network, a short-term traffic flow forecasting model based on multi-source data fusion and genetic wavelet neural network (GA-WNN) was put forward. The results of case study show that: the traffic data fusion method based on GA-WNN has more advantages than others, and the prediction accuracy of multi-source data fusion is better than that of single data source short-term traffic flow prediction sequence, which provides more precise and comprehensive traffic information for the decision-making of traffic managers and the choice of path for traffic travelers.

Key words: traffic engineering, multi-source data fusion, short-term traffic flow forecasting, genetic algorithm, wavelet neural network

摘要： 不同类型交通检测器所获取的交通数据中包含了不同的交通信息，交通流预测在交通管理与控制中具有重要作用，基于此，进行了多源交通数据动态加权融合和短时交通流预测。在综合分析多源数据特性及其融合优势的前提下结合遗传算法的全局搜索及小波神经网络的自适应学习优点，提出了基于多源数据融合与遗传-小波神经网络(GA-WNN)的短时交通流预测模型。通过实例验证分析，基于GA-WNN的交通数据融合方法比其他方法更有优势；同时，多源数据融合的预测精度优于单一数据源的短时交通流预测序列，从而能为交通管理者的判断决策与交通出行者的路径选择提供更准确、全面的交通信息。

关键词: 交通工程, 多源数据融合, 短时交通流预测, 遗传算法, 小波神经网络