The invention provides a traffic flow prediction method based on big data, and the method comprises the steps: dividing a monitoring region into M sample road sections through constructing a traffic flow prediction model, and determining historical traffic data (including a corresponding traffic flow label) corresponding to each sample road section through employing a historical traffic data set; determining a real-time feature and a global feature corresponding to each sample road section; carrying out feature fusion to obtain N training features corresponding to each sample road section, and forming a training data set containing MN training features; and constructing a GRU model, and carrying out training and testing to obtain a trained traffic flow prediction model. The real-time features and the global features are combined, and the GRU model is adopted for prediction, so that the accuracy of traffic flow prediction can be improved, and the prediction result is more reliable.
本申请提供一种基于大数据的交通流量预测方法,通过构建交通流量预测模型,对监测区域进行划分形成M个样本路段,利用历史交通数据集确定出每个样本路段对应的历史交通数据(包含对应的交通流量标签);确定出每个样本路段对应的实时特征和全局特征;再进行特征融合,得到每个样本路段对应的N个训练特征,形成包含MN个训练特征的训练数据集;构建GRU模型,进行训练和测试,得到训练好的交通流量预测模型。通过结合实时特征和全局特征,并采用GRU模型进行预测,能够提高交通流量预测的准确性,使预测结果更加可靠。
Traffic flow prediction method based on big data
基于大数据的交通流量预测方法
24.05.2024
Patent
Elektronische Ressource
Chinesisch
AIS data-based channel traffic flow prediction method
Europäisches Patentamt | 2022
|Traffic flow prediction system based on traffic big data
Europäisches Patentamt | 2023
|Traffic flow prediction method for traffic flow spatio-temporal data information
Europäisches Patentamt | 2024
|Traffic speed prediction method based on traffic flow data fusion and system
Europäisches Patentamt | 2021
|Traffic Flow Prediction Method Based on Big Data Mining
IEEE | 2022
|