The invention provides a space-time ARMA graph convolutional network traffic flow prediction method, which comprises the following steps: collecting traffic flow data recorded by a traffic network sensor, preprocessing the traffic flow data, establishing a data set, and dividing the data set into a training set, a verification set and a test set; constructing a spatial adjacency matrix according to the adjacent relation of the traffic sensors; extracting time characteristics in the traffic flow data by using a gating circulation unit; extracting spatial features by using a graph neural network based on an ARMA filter; and training the space-time ARMA graph convolutional network by using the training set, selecting model hyper-parameters by using the verification set, and evaluating the prediction precision of the model by using the test set. By constructing the space-time ARMA graph convolutional network, the space features and the time features in the traffic flow data are effectively extracted, and the prediction precision of the model is improved.
本发明提供了一种时空ARMA图卷积网络交通流预测方法,步骤如下:收集交通路网传感器记录的交通流数据,对交通流数据进行预处理,建立数据集,并将其划分为训练集、验证集和测试集;根据交通传感器相邻关系构造空间邻接矩阵;使用门控循环单元提取交通流数据中的时间特征;使用基于ARMA过滤器的图神经网络提取空间特征;使用训练集训练时空ARMA图卷积网络,使用验证集选择模型超参数,使用测试集评估模型的预测精度。本发明通过构建时空ARMA图卷积网络,有效提取了交通流数据中的空间特征和时间特征,提高了模型的预测精度。
Space-time ARMA graph convolutional network traffic flow prediction method
一种时空ARMA图卷积网络交通流预测方法
2024-07-05
Patent
Electronic Resource
Chinese
Space-time adaptive dynamic graph convolutional network traffic flow prediction method
European Patent Office | 2024
|European Patent Office | 2023
|European Patent Office | 2023
|European Patent Office | 2024
|Urban traffic flow space-time prediction scheme based on graph convolutional neural network
European Patent Office | 2021
|