The invention discloses a traffic flow prediction method and system based on a multi-factor causal fusion enhanced time sequence diagram convolutional network, and the method comprises the following steps: firstly, analyzing the causal relationship among multiple factors of a node through employing a convergent cross mapping algorithm; and a causal coefficient matrix fusing multiple traffic states is obtained to enhance the expression ability of node association features. Then, designing a spatial feature extraction module based on a gating fusion network, and improving the processing capability of non-Euclidean data; in addition, a time feature extraction module based on a time sequence convolutional network is designed to effectively capture long-term time dependence. And finally, constructing a fusion output module based on an attention mechanism to obtain a predicted value. Compared with ASTGCN, DCRNN and GWN based on attention, the prediction errors of the MCF-TGCN model are respectively reduced by 11.623%, 18.783% and 21.456% on a PEMS04 data set by taking a root mean square error as an evaluation index, and the prediction errors of the MCF-TGCN model are respectively reduced by 11.623%, 18.783% and 21.456%. On a PEMS08 data set, compared with an ASTGCN model, a DCRNN model and a GWN model, the prediction errors of the MCF-TGCN model are reduced by 14.513%, 20.766% and 29.614% respectively.
本发明公开的一种基于多因素因果融合增强时序图卷积网络的交通流预测方法及系统,包括以下步骤:首先,采用收敛交叉映射算法来分析节点多因素之间的因果关系,得到融合多种交通状态的因果系数矩阵来增强节点关联特征的表达能力。然后,设计基于门控融合网络的空间特征提取模块,提髙对非欧几里得数据的处理能力。此外,设计基于时序卷积网络的时间特征提取模块以有效捕获长期时间依赖。最后,构建基于注意力机制的融合输出模块得到预测值。以均方根误差为评价指标,在PEMS04数据集上,本发明的MCF‑TGCN模型相比基于注意力的ASTGCN、DCRNN和GWN,其预测误差分别降低11.623%、18.783%和21.456%;在PEMS08数据集上,MCF‑TGCN模型相比ASTGCN、DCRNN和GWN模型预测误差分别降低14.513%、20.766%和29.614%。
Traffic flow prediction method and system based on multi-factor causal fusion enhanced time sequence diagram convolutional network
基于多因素因果融合增强时序图卷积网络的交通流预测方法及系统
2024-09-24
Patent
Electronic Resource
Chinese
IPC: | G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS / G06F ELECTRIC DIGITAL DATA PROCESSING , Elektrische digitale Datenverarbeitung / G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen |
Attribute-enhanced perceptual sequence diagram convolutional network traffic speed prediction method
European Patent Office | 2023
|Traffic speed prediction method based on multi-space-time diagram fusion convolutional network
European Patent Office | 2023
|Traffic flow prediction method based on multi-space-time diagram convolutional network
European Patent Office | 2021
|European Patent Office | 2023
|Traffic flow prediction method based on space-time diagram convolutional network
European Patent Office | 2022
|