Abstract:Traffic flow prediction, which provides dynamic data support for intelligent transportation systems, has always been a topic of great interest. Accurate prediction of traffic flow is vital for optimizing traffic management and planning and improving traffic efficiency. Existing traffic flow prediction models usually consider only a single influencing factor when modeling spatial-temporal correlation and fail to comprehensively consider the multiscale factors corresponding to each module, resulting in incomplete spatial-temporal modeling and thus affecting prediction accuracy. For this reason, this paper proposes a spatial-temporal multifactor fusion graph neural network (STMFGNN), which aims to characterize spatial-temporal features more comprehensively. First, in the spatial feature extraction module, we parallelly utilize dynamic similarity graphs and static adjacency graphs to perform graph convolution and introduce the gated fusion module to self-learn the dynamic influence weight so that our model can integrate the information of global hidden knowledge and local prior knowledge and capture the multiscale spatial dependencies between nodes. Second, the model combines gated tanh unit convolution with a multireceptive field and gated recurrent units in the temporal feature extraction module. Similarly, it utilizes the gated fusion module to adaptively and dynamically adjust the importance weights of the two components. This enables the acquisition of multiscale temporal dependency information at both short-term and long-term levels. In addition, we employ an improved generative adversarial imputation network for incomplete traffic flow data. The experimental results on four real-world datasets show that our proposed method consistently outperforms other baseline models, achieving state-of-the-art performance. The key source code and data are available at https://github.com/Keaiii3/STMFGNN.

MAPredRNN: multi-attention predictive RNN for traffic flow prediction by dynamic spatio-temporal data fusion

Spatial–temporal combination and multi-head flow-attention network for traffic flow prediction

Features Injected Recurrent Neural Networks for Short-Term Traffic Speed Prediction.

Convolutional Long-Short Term Memory Network with Multi-Head Attention Mechanism for Traffic Flow Prediction

Fuzzy Hybrid Framework with Dynamic Weights for Short‐term Traffic Flow Prediction by Mining Spatio‐temporal Correlations

Dynamic Spatial-Temporal Representation Learning for Traffic Flow Prediction

A multi-feature spatial–temporal fusion network for traffic flow prediction

Attention Mechanism With Spatial-Temporal Joint Model for Traffic Flow Speed Prediction

Spatial-temporal multi-factor fusion graph neural network for traffic prediction

Dynamic Graph Convolutional Network with Attention Fusion for Traffic Flow Prediction

A Hybrid Deep Learning Model With Attention-Based Conv-LSTM Networks for Short-Term Traffic Flow Prediction

Deep spatio-temporal neural network based on interactive attention for traffic flow prediction

Dynamic multi-scale spatial-temporal graph convolutional network for traffic flow prediction

Multi-Step Traffic Flow Prediction Using Recurrent Neural Network

A time-dependent attention convolutional LSTM method for traffic flow prediction

Traffic Flow Prediction via Spatial Temporal Graph Neural Network

Spatial-Temporal Traffic Flow Prediction with Fusion Graph Convolution Network and Enhanced Gated Recurrent Units

A Multifeature Fusion Short-Term Traffic Flow Prediction Model Based on Deep Learnings

A Multi-Scale Residual Graph Convolution Network with hierarchical attention for predicting traffic flow in urban mobility

An Effective Dynamic Spatio-temporal Framework with Multi-Source Information for Traffic Prediction

SPRNN: A spatial–temporal recurrent neural network for crowd flow prediction