FDSA-STG: Fully Dynamic Self-Attention Spatio-Temporal Graph Networks for Intelligent Traffic Flow Prediction

Duan, Youxiang; Chen, Ning; Shen, Shigen; Zhang, Peiying; Qu, Youyang; Yu, Shui

doi:10.1109/tvt.2022.3178094

Cited by 28 publications

(9 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, in different scenarios, the demands placed on the grid are also different [7]. For example, in scenarios such as fault identification and prediction, a large amount of AI computing and network bandwidth energy is required for real-time perception and accurate prediction [8]. However, these energy sources are not abundant at this stage, so how reasonably to allocate AI computing and bandwidth energy has become a difficulty in the development of smart grids.…”

Section: B Pose Challengesmentioning

confidence: 99%

Multi-Target-Aware Energy Orchestration Modeling for Grid 2.0: A Network Virtualization Approach

et al. 2023

Self Cite

View full text Add to dashboard Cite

Smart Grid -Grid 2.0, is a promising grid scenario due to its low cost, high reliability, and high security. However, the high-intensity complex task interaction and the differentiated quality of service (QoS) requirements pose challenges to the energy orchestration of the dynamic and time-varying smart grid. In addition, Network Virtualization (NV) technology is widely utilized to decouple physical network resources. In this work, based on Virtual Network Embedding (VNE), we propose an energy orchestration strategy to better support the operation and maintenance of Grid 2.0. Specifically, a series of constraint indicators for energy orchestration is first formulated. Second, based on the advantages of deep reinforcement learning (DRL), a five-layer architecture energy orchestration model is proposed. It frequently extracts the grid energy matrix as the state input, adopts two-stage actions to reasonably orchestrate the grid energy, and combines the reward mechanism of DRL to effectively interact with the environment. Finally, we conduct experiments in a simulation environment and demonstrate the advantages of the proposed strategy through analysis and comparison.

show abstract

Section: B Pose Challengesmentioning

confidence: 99%

Multi-Target-Aware Energy Orchestration Modeling for Grid 2.0: A Network Virtualization Approach

et al. 2023

Self Cite

View full text Add to dashboard Cite

show abstract

“…Chen et al [37] developed a prediction model called a location graph convolutional network (location-GCN) by designing a novel graph convolution to construct a dynamic adjacency matrix, and combined it with LSTM. Although these models have achieved some success in traffic flow prediction, they have limitations in capturing the important spatio-temporal features [38].…”

Section: Spatio-temporal Feature Extractionmentioning

confidence: 99%

“…As shown in Figure 4, the decoder focuses on modelling both the seasonal periodic and trend periodic part, resulting in a more complex structure in contrast to the encoder. There are M layers in the decoder, and the overall equations for the l th layer of the decoder are summarized as Equation (38). Details are shown in Equations ( 39) to (42).…”

Section: Decodermentioning

confidence: 99%

HD‐Net: A hybrid dynamic spatio‐temporal network for traffic flow prediction

Liu,

Wang,

Liu

et al. 2023

IET Intelligent Trans Sys

View full text Add to dashboard Cite

Accurately predicting traffic flow is crucial for intelligent transportation systems (ITS). In recent years, many deep learning‐based prediction models have been widely applied in traffic flow prediction, and various spatio‐temporal networks have been proposed. However, most of the existing models follow a general technical route to extract the spatio‐temporal features, which lack the capacity of extracting the important historical information with the high spatial and temporal correlations dynamically and deeply. How to develop a well‐performance traffic flow prediction model for a complex transportation network is still facing some challenges. In this paper, a hybrid dynamic spatio‐temporal network (HD‐Net) for traffic flow prediction is proposed. In HD‐Net, the authors first extract the dynamic spatio‐temporal features using dynamic graph convolution and bidirectional gate recurrent uni (BiGRU). Subsequently, the authors extract the important features with high spatial and temporal correlations from the obtained dynamic spatio‐temporal features using an auto‐correlation mechanism from a local perspective, and self‐attention mechanism from a global perspective, respectively. Extensive experiments have been conducted on two real‐world traffic datasets. The experimental results demonstrate that the proposed HD‐Net outperforms the baselines in the field of capturing the dynamic and important spatio‐temporal features with high correlations.

show abstract

“…Jin et al [28] put forward GAN-Based Short-Term Link traffic prediction under a parallel learning framework (PL-WGAN) for urban networks which added spatial-temporal attention mechanism to adjust the importance of different temporal and spatial contexts. In addition, Duan et al [29] applied fully dynamic self-attention spatiotemporal graph network (FDSA-STG) by improving the attention mechanism using graph attention networks (GATs). This model jointly modified the GATs and the self-attention mechanism that fully dynamically focused and integrated spatial, temporal and periodic correlations.…”

Section: B Attention Mechanism In Time Series Data Predictionmentioning

confidence: 99%

Dynamic Spatial-Temporal Graph Attention Graph Convolutional Network for Short-Term Traffic Flow Forecasting

Tang

Sun

2020

2020 IEEE International Symposium on Circuits and Systems (ISCAS)

View full text Add to dashboard Cite

Accurate traffic prediction in real time plays an important role in Intelligent Transportation System (ITS) and travel navigation guidance. There have been many attempts to predict short-term traffic status which consider the spatial and temporal dependencies of traffic information such as temporal graph convolutional network (T-GCN) model and convolutional long short-term memory (Conv-LSTM) model. However, most existing methods use simple adjacent matrix consisting of 0 and 1 to capture the spatial dependence which can not meticulously describe the urban road network topological structure and the law of dynamic change with time. In order to tackle the problem, this paper proposes a dynamic temporal self-attention graph convolutional network (DT-SGN) model which considers the adjacent matrix as a trainable attention score matrix and adapts network parameters to different inputs. Specially, selfattention graph convolutional network (SGN) is chosen to capture the spatial dependence and the dynamic gated recurrent unit (Dynamic-GRU) is chosen to capture temporal dependence and learn dynamic changes of input data. Experiments demonstrate the superiority of our method over state-of-art model-driven model and data-driven models on real-world traffic datasets.

show abstract

FDSA-STG: Fully Dynamic Self-Attention Spatio-Temporal Graph Networks for Intelligent Traffic Flow Prediction

Cited by 28 publications

References 37 publications

Multi-Target-Aware Energy Orchestration Modeling for Grid 2.0: A Network Virtualization Approach

Multi-Target-Aware Energy Orchestration Modeling for Grid 2.0: A Network Virtualization Approach

HD‐Net: A hybrid dynamic spatio‐temporal network for traffic flow prediction

Dynamic Spatial-Temporal Graph Attention Graph Convolutional Network for Short-Term Traffic Flow Forecasting

Contact Info

Product

Resources

About