AST-GIN: Attribute-Augmented Spatiotemporal Graph Informer Network for Electric Vehicle Charging Station Availability...

Area of Science:

• Intelligent Transportation Systems
• Machine Learning
• Environmental Science

Background:

• Forecasting electric vehicle (EV) charging demand and station availability presents a significant challenge in intelligent transportation systems.
• Existing deep learning methods often overlook complex external factors like points of interest (POIs) and weather, limiting prediction accuracy.
• Accurate EV station availability predictions are essential for mitigating range anxiety and optimizing charging behaviors.

Purpose of the Study:

• To enhance the accuracy and interpretability of EV charging station availability predictions.
• To develop a novel deep learning model that incorporates external factors influencing spatiotemporal transportation data.
• To evaluate the proposed model's performance against existing baseline methods.

Main Methods:

• Introduction of the Attribute-Augmented Spatiotemporal Graph Informer (AST-GIN) model.
• Integration of Graph Convolutional Network (GCN) and Informer layers to capture internal and external spatiotemporal dependencies.
• Modeling external factors as dynamic attributes using an attribute-augmented encoder for enhanced training.

Main Results:

• The AST-GIN model demonstrated superior performance in predicting EV charging station availability.
• The study confirmed the significant influence of external factors on prediction accuracy across various time horizons.
• Experimental results on Dundee City data validated the model's effectiveness compared to baseline approaches.

Conclusions:

• The AST-GIN model offers a significant advancement in EV charging station availability forecasting.
• Incorporating external factors through attribute augmentation substantially improves prediction accuracy and interpretability.
• The findings support the integration of advanced deep learning techniques for more reliable intelligent transportation systems.