WeTRaC: Scalable EV Charging Demand Forecasting for Heavy-Duty Fleets – Data Repository

This repository contains the data used to develop and validate the WeTRaC framework, a model for predicting heavy-duty electric vehicle (EV) charging demands under diverse operating conditions. Three primary datasets are included, each corresponding to specific experiments or validation steps discussed in our paper: 1. Graphs.zip This compressed folder contains graph-structured data used for training the WeTRaC model. Each subfolder is named after a particular city or region; inside, you will find numerous .pt files, each representing a distinct route. The naming convention follows this format: [CityName]-[Identifier1]-[Identifier2]_[InitialSOC]_[InitialTime]_v0_1.pt Use PyTorch’s torch.load() function in Python to import these graph files. Refer to Table 5 in the paper for a detailed description of the graph attributes, node features, and edge labels. Note: aux_hvac_sum corresponds to aux_power. SOC and t in the files correspond to abs_SOC and abs_t mentioned in the paper. 2. Field-Validation.zip This folder includes the data used for real-world truck route validation: Lines/: Contains CSV files describing the road network segments. Each line entry has columns for segment IDs, length, speed, geometric attributes, and other parameters. Truck-Measurements/: Provides CSV files with timestamps, truck SOC, GPS coordinates, and other sensor measurements, where each row corresponds to a specific time-stamped reading. 3, Virtual-Corridor-Lines.zip Folders in this archive correspond to individual “virtual corridors” examined in the paper, simulating possible route layouts and charging infrastructure. Each folder contains a single route file with CSV-formatted rows. These corridor files enable simulation-based analysis of charging demand under hypothetical route conditions. Citation: If you use this dataset, please cite our paper: