Real World and Synthetic Datasets from "Predicting Extreme Weather-Induced Outages in Distribution Networks Using Graph-Based Neural Network Models"

This contains synthetically generated distribution networks and weather data, as well as real world weather data that was use to trained the model in the paper "Predicting Extreme Weather-Induced Outages in Distribution Networks Using Graph-Based Neural Network Models".<br>Six synthetic distribution networks were generated, using NREL Shift in Orange County, FL and consist of mainly urban topologies. Synthetic weather events were generated using uniform distributions for weather features to explore a range of potential conditions. Wind speeds were sampled from 0 to 120 mph, and precipitation levels were sampled from 0 to 70 mm. Real-world weather events were selected from the NOAA Storm Events Database. The selected events included tropical storms, flash floods, and high winds in Orlando, FL.

本数据集包含合成生成的配电网（distribution networks）与气象数据，以及用于训练论文"基于图神经网络模型（Graph-Based Neural Network Models）预测配电网极端天气引发的停电事故"中所用模型的真实气象数据。研究团队基于佛罗里达州奥兰治县的NREL Shift工具生成了6个合成配电网，其拓扑结构以城区为主。为探索各类潜在气象场景，研究人员通过对气象特征采用均匀分布的方式生成了合成气象事件，其中风速采样范围为0至120英里每小时，降水量采样范围为0至70毫米。真实气象事件数据集源自美国国家海洋和大气管理局（National Oceanic and Atmospheric Administration, NOAA）风暴事件数据库，所选事件涵盖佛罗里达州奥兰多市发生的热带风暴、山洪暴发与强风天气。