科研速览 · Science Skim继续刷下去 · Keep skimming →
◇ Harvard Dataverse2026-07-31· Computer science

Team Digantara Code Implementation

Bhargav M. Joshi

原始摘要(英文原文)· Original abstract
This repository contains the codebase developed for the MIT ARCLab STORM-AI Challenge 2025, focused on forecasting thermospheric orbit-averaged neutral mass density using AI/ML methods. Accurate density prediction is critical for satellite drag modeling, orbit propagation, and space situational awareness—especially during geomagnetically active periods. The approach integrates physics-informed feature engineering with state-of-the-art machine learning techniques, including: 🌡️ XGBoost-based Conditional Models for initial density estimation 🔄 TSLANet (Time Series Lightweight Adaptive Network) for 72-hour density trend forecasting ⚙️ Tools for data preprocessing, scaling, and feature extraction based on OMNI2 and ESA SWARM datasets 📊 Benchmarking against truth data using custom evaluation metrics such as OD-RMSE This repository is organized for transparency and reproducibility, and includes: Training scripts for decision tree and neural models Inference pipelines for both Phase 1 and Phase 1.2 datasets Environment setup (.yml) and dependencies Example visualizations and evaluation outputs We hope this project helps advance data-driven techniques in thermospheric modeling and supports broader applications in space weather forecasting and satellite operations.
读原文 · Read the paper ↗

AI 追问PRO

登录后使用 AI 追问

讨论区

登录后参与讨论

相关论文 · Related

Team Digantara Code Implementation — 科研速览 Science Skim