Space Debris Detection and Trajectory Prediction

Project Overview

Space debris poses a significant threat to satellites, space missions, and other orbital assets. This project aims to tackle two critical challenges in space debris management:

1. Debris Detection: Identifying space debris using machine learning models based on orbital parameters extracted from TLE (Two-Line Element) data.
2. Trajectory Prediction: Developing a system to predict debris trajectories using time-series analysis of orbital data.

By addressing these challenges, this project contributes to safer and more efficient management of Earth's orbital space.

---

Objectives

1. Debris Detection:

   • Classify orbital objects as either active satellites or debris based on their orbital characteristics.
   • Achieved a detection accuracy of over 98% using Random Forest Classifiers.

2. Trajectory Prediction:

   • Build time-series models to predict future debris positions based on historical orbital parameters.

---

Data Sources

The data for this project is derived from CelesTrak and other open sources:

• Debris TLE Data:
  • COSMOS 1408 Debris: Link
  • FENGYUN 1C Debris: Link
  • IRIDIUM 33 Debris: Link
  • COSMOS 2251 Debris: Link
• Active Satellite TLE Data:
  • Active Satellites

---

Methodology

1. Debris Detection

• Extracted orbital parameters from TLE data:
  • Semi-major axis, eccentricity, inclination, RAAN, argument of perigee, mean anomaly, mean motion, and derived features (e.g., velocity, altitude).
• Trained machine learning models (Random Forest, Gradient Boosting) to classify objects as satellites or debris.
• Achieved:
  • Accuracy: 99.62% (on validation data)
  • Recall: 98% (balanced dataset)

2. Trajectory Prediction

• Created time-series data from multiple TLE snapshots for each debris object.
• Planned future models:
  • Use LSTMs, GRUs, or Kalman filters to predict debris trajectories.
  • Include visualization of orbital paths.

---

Key Features

• Derived Features:
  • Orbital period, velocity magnitude, specific orbital energy.
• Balanced vs. Unbalanced Data:
  • Experimented with class balancing to optimize precision and recall.
• Evaluation:
  • Metrics: Accuracy, Precision, Recall, F1-score, OOB Score.

---

Results

Detection

• Unbalanced Data:
  • Accuracy: 0.9962
  • Precision (Class 1 - Debris): 97%
  • Recall (Class 1 - Debris): 97%
• Balanced Data:
  • Accuracy: 0.9856
  • Precision (Class 1 - Debris): 95%
  • Recall (Class 1 - Debris): 98%

Trajectory Prediction

• In Progress: Data collection for time-series modeling underway.

---

Next Steps

1. Expand trajectory prediction using hourly TLE snapshots.
2. Implement advanced predictive models like LSTMs or Kalman Filters.
3. Visualize trajectories in 3D to demonstrate debris behavior.

---

Usage

1. Setup

• Clone the repository:

  git clone https://github.com/your-repo/space-debris-detection.git