Machine Learning Satellite Drag Model With Uncertainty Quantification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems for predicting space debris density in Low Earth Orbit (LEO) are inefficient and lack predictive capabilities, making collision avoidance challenging for satellites, especially during solar storms, and are not accessible to the commercial sector due to reliance on military-grade hardware and expertise.
Innovation Solution
A machine learning-based space density prediction system, HASDM-ML, uses the High Accuracy Satellite Drag Model (HASDM) database to generate accurate atmospheric drag models, incorporating Principal Component Analysis and Bayesian Deep Learning for dimensionality reduction and uncertainty quantification, enabling rapid access and commercial use of the data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine learning models are trained on the complete high-dimensional HASDM database, then prediction accuracy improves, but computational time and processing resources increase significantly
Solution Approach 1:
The patent divides the high-dimensional HASDM database into multiple subsets based on different spatial and temporal characteristics. Multiple specialized machine learning models are trained on these segmented datasets, allowing each model to focus on specific patterns and conditions. This segmentation enables faster inference while maintaining overall prediction accuracy across diverse space weather conditions.
Solution Approach 2:
The patent transforms the high-dimensional spatial and temporal data into a reduced-dimensional representation by identifying and modeling key dominant patterns. This dimensionality reduction allows the system to capture essential atmospheric density variations without processing the complete high-dimensional dataset, significantly reducing computational time while preserving prediction accuracy.
2Reliability
If complex machine learning models with uncertainty quantification are implemented, then reliability of predictions improves, but device complexity and computational resources required increase
Solution Approach 1:
The patent implements uncertainty quantification selectively rather than uniformly across all predictions. Ensemble methods and Bayesian approaches are applied to critical prediction scenarios where reliability is most important, such as during geomagnetic storms or for satellites in sensitive orbits. This partial application of complex uncertainty modeling maintains system reliability for critical cases while avoiding the full computational overhead of applying these methods universally.
3Productivity
If the HASDM database is made accessible to the commercial sector, then market application and productivity improve, but security risks and control difficulties increase
Solution Approach 1:
The patent establishes an intermediary access architecture where commercial users interact with the HASDM database and machine learning models through controlled interfaces and APIs. This intermediary layer enables commercial sector productivity by providing data access and prediction capabilities, while simultaneously maintaining security controls, authentication mechanisms, and monitoring to mitigate security risks and maintain governmental oversight.
Data Source
AI summary
The present disclosure relates to an upper-atmospheric mass density prediction model with robust and reliable uncertainty estimates in accordance with various embodiments of the present disclosure. The upper-atmospheric mass density model is developed based on the SET HASDM density database. In various embodiments, PCA is used to reduce the spatial dimension of the dataset. The input sets used to train the mass density model contains a time series for the geomagnetic indices. The mass density prediction model is trained to output a mass density map for accurately prediction trajectories of satellites. For example, a likelihood of collision associated with a given object can be determined based at least in part on the mass density map. Analysis of the mass density map along with the likelihood of collision can used to determine a trajectory for the given object in space.