Orbit State Prediction Using Multi-Model Ensemble Density Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for predicting the orbital state of objects in low Earth orbit lack accurate uncertainty quantification, particularly due to uncertainties in atmospheric density, which are not adequately addressed by existing models, leading to inaccuracies in collision probability assessments and space traffic management.
Innovation Solution
The development of a multi-model ensemble approach that combines the predictions of multiple atmospheric density models, such as HASDM-ML-DP, CHAMP-ML-DP, and MSIS-UQ-DP, using machine learning techniques to provide a more comprehensive and realistic estimation of orbital state uncertainties through ensemble modeling and filtering methods like the Consider Covariance Sigma Point filter.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If single atmospheric density models are used for orbit prediction, then computational simplicity is maintained, but uncertainty quantification accuracy deteriorates
Solution Approach 1:
The patent combines multiple atmospheric density models (HASDM-ML-DP, CHAMP-ML-DP, MSIS-UQ-DP) into an ensemble system that integrates their predictions through machine learning techniques. This merging approach captures the strengths of individual models while quantifying uncertainties more accurately, resolving the contradiction between model complexity and prediction accuracy by showing that the ensemble benefits from diversified model inputs without requiring excessive computational resources.
Solution Approach 2:
The patent creates a composite modeling approach by integrating multiple density models with different characteristics (machine learning-based models and physics-based models). Each model contributes its unique strengths, and the ensemble combines them into a unified prediction system that achieves superior uncertainty quantification compared to any single model, analogous to how composite materials combine different materials to achieve enhanced properties.
2Measurement precision
If multiple density prediction models are combined using ensemble approach, then orbital state prediction accuracy is improved, but computational resources required increase
Solution Approach 1:
The patent applies partial action by selectively combining a limited number of density prediction models (three specific models) rather than using all available models. This selective approach achieves sufficient accuracy improvement while avoiding the excessive computational burden that would result from incorporating too many models, thus resolving the contradiction between accuracy and computational cost.
3Reliability
If traditional orbit propagation methods are used, then computational speed is maintained, but collision probability assessment reliability deteriorates
Solution Approach 1:
The patent replaces traditional mechanical orbit propagation methods with a machine learning-based ensemble approach. The machine learning models learn complex atmospheric density patterns from training data and provide more reliable uncertainty quantification, substituting the traditional physics-based propagation with a data-driven approach that achieves both reliability and computational efficiency.
Solution Approach 2:
The patent changes the fundamental parameters of the orbit prediction system by transitioning from deterministic single-model propagation to probabilistic ensemble propagation. This involves changing from fixed orbital parameters to probability distributions of orbital parameters, enabling more reliable collision probability assessments while maintaining computational tractability through efficient ensemble methods.
Data Source
AI summary
The present disclosure relates to the quantification and propagation of orbital state uncertainties and accurately determining an orbit state of an orbiting object using multi-model ensemble analysis. Input data (e.g., solar indices, geomagnetic indices, space weather parameters, temporal parameters, etc.) associated with the solar environment and orbiting object can be provided as inputs to multiple trained density prediction models. The trained density prediction models can be configured to output atmospheric density data associated with the orbiting object (e.g., satellite). Using orbit propagation for the respective atmospheric density data, orbit data (e.g., position, velocity) can be predicted. The predicted orbit data associated with the multiple density prediction models can then be analyzed in an ensemble approach to accurately predict the orbit state of the orbiting object.