Summary of Physics-informed Machine Learning Towards a Real-time Spacecraft Thermal Simulator, by Manaswin Oddiraju et al.
Physics-Informed Machine Learning Towards A Real-Time Spacecraft Thermal Simulator
by Manaswin Oddiraju, Zaki Hasnain, Saptarshi Bandyopadhyay, Eric Sunada, Souma Chowdhury
First submitted to arxiv on: 8 Jul 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Computational Engineering, Finance, and Science (cs.CE)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The proposed paper presents a novel hybrid modeling architecture that combines simplified physics models with machine learning (ML) models for thermal state estimation in complex space missions. The emerging paradigm of physics-informed machine learning (PIML) addresses the challenge of high computation required for thermal modeling, enabling designs with reduced mass and power through onboard thermal-state estimation and control. The PIML model consists of a neural network that predicts reduced nodalizations given on-orbit thermal load conditions, followed by a finite-difference model operating on this mesh to predict thermal states. The hybrid approach provides significantly better generalization than data-driven neural net models and high-fidelity finite-difference models while reducing computing cost. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary For space missions, it’s crucial to understand the temperature of surfaces in airless bodies. Currently, these calculations are too slow for real-time use on spacecraft. This paper shows how combining physics and machine learning can create faster and more accurate thermal models. The approach uses a neural network to predict simplified models that can then be used to calculate temperatures quickly. This method is better than other approaches at predicting temperatures and also takes less computer time. |
Keywords
* Artificial intelligence * Generalization * Machine learning * Neural network * Temperature
