Stepwise Identification Method of Thermal Load for Box Structure Based on Deep Learning.

Area of Science:

• Aerospace Engineering
• Structural Mechanics
• Computational Science

Background:

• Accurate thermal load identification is vital for spacecraft on-orbit monitoring.
• Limited measurement points pose challenges for traditional methods.

Purpose of the Study:

• To develop a stepwise deep learning method for rapid and accurate identification of structural thermal loads on spacecraft.
• To enhance the mapping of local responses to overall thermal loads in box structures.

Main Methods:

• A cascade of deep learning models is employed, segmenting the structure into subregions to progressively narrow the solution domain.
• Boundary conditions are incorporated into the deep learning models to improve generalization.
• A large simulated dataset with varied load positions and intensities is utilized, with structural displacements as input and thermal load parameters as output.

Main Results:

• The proposed stepwise identification method reduces thermal load parameter identification error by over 45% compared to a single deep learning network.
• The method accurately identifies both the location and magnitude of thermal loads.

Conclusions:

• The developed deep learning approach offers a promising solution for optimizing spacecraft structure design and analysis.
• This method contributes to improved performance and reliability in future space missions through enhanced thermal load identification.