Multi-Damage Detection in Composite Space Structures via Deep Learning.

Area of Science:

• Aerospace Engineering
• Materials Science
• Data Science

Background:

• Spacecraft rely on large, lightweight composite structures like solar panels, increasing vulnerability to orbital debris impacts.
• Detecting damage in these extensive structures is challenging due to subtle changes in global dynamics.
• Advanced structural health monitoring is crucial for ensuring the operational safety of space platforms.

Purpose of the Study:

• To develop and assess a data-driven methodology for diagnosing environmentally induced damages in composite solar arrays.
• To compare the effectiveness of accelerometers and piezoelectric sensors in identifying damage locations.
• To enhance the structural health monitoring capabilities for large space structures.

Main Methods:

• Utilized Long Short-Term Memory (LSTM) networks for damage detection in solar arrays.
• Employed finite element models to simulate damage locations in critical risk areas.
• Generated datasets from simulated attitude maneuvers using local accelerations and piezoelectric voltages.
• Trained LSTM networks to associate time-series sensor data with specific damage labels.

Main Results:

• The LSTM-based framework effectively identified the location of damaged elements in solar arrays.
• Both accelerometer and piezoelectric sensor data yielded accurate damage localization.
• The methodology proved effective even with limited measured time samples.

Conclusions:

• The data-driven LSTM approach offers a promising solution for structural health monitoring of large composite structures in space.
• The study validates the efficacy of using either accelerometers or piezoelectric sensors for damage detection.
• This method contributes to ensuring the operational safety of space platforms by enabling prompt damage identification.