Publications
AI-powered Bi-LSTM detect structural damage in flexible satellite antennas with over 99% accuracy using onboard sensor data.
Artificial Intelligence in Space: Structural Health Monitoring for Flexible Satellite Antennas
When we think of Artificial Intelligence (AI) in space, we often picture autonomous rovers on Mars or satellites processing images in real time. Yet one of the most concrete and strategic applications is emerging in a less visible, but crucial, domain: the structural monitoring of large space antennas.
The Challenge of Monitoring Fragile Components in Orbit
As scientific missions grow more complex, satellites are equipped with increasingly wide and lightweight appendages, such as Large Mesh Reflector Models. These structures are essential for data collection, but also highly vulnerable to localized damage which, if undetected, can irreversibly compromise the mission.
How Deep Learning Detects Structural Anomalies
The answer lies in AI, specifically, Deep Learning. A recent study has shown how an advanced recurrent neural network, a Bi-LSTM (Bidirectional Long Short-Term Memory), can accurately detect breaks and partial damage in structural components of space antennas by analyzing data from a distributed network of accelerometers installed on the structure.
From Realistic Data to High-Accuracy Detection
The system simulates the satellite’s dynamic behavior during orbital maneuvers and collects realistic data, including noise. The signals are normalized, segmented at key moments, and used to train the AI model. The outcome? In some cases, the model achieves over 99% accuracy in distinguishing between intact, damaged, or broken antenna states.
Sensor Placement as an AI-Driven Design Variable
One key insight concerns sensor placement: damage near the antenna’s attachment point to the satellite body is more easily detectable. For more peripheral injuries, such as those on the support truss, the AI has difficulty distinguishing between healthy and compromised conditions. This opens up new possibilities: using AI not just for data analysis, but also for optimizing sensor layout during the design phase.
Toward Autonomous, Resilient Satellite Architectures
In an era where every satellite component must deliver maximum reliability and resilience, intelligent monitoring systems like this may become an integral part of future space architectures. No longer limited to ground-based control or occasional inspections, they act as a true “orbital sixth sense,” capable of recognizing even the slightest sign of failure in real time. Because in space, even the smallest damage can make a huge difference.
Authors
F. Angeletti, P. Iannelli, P. Gasbarri, M. Panella, A. Rosato
