A
Digital twin
Reduced-order FE model of the structure, updated from measurements.
We know where the satellite is. Hevelia will tell you what state it’s in.
Visualisation
Project
From monitoring a satellite’s operational parameters to monitoring its actual mechanical condition.
Hevelia is a research and development project that aims to enable reliable assessment of the mechanical condition of large deployable satellite structures from limited indirect measurements.
Modern satellites increasingly rely on multi-segment solar arrays, deployable reflector antennas and lightweight booms. They must be light and compact, so they have low stiffness, complex dynamics, and respond to manoeuvres, temperature changes and wear in their mechanical joints.
Today an operator knows a great deal about a satellite’s attitude, temperatures and subsystems. Far less about whether its structure behaves as designed. Hevelia aims to close that gap.
Problem
Structures in orbit cannot simply be measured.
- Few measurements
- Coupled dynamics
- Thermal effects
- Uncertain parameters
- Limited observability
- No failure data
Technology
Physics-Informed Structural Intelligence
We read structural condition from the response of the whole satellite, through a physical model.
B
Sparse data fusion
Gyroscopes, attitude, temperatures, optionally cameras.
C
Physics-informed AI
Inverse identification, Bayesian estimation, machine learning.
D
Explicit uncertainty
When data are insufficient, we show a range of possible explanations instead of guessing.
Application
First target: deployable solar arrays.
We are analysing how to detect changes in panel joint dynamics from attitude determination and control system (ADCS) data.
- gyroscopes
- reaction wheels
- attitude
- temperatures
- 01
On the ground
qualification and acceptance testing
- 02
After deployment
assessment immediately after deployment in orbit
- 03
In operation
monitoring throughout the mission
Results
Hevelia Structural Diagnostics Engine
- sensitivity to stiffness changes of 20% or more
- min. 90%
- false alarms
- max. 5%
- correct localisation of changes
- min. 80%
- of full-instrumentation sensitivity with limited measurements
- min. 90%
- fewer thermally induced false alarms
- min. 30%
- validation on a new configuration
- min. 1
Who it’s for
- Satellite platform manufacturers
- Deployable structure suppliers
- System integrators
- Satellite operators
Space Intelligence.
Physically grounded knowledge of a satellite’s condition.
FAQ
01 Do additional sensors need to be mounted on the satellite?
In the baseline approach, no. We use data the satellite already collects: gyroscopes, attitude and temperatures.
02 How does Hevelia differ from conventional SHM?
Conventional SHM requires sensors on the component itself. Hevelia infers condition indirectly: from the response of the whole platform, through a physical model.
03 Is this just an AI model?
No. Machine learning operates within physical constraints, together with inverse identification and Bayesian estimation.
04 What happens when there is not enough data?
The system says so explicitly and indicates the range of possible explanations.
05 How do you check that the diagnosis is correct?
On a laboratory test rig we introduce known changes and measure them with an independent reference configuration.
06 When will the technology be available?
The project will run for 30 months and aims to bring the technology to TRL 4, with a path to TRL 5.
Building structures
or satellite platforms?
We are looking for industrial and research partners, and operators. Let’s talk about data and integration.
[TO BE COMPLETED: e-mail]
Hevelia Sp. z o.o. · Sopot