Spacecraft real-time thermal simulation using artificial neural networks

Abstract

Spacecraft Operational Simulators are mainly used for training satellite operators, to test the ground control system, and the evaluation of operational and onboard procedures before their execution in the real satellite. To achieve these objectives, all the internal models of the Operational Simulator must provide information in real time. Traditionally, the thermal simulation in these simulators is accomplished through interpolation on a set of pre-calculated scenarios or by the integration of a very simplified mathematical model. Both approaches, however, have limitations in both fidelity and runtime. In order to overcome these limitations, in this work it is proposed to build the thermal model of a Spacecraft Operational Simulator using artificial neural networks. This approach was applied to the Amazonia-1, a medium size satellite currently being developed at the Brazilian National Institute for Space Research. The obtained results show increased fidelity and an extremely short execution time, evidencing the potential of the approach to simulate satellite thermal behavior in Operational Simulators.

Including AI experiments onboard the UPMSat-3 satellite mission

“…This amount of data may seem to be scarce. However, previous studies like [13] achieved satisfactory results in ML thermal modeling with 707 samples and a longer sampling period of 60 seconds.…”