Quantifying the Effects of Gyroless Flying of the Mars Express Spacecraft with Machine Learning

The gyroscopes on-board the Mars Express (MEX) spacecraft, responsible for orientation and pointing actions, are slowly decaying. On 16th April 2018, a new software was deployed to MEX which reduces gyroscope usage by 90%. In this paper, we investigate the effect of gyroless flight on the power consumption of MEX's thermal subsystems. In particular, we train predictive models from telemetry data obtained before the event on 16th April 2018 and estimate their performance on time periods before and after the event. This offers the chance to evaluate machine learning models on new situations (e.g., gyroless scenarios) while these models have been trained on scenarios that were yet unencountered. The results show that the predictive performance of the models estimated in the gyroless period is lower when compared to the estimated performance before the event. Notwithstanding, the estimated performance in both scenarios is very good and the differences are not practically significant. Hence, the models can be utilized for accurate prediction of the thermal power consumption in practical sense. Given MEX's new gyroless situation, which is likely to occur also on any other long lasting mission, we investigate the impact of the gyroscopes on the thermal power consumption and quantify the effects. Such modelling and analysis provides important insight into the spacecraft's new behavior, and still allows for planning optimization of MEX's operations, despite the radical change in the operational methodology.