Open data from Mars – Machine learning competition using Mars Express data #MarsExpressPower
UPDATE 31 March: The formal call for participation is now live via the main ESA website.
The Mars Express control team working at ESA’s ESOC operations centre in Darmstadt, Germany, works every day to keep the orbiter functioning and in good health. The telemetry received from Mars Express brings to the spacecraft controllers awareness on the states of all the different internal subsystems (thermal, propulsion, power management, radio communication, attitude control, etc.); with tens of thousands of parameters and the (welcome!) longevity of the spacecraft’s subsystems, the maintenance work done by the engineers has become more demanding. In addition, modelling the power and thermal status of the spacecraft is a challenge at any time, yet this is an extremely important factor in managing and maintaining Mars Express.
With recent developments in data mining and machine learning, the Mars Express mission control team have provided a lot of data to the Data Analytics experts on the Advanced Mission Concepts team at ESOC. This data, primarily related to Mars Express’ power and thermal status over the years, will be released in a machine-learning competition that will invite data miners and analysts – or any interested person – to find valuable information that can be shared with the mission control team and indeed with the wider data-mining community (watch this space for news – Ed.)
The competition participants will be provided with three Martian years’ worth historical data to ‘train’ their models. A testing set (context data) of the year ahead will also be provided so that participants can demonstrate the ability of their models to predict the thermal subsystem power consumption for the year ahead.