University of Amsterdam
Institute of Informatics, SNE
EDL P16-25 P4: Deep Learning for High-Tech Systems and Materials
|Efficient Deep Learning for System Health Management in radio Astronomy|
Modern radio telescopes generate an ever growing amount of data. To improve spatial resolution, sensitivity, and field of view, larger telescope arrays are being constructed. The increased size and capabilities of the instruments lead to more data, a higher system complexity, and ever-growingerror rates.
System health management (SHM) is the process of detecting, diagnosing, and remedying system failures to maximize system uptime. SHM and pinpointing error sources in radio telescopes today still rely on manual inspection and human interpretation of the data. This is error-prone, and large-scale spatially distributed radio telescopes such as LOFAR face reliabilityand up-time issues. These issues stem from the scale and complexity of the systems and processing chains involved. In LOFAR, for example, we expect that, due its scale and its somewhat harsh operating conditions, at any given time several components in the systems will not operate correctly.
Our goal is to apply deep learning based models to effectively detect and
diagnose failures and anomalies within radio telescopes.
Deep learning assisted data inspection for radio astronomy. Michael Mesarcik, Albert-Jan Boonstra, Christiaan Meijer, Walter Jansen, Elena Ranguelova, Rob V van Nieuwpoort . Monthly Notices of the Royal Astronomical Society, Volume 496, Issue 2, August 2020. Journal publication. Open Access-gold. https://doi.org/10.1093/mnras/staa1412; https://drive.google.com/open?id=1tH-2_VAY-jpv-P-a4H3umRvUNkSPP9Ui