Automatic identification of MHD modes in magnetic fluctuation spectrograms using deep learning techniques

The control and mitigation of magnetohydrodynamic (MHD) oscillation modes is an issue in fusion science because these modes can contribute to the outward particle/energy flux and can drive the device away from ignition conditions. It is of general interest to extract the mode information from large...

Descripción completa

Detalles Bibliográficos
Autores: Bustos, A., Ascasíbar, E., Cappa, A., Mayo-García, R.
Tipo de recurso: artículo
Fecha de publicación:2021
País:España
Institución:Centro de Investigaciones Energéticas, Medioambientales y Tecnológicas (CIEMAT)
Repositorio:Docu-menta. Repositorio Institucional del CIEMAT
Idioma:inglés
OAI Identifier:oai:dnet:documenta___::fde8dcce812d8b81acfcadd32ca9a065
Acceso en línea:https://hdl.handle.net/20.500.14855/1601
Access Level:acceso abierto
Descripción
Sumario:The control and mitigation of magnetohydrodynamic (MHD) oscillation modes is an issue in fusion science because these modes can contribute to the outward particle/energy flux and can drive the device away from ignition conditions. It is of general interest to extract the mode information from large experimental databases in a fast and reliable way. We present a software tool based on deep learning that can identify these oscillation modes taking Mirnov coil spectrograms as input data. It uses convolutional neural networks that we trained with manually-annotated spectrograms from the TJ-II stellarator database. We have tested several detector architectures, resulting in a detector area under the curve score of 0.6 on the test set. Finally, it is applied to find MHD modes in our spectrograms to show how this new software tool can be used to mine large databases.