PULSE
PULSE

PULSE
PULSE
 : 
X-ray pulse detection using AI
X-ray pulse detection using AI

A Project coordinated by IIIA.

Web page:

Principal investigator: 

Collaborating organisations:

Funding entity:

European Comission
European Comission

Funding call:

Funding call URL:

Project #:

AHEAD 2020
AHEAD 2020

Funding amount:

60.000,00€
60.000,00€

Duration:

01/Aug/2020
01/Aug/2020
31/Jul/2021
31/Jul/2021

Extension date:

Astrophysics are interested in the observations of remote objects in the X-ray spectrum. But X rays are blocked by the upper atmosphere, then it is necessary to use telescopes installed on artificial satellites. 

ATHENA X-ray Observatory is a mission of the European Space Agency (ESA) science program (2014), to be launched in the early 2030s. ATHENE will contains X-IFU X-ray Integral Field Unit, that is, a cryogenic X-ray spectrometer, based on a large array of Transition Edge Sensors (TES):

  • Range: 0.2 – 12 keV
  • Spectral resolution: 2.5 eV
  • Time resolution: 10μs
  • Field of view: with 5’’ pixels, over a of 5 arc minutes (0.083 degrees) in equivalent diameter.

The Institute of Physics of Cantabria (IFCA) participate in the Horizon 2020 (Integrating and opening research infrastructures of European interest) project AHEAD2020: Integrated Activities for the High Energy Astrophysics Domain. Task14.4, coordinated by  Maite Ceballos (IFCA), Machine learning techniques for micro-calorimeter data-reduction, include a prospective analysis of the different machine learning techniques applicable to the triggering and reconstruction of the X-IFU current pulses aimed at improving S/N and energy resolution, comparing its performance with already established traditional techniques such optimal filtering, using figures of merit like the energy resolution or the processing requirements.

IIIA collaborates in this task offering its knowledge and experience in Artificial Intelligence to deal with a noise signal containing X-ray pulses of different shapes and sometimes overlapping. We are working with synthetic and real signal and we are designing algorithms to detect the pulses, to separate them when overlapping, and to measure the energy of each pulse. We are exploring machine learning techniques as mainly CNN and also SVM. We have to take into account that it is not possible to send all the data to earth implying that the detection and analysis has to be made on board with very limited computational resources.

Astrophysics are interested in the observations of remote objects in the X-ray spectrum. But X rays are blocked by the upper atmosphere, then it is necessary to use telescopes installed on artificial satellites. 

ATHENA X-ray Observatory is a mission of the European Space Agency (ESA) science program (2014), to be launched in the early 2030s. ATHENE will contains X-IFU X-ray Integral Field Unit, that is, a cryogenic X-ray spectrometer, based on a large array of Transition Edge Sensors (TES):

  • Range: 0.2 – 12 keV
  • Spectral resolution: 2.5 eV
  • Time resolution: 10μs
  • Field of view: with 5’’ pixels, over a of 5 arc minutes (0.083 degrees) in equivalent diameter.

The Institute of Physics of Cantabria (IFCA) participate in the Horizon 2020 (Integrating and opening research infrastructures of European interest) project AHEAD2020: Integrated Activities for the High Energy Astrophysics Domain. Task14.4, coordinated by  Maite Ceballos (IFCA), Machine learning techniques for micro-calorimeter data-reduction, include a prospective analysis of the different machine learning techniques applicable to the triggering and reconstruction of the X-IFU current pulses aimed at improving S/N and energy resolution, comparing its performance with already established traditional techniques such optimal filtering, using figures of merit like the energy resolution or the processing requirements.

IIIA collaborates in this task offering its knowledge and experience in Artificial Intelligence to deal with a noise signal containing X-ray pulses of different shapes and sometimes overlapping. We are working with synthetic and real signal and we are designing algorithms to detect the pulses, to separate them when overlapping, and to measure the energy of each pulse. We are exploring machine learning techniques as mainly CNN and also SVM. We have to take into account that it is not possible to send all the data to earth implying that the detection and analysis has to be made on board with very limited computational resources.

2020
Jesús Vega,  M. Ceballos,  Josep Puyol-Gruart,  Pere García,  B. Cobo,  & F. J. Carrera (2020). TES X-ray pulse identification using CNNs. ADASS XXX . [BibTeX]  [PDF]
Pere Garcia Calvés
Tenured Scientist
Phone Ext. 216

Josep Puyol-Gruart
Tenured Scientist
Phone Ext. 210

Jesús Vega
Contract Researcher
Phone Ext. 232

Sharika Vijayakumar
Masters Student