Gravitational wave signal recognition and ring-down time estimation via Artificial Neural Networks

• Published in: Expert systems with applications 2022-11, Vol.207, p.117931, Article 117931
• Main Authors: Santos, Gerson R., Santos, Antonio de Pádua, Protopapas, Pavlos, Ferreira, Tiago A.E.
• Format: Article
• Language: English
• Subjects: Artificial Neural Network; Gravitational waves; Pattern recognition; Time series analysis
• ISSN: 0957-4174; 1873-6793
• DOI: 10.1016/j.eswa.2022.117931

Laser Interferometer Gravitational-Wave Observatory (LIGO) was the first laboratory to measure the gravitational waves successfully. An exceptional experimental design was needed to measure distance changes less than an atomic nucleus. In the same way, the data analyses to confirm and extract information is a tremendously challenging task. This article shows a computational procedure based on Artificial Neural Networks (ANN) to recognize a black hole-black hole gravitation wave event signal from the LIGO data. With the ANN introduced methodology, it is possible to define a numerical score, like a thermometer. High score values are associated with gravitational wave observation and small values with noise. Building a time series from these scores values, physical information about the astronomical system’s damping time, the ring-down time, can be estimated at a first approximation, based on a damped harmonic oscillator modeling. Here, the ring-down time is estimated, at a first approximation, with a direct data measure on the ANN score time series, without using numerical relativity techniques and high computational power. •Artificial Neural Network application for signal recognition in very noisily data.•Recognition of black hole-black hole gravitational wave event from de LIGO’s data.•Score time series definition for gravitational wave event recognition.•Score time series analysis from multiple and synchronized data sources.•Ring-Down time estimation of the black hole-black hole system.