Deep Neuron Networks on Gravitational Wave Data Analysis

No Thumbnail Available



Journal Title

Journal ISSN

Volume Title



As the increasing sensitivity of gravitational wave detectors, the detection of gravitational wave events will be more and more frequent. Due to the fact that the gravitational waves of BNS and NSBH will arrive at the earth before the electromagnetic wave counterparts. The gravitational wave-triggered multi- messenger observation becomes a promising region of Astronomy. To meet the requirements of multi-messenger observation, the latency of search algorithm must be within few seconds. In this thesis, we constructed two convolutional neuron networks, one for detecting the present of the gravitational waves and the other for estimating physical parameters of the gravitational waves. Our neuron networks take only 4 seconds to process 3944 seconds data. The accuracy of our neuron network is larger than 99 % for detection when the SNR is larger than 12, and the mean relative errors are less than 10 % when the SNR is larger than 9. We also tested our DNNs with four gravitational wave events: GW150914, GW151226, GW170104 and GW170814. Keywords: Gravitational wave, Deep neural network.



None, Gravitational wave, Deep neural network