Deep-Learning-Based Earthquake Detection for Fiber-Optic Distributed Acoustic Sensing

In this paper, deep learning models trained with real seismic data are proposed and proven to detect earthquakes in fiber-optic distributed acoustic sensor (DAS) measurements. The neural network architectures cover the three classical paradigms: fully connected artificial networks (FC-ANNs), convolutional (CNNs) recurrent (RNNs). Results demonstrate that training these waveforms measured by traditional broadband seismometers can extract learn relevant features of earthquakes, enabling reliable detection waves DAS intrinsic differences between seismograph waveforms, eventual errors labelling data, slightly reduce performance when tested Despites that, still reach up 96.94% accuracy case CNN 93.86% CNN+RNN. method results here reported could represent an important contribution development early warning earthquake system based on technology.