
Convolutional neural network microseismic event detection based on variance fractal dimension
Author(s) -
Guoqing Han,
Yan Song,
Zejie Chen,
Lin Chen
Publication year - 2022
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/2196/1/012016
Subject(s) - microseism , convolutional neural network , fractal dimension , computer science , fractal , noise (video) , pattern recognition (psychology) , artificial intelligence , algorithm , mathematics , geology , image (mathematics) , seismology , mathematical analysis
Microseismic event detection helps to predict outbreak catastrophic problems and has essential applications in resource exploration. Low SNR microseismic signal detection is a challenging task in microseismic detection. In this paper, we propose a (convolutional neural network microseismic detection method based on variance fractal dimension) VFD-CNN method based on the variance fractal dimension (VFD). In this method, signals and background noise are first measured by variance fractal dimension, which can effectively extract seismic nonlinear features. These fractal features are then fed into VFD-CNN to distinguish signal and noise. Finally, the variance fractal dimension of the test data is fed into the optimal model to detect microseismic events. The VFD-CNN method can significantly improve the detection capability of low SNR microseismic signals. To verify the performance of the VFD-CNN method, We use the VFD-CNN method to synthesize microseismic data. Furthermore, the comparison experiments were conducted using VFD-CNN and short-term averaging to long-term averaging (STA/LTA) algorithms. The results show that the VFD-CNN method can significantly improve the detection of low SNR microseismic signals, and its precision is substantially higher than the STA/LTA algorithm.