Automatic Waveform Classification and Arrival Picking Based on Convolutional Neural Network

Abstract

Automatic waveform classification and arrival picking methods are widely studied to reduce or replace the manual works. Machine learning based methods, especially neural networks, and clustering based methods have shown great potentials in previous studies. However, most of the existing methods are sensitive to noise. The convolution neural networks (CNNs), developed from the traditional neural networks, have been successfully applied in many different fields, but are rarely studied in seismic waveform classification. In this paper, we propose a novel antinoise CNN architecture for waveform classification and also propose to combine k‐means clustering (KC) with CNN classification to pick arrivals (CNN‐KC). Seismic data are sampled to 1‐D vectors using a specific time window. Using the trained CNN classifier, these 1‐D vectors are classified into two categories: waveform and nonwaveform. With the constraint of the first waveform, CNN‐KC can pick the arrival more accurately. We also apply the proposed methods to the synthetic microseismic data with different noise levels and the actual field microseismic data to test their robustness. CNNs perform much better than the traditional multilayer perceptron on the waveform classification of the noisy microseismic data. Based on the analysis of the CNN internal architecture, we also conclude that the main reason that CNN is insensitive to noise is the convolution and pooling layers which behave like smooth operation in some ways. The final results show that the CNN and CNN‐KC are effective and robust methods for waveform classification and arrival picking.

TBM-Mounted Seismic Ahead-Prospecting for Fast Detecting Anomalous Geology Ahead of Tunnel Face

“…One of the most used methods for seismic data processing is the migration imaging method. This method conventionally requires picking the first arrivals of the direct waves on each source-receiver pair, for the velocity analysis, which is important to determine the source-reflector-receiver distance [25] [31]. However, following factors makes manual picking time consuming (1) The amplitudes of both the signal and noise is different from trace to trace.…”

A high accurate automated first‐break picking method for seismic records from high‐density acquisition in areas with a complex surface

“…Chen et al . () proposed a novel anti‐noise CNN architecture for waveform classification and propose to combine k‐means clustering (KC) with CNN classification to pick arrivals (CNN‐KC). A group of synthetic and real high S/N data sets shows that the proposed methods are much more robust than the state‐of‐the‐art STA/LTA method in picking microseismic events; however, features of primary wave in low S/N data are not clear, and the first‐break time is drown by noise.…”

“…Based on the feature that the S/N of primary wave in seismic data from high‐density acquisition in areas with a complex surface is low and on the base of fuzzy clustering algorithms researched by Zhu, Li and Zhang () and Chen (), this paper studies the automated first‐break picking method that combines the design of time window of primary wave, algorithms based on clustering, multi‐angle comprehensive quality evaluation and ant colony algorithm. Firstly, multi‐azimuth spatial interpolation technology is used to determine the range of time window of primary wave, and then the improved clustering algorithm is used to initially pick first breaks.…”

Enhancing Seismic P-Wave Arrival Picking by Target-Oriented Detection of the Local Windows Using Faster-RCNN

“…Chen [20] tried to use unsupervised machine learning to promote the quality of P-wave arrival picking by dividing data into noise and signal segments. With the rise of deep learning, Perol et al [21] proposed a CNN (convolutional neural network) model for earthquake detection and location; Ross et al [22] trained two CNN models to pick the first seismic arrival and get the first motion, respectively; and Zhu [23] and Chen et al [24] creatively treat the phase picking as a classification problem and enhance the P-wave picking after a rough classification using a CNN. In addition to CNN, RNN (recurrent neural network) is also used in P-wave arrival picking attempts, such as Mousavi et al [25] combined a CNN and an RNN to detect earthquake signals and achieved high-detection accuracy with a low rate of false positives, and Zhou et al [26] developed a hybrid algorithm using both convolutional and recurrent neural networks to pick phases from archived continuous waveforms.…”