Denoising Method for Microseismic Signals with Convolutional Neural Network Based on Transfer Learning

Abstract

Microseismic signals contain various information for oil and gas developing. Increasing the signal-to-noise ratio of microseismic signals can successfully improve the effectiveness of oil and gas resource exploration. The lack of sufficient labeled microseismic signals makes it difficult to train neural network model. Transfer learning can solve this problem using image data sets to pre-train the denoising model and the learned knowledge can be transferred into microseismic signals denoising. In addition, a convolutional neural network (CNN) model with 16 layers is designed for noise reduction. Considering the strong similarity between noisy signals and denoising signals, residual learning is utilized to optimize the denoising model. The simulation experiment results show that the proposed denoising model eliminates the noise in the microseismic signals effectively and quickly, restores the amplitude of the microseismic signals with high accuracy, and has excellent effect in denoising on the information at the edge.

Research on U-Net seismic signal denoising combined with residual dense blocks

“…In formula (1), Y (t, f) represents a noisy seismic signal, X (t, f) represents a clean seismic signal, and N (t, f) represents noise. The goal of noise reduction is to recover a clean seismic signal X (t, f) from the noisy seismic signal Y (t, f) with as little distortion as possible.…”

“…These noises exist in the same frequency band as the seismic signal and are similar to the seismic signal in amplitude, frequency, and frequency range. The denoising quality of seismic signals directly affects the accuracy of subsequent seismic event processing [1]. * Author to whom any correspondence should be addressed.…”

Application of improved AlexNet for seismic data denoising

“…The main principle of the algorithm is to transform the signal into the frequency-wavnumber domain, and use the set filter to separate the signal and noise to achieve the purpose of denoising. This algorithm is mainly applicable to the situation where the characteristics of the signal and noise are significantly different, and it cannot be denoised when the difference is not obvious [2] . Transform domain algorithms include wavelet transform, SVD decomposition, curve-wave transform, etc.…”

Self-Supervised Representation Learning for Quasi-Simultaneous Arrival Signal Identification Based on Reconnaissance Drones

“…However, QSAS is not simply a combination of different modulations and contains a lot of noise. Therefore, it is crucial to explore how to train the network and perform transfer learning (TL) [35,36] to adapt to different signal-to-noise ratios (SNR) [37,38] and improve noise adaptation.…”