Abstract:This study developed a noise-reduction method for acoustic logging array signals using a deep neural network algorithm in the time-frequency domain. Initially, we derived analytical solutions for the received waveforms when the acoustic logging tool was positioned either at the centre or eccentrically within the borehole. To simulate the received waveforms across various formations, we developed a real-axis integration algorithm. Subsequently, we devised a noise-reduction algorithm workflow based on a convolutional neural network (CNN) and configured the structure and parameters of the U-net using TensorFlow. To address the scarcity of open datasets, we established both signal and noise datasets. The signal dataset was generated using theoretical simulation encompassing various model parameters, while the noise dataset was collected during tool testing and downhole operations. The trained model demonstrated substantial noise-reduction capabilities during validation. To validate the effectiveness of the algorithm, we applied noise reduction to actual data collected during downhole operations in the TangGu oilfield, yielding impressive results across different types of noisy data. Therefore, the U-net-based time-domain noise-reduction algorithm proposed in this paper holds the potential to significantly improve the quality of acoustic logging array signals.

What problem does this paper attempt to address?

This paper aims to solve the problem that the received signals are interfered by complex downhole noises during the process of Acoustic Logging While Drilling (ALWD). Specifically, the goal of the paper is to develop a time - frequency domain denoising method based on the deep neural network algorithm to improve the quality of ALWD array acoustic signals. ### Background and Problem Description of the Paper In the exploration and development of oil and gas resources, ALWD technology provides the measurement of formation acoustic velocity and elastic parameters by actively emitting acoustic waves and measuring the propagation time differences in the formations around the well. However, due to the extremely harsh downhole operating environment, including factors such as high temperature, high pressure, strong vibration and electromagnetic interference, the acoustic signals received by ALWD tools are often severely interfered. These interferences not only reduce the signal quality but also affect the accuracy of formation velocity measurement. ### Existing Challenges Traditional denoising methods, such as band - pass filters and slowness - time coherence methods, although they can remove noise to a certain extent, are not effective in dealing with complex downhole noises. Complex noise types, such as similar - frequency noise, broadband noise, sudden disturbances and high - energy pulses, significantly reduce the quality of the received signals, thereby affecting the performance of acoustic processing algorithms. ### Solution To overcome the above challenges, the paper proposes a convolutional neural network (CNN) denoising method based on the U - net architecture. This method first converts the noisy signal into the time - frequency domain through the short - time Fourier transform (STFT), then uses the U - net network to extract the features of the signal and noise and generate corresponding masks. Finally, the separated signal and noise are converted back from the time - frequency domain to the time domain through the inverse STFT. ### Dataset Construction Due to the scarcity of publicly available high - quality labeled data, the research team constructed a signal library containing 61,272 data points and a noise library containing 56,862 noise signals through theoretical simulation and actual collection. These data are randomly divided into training sets, validation sets and test sets for training and evaluating the U - net model. ### Experimental Results The experimental results show that the U - net model exhibits significant denoising ability in the validation stage. In particular, when denoising the actual downhole data from the Tanggu Oilfield in China, the model has achieved satisfactory results on different types of noise data. ### Conclusion The time - frequency domain denoising algorithm based on U - net proposed in this paper has great potential in improving the quality of ALWD array acoustic signals and is expected to play an important role in future acoustic logging data processing.

Acoustic logging array signal denoising using U-net and a case study in TangGu oil field

Active-source Ocean Bottom Seismometer Data Denoising Based on Multiresolution Convolutional Neural Networks

Efficient Seismic Data Denoising via Deep Learning With Improved MCA-SCUNet

An Intelligent Denoising Method for Nuclear Magnetic Resonance Logging Measurement Based on Residual Network

Automatic Identification of Strong Energy Noise in Seismic Data Based on U-Net

A novel denoising method for low SNR NMR Logging echo signal based on deep learning

An Underwater Wideband Sound Source Localization Method Based on Light Neural Network Structure

Low-Frequency Magnetotelluric Data Denoising Using Improved Denoising Convolutional Neural Network and Gated Recurrent Unit

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

Underwater Acoustic Signal Noise Reduction Based on a Fully Convolutional Encoder-Decoder Neural Network

Coupled Noise Reduction in Distributed Acoustic Sensing Seismic Data Based on Convolutional Neural Network

Attenuation of marine seismic interference noise employing a customized U-Net

Simultaneous reconstruction and denoising for DAS-VSP seismic data by RRU-net

Distributed Acoustic Sensing Vertical Seismic Profile Data Denoiser Based on Convolutional Neural Network

Unsupervised Learning with Waveform Multi-Branch Attention Mechanism for Erratic Noise Attenuation

Study of Acoustic Emission Signal Noise Attenuation Based on Unsupervised Skip Neural Network

Denoising Distributed Acoustic Sensing Data Using Unsupervised Deep Learning

Automatic bioacoustics noise reduction method based on a deep feature loss network

A Deep Neural Network Approach to Acoustic Source Localization in a Shallow Water Tank Experiment

TEMDnet: A Novel Deep Denoising Network for Transient Electromagnetic Signal With Signal-to-Image Transformation

A Novel Iterative PA-MRNet: Multiple noise suppression and weak signals recovery for downhole DAS data