Seismic noise attenuation by signal reconstruction: an unsupervised machine learning approach

Deep learning has shown excellent performance in simulating complex nonlinear mappings from the seismic data to elastic parameters. However, seismic acoustic impedance estimated from a direct mapping from seismic waveform data to P‐wave impedance (single‐input network) is hampered by the limited frequency bands. In this paper, we propose to incorporate the low‐frequency impedance model to constrain the inversion (multi‐input network). We add a feature fusion layer to force the lateral smoothness. Besides, usually, a given seismic survey is likely to contain only a few well logs, which is insufficient for conventional deep‐learning ‐based methods to learn the complex mapping from seismic data to elastic parameters. The problem is compounded by the fact that a network trained with synthetic data (compensated for the lack of logs) cannot be directly used for field data. Therefore, we propose to use transfer learning to mitigate this issue. The multi‐input neural network is trained using synthetics and real data in two stages. We carry out experiments to demonstrate that the two‐step training multi‐input network approach has high accuracy in the time direction, excellent continuity in the lateral direction and favourable robustness. Synthetic and field data examples demonstrate that the proposed network can accurately predict impedance even with limited logging data, which provides a reference for oil and gas exploration in the actual production process.

Section: Introductionmentioning

confidence: 99%

Seismic impedance inversion using a multi‐input neural network with a two‐step training strategy

Meng

Wang

Niu

et al. 2022

“…Supervised deep learning needs a large number of training samples. However, not only the field data are limited (Gao et al., 2021), but it is also difficult to obtain the label data without SDWs by traditional denoising methods, so we use seismic wave modelling technology to produce a large amount of synthetic data for network training. We use synthetic seismic records on a rugged surface as the input, and the simulated seismic records of a uniform half‐space model with the same rugged surface as the output of the CAE, namely, the modelled SDWs without reflections as label data of the network.…”

Section: Introductionmentioning

confidence: 99%

A deep learning approach to separating scattered direct waves from reflection seismic records on rugged surface: Synthetic data examples

Gong

Wang

Geng

2022

Seismic records from reflection seismic exploration in complex surface areas often contain scattered direct waves, a type of scattered waves generated by direct waves propagating on a rugged surface. For imaging reflections, scattered direct waves are regarded as noise, which lowers the signal‐to‐noise ratio of the reflected waves and deteriorates the quality of the seismic profile. Importantly, it is very challenging to separate scattered direct waves with strong energy and complex wave field characteristics owing to the rugged surface. In recent years, the rapid development of machine‐learning technology has broadened and advanced the application of deep learning in denoising seismic data. In this context, we propose an approach to using the convolutional autoencoder, a deep learning network, to intelligently separate scattered direct waves from seismic records on a rugged surface. The spectral element method is applied to simulate the elastic wave seismic records and scattered direct waves on a rugged surface. The synthetic shot gathers on the rugged ground are employed as the input for training the convolutional autoencoder network, while the simulated scattered direct waves on the uniform half‐space with the same rugged ground are employed as the label data for training. After the network training is completed, the trained convolutional autoencoder network can be applied to predict scattered direct waves from seismic records. The numerical experiments demonstrate that the proposed approach has good potential for suppressing complex scattered direct waves generated by direct waves propagating on a rugged surface.

“…Deep learning methods are gradually developed and applied to various aspects of the seismic problems, such as denoising (Chen et al ., 2019; Sun et al ., 2019; Saad and Chen, 2020; Gao et al ., 2021), interpolation (Wang et al ., 2020) and inversion (Das et al ., 2018; Chen et al ., 2020; Li et al ., 2020). Facies identification can be solved as a supervised problem.…”

Section: Introductionmentioning

confidence: 99%

Semi‐supervised deep autoencoder for seismic facies classification

Liu

et al. 2021

Facies boundaries are critical for flow performance in a reservoir and are significant for lithofacies identification in well interpretation and reservoir prediction. Facies identification based on supervised machine learning methods usually requires a large amount of labelled data, which are sometimes difficult to obtain. Here, we introduce the deep autoencoder to learn the hidden features and conduct facies classification from elastic attributes. Both labelled and unlabelled data are involved in the training process. Then, we develop a semi‐supervised deep autoencoder by taking the mean of intra‐class and the whole population of facies into account to construct a classification regularization term, thereby improving the classification accuracy and reducing the uncertainty. The new method inherits the profits of deep autoencoder and absorbs the information provided by labelled data. The proposed method performs well and produces promising results when it is used to address problems of reservoir prediction and facies identification. The new method is evaluated on both well and seismic data and compared with the conventional deep autoencoder method, which demonstrates its feasibility and superiority with respect to classification accuracy.