Continuously improving the accuracy is a hot topic in hyperspectral image (HSI) classification with small-scale samples, due to the high label noise of traditional labeling systems and the high cost of expert labeling systems. We focus on constructing a smaller and more informative training sample set, so an iterative sample selection method guided by uncertainty measurement (ISS-Un) is proposed. The method learns shallow and deep features in the spectral and spatial domains via a convolutional neural network (CNN), where an uncertainty measurement algorithm such as least confidence (LC), marginal sampling (MS) or entropy (Ent) is used to iteratively select high-quality samples for the training set. In addition, we propose a more efficient uncertainty measurement algorithm named margin-entropy fusion (MEF) algorithm to integrate multiple-criteria information. The proposed method is compared with the conventional random sampling method. Experimental results on three HSI datasets show that the proposed ISS-Un method can significantly alleviate the redundancy of training samples and form a more compact and efficient training set, thus improving the classification performance of pixel-oriented HSI. Meanwhile, training sets constructed based on different uncertainty measurement algorithms are applied to five popular CNN models to verify the quality and generalizability of the selected samples. The results show that these training sets work better than random training sampling. Moreover, the proposed MEF algorithm outperforms the LC, MS, and Ent algorithms in selecting samples and is the main recommended scheme.
Shear wave (S-wave) velocity prediction is important for the evaluation of shale oil and gas reservoir. However, there are some problems with traditional models: the parameters of the petrophysical model are relatively fixed, and the machine learning models do not consider the sequence information of the log data. Therefore, the S-wave velocity prediction model based on Temporal Convolutional Network (TCN) for shale reservoir is proposed. The model can flexibly extract the sequence features by adopting causal convolution and dilation factors and mine the inner relationship between the well logs and the reservoir S-wave velocity to achieve a better prediction performance. Two wells of MY1 and FN4 in shale reservoir in the Permian Fengcheng Formation in Mahu Sag of Junggar Basin, Xinjiang Oilfield are taken as an example. The TCN model achieves optimal results on both MY1 and FN4 with mean relative error (MRE) of 0.84% and 1.39%, respectively, when compared with the results of traditional petrophysical models, machine learning models and conventional deep learning models. This indicates that the TCN model has strong effectiveness and generalization in Swave velocity prediction, which provides a new idea for S-wave velocity prediction in shale reservoir.
As an extremely important target for unconventional oil and gas resources exploration at present, shale reservoir differs significantly from conventional clastic and carbonate reservoirs due to their diverse mineral composition, complex pore characteristics, and severe heterogeneity, which makes the conventional theoretical petrophysical models not accurate enough to characterize shale reservoirs. For this reason, machine learning and deep learning methods are introduced to construct a more intelligent petrophysical modeling process, which uses a data-driven approach. And taking the shale reservoirs of the Permian Fengcheng Formation in Mahu Depression of Junggar Basin as an example, we achieve high accuracy Shear wave velocity prediction based on conventional well logs, and the mean relative error (MRE) of prediction is reduced by 2.78-3.88% and the method has good applicability and generalization compared with conventional petrophysical model.
Volcanic rock formations, as an important oil and gas resource reservoir, have received the focus of the energy industry in recent years. Shear wave logging is essential geophysical data for the exploration and evaluation of volcanic rock oil and gas reservoirs. Due to the strong nonlinear relationship between reservoir logging parameters and S-wave velocity, the conventional point-to-point machine learning methods can not effectively construct the feature space. Deep learning adds neighborhood information to learn the depth features relationship, and builds the mapping of S-wave velocity and wireline logs with its powerful nonlinear solving capability, achieves S-wave velocity prediction. Taking the volcanic reservoir in Xujiaweizi area of Songliao Basin in Northeast China as an example, thirteen logging parameters sensitive to S-wave velocity are selected, and the S-wave velocity prediction models are based on deep learning methods (represented by CNN, ViT, and MLP-Mixer) are proposed. The research demonstrates that the proposed deep learning models are able to predict S-wave velocity with more precision, and the modeling method can give great significance for the exploration of the volcanic reservoir.
Shear wave velocity (S-wave velocity) is the essential data for rock mechanics parameter prediction and reservoir compressibility evaluation in shale oil and gas sweet spot optimization. Owing to the extremely complex rock components and pore structure of shale reservoirs, it is usually difficult to represent the relationship between well logs and S-wave velocity accurately for theoretical petrophysical models and conventional empirical formulas. Within this context, a novel architecture of S-wave velocity estimation based on N-BEATS model was proposed. It can help improve the estimation accuracy by effectively incorporating sequence features of well logs. To illustrate its performance, a case study for shale reservoir in the Permian Fengcheng Formation in Mahu Sag of Junggar Basin, Xinjiang Oilfield, was performed. Seven kinds of conventional well logs were selected to establish the regression model. Compared with Xu-White model and eleven machine learning methods (MLs) and deep learning methods (DLs), the mean relative error (MRE) of N-BEATS has been reduced to 0.946%. The case study showed that N-BEATS model proposed can achieve better performance and generalization, which indicated its widespread application value to the other oil and gas exploration area.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.