Reservoir Stochastic Simulation Based on Octave Convolution and Multistage Generative Adversarial Network
Abstract For finely representation of complex reservoir units, higher computing overburden and lower spatial resolution are limited to traditional stochastic simulation. Therefore, based on Generative Adversarial Networks (GANs), spatial distribution patterns of regional variables can be reproduced...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2024-12-01
|
| Series: | Scientific Reports |
| Subjects: | |
| Online Access: | https://doi.org/10.1038/s41598-024-80317-1 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1841559423416270848 |
|---|---|
| author | Xuechao Wu Wenyao Fan Shijie Peng Bing Qin Qing Wang Mingjie Li Yang Li |
| author_facet | Xuechao Wu Wenyao Fan Shijie Peng Bing Qin Qing Wang Mingjie Li Yang Li |
| author_sort | Xuechao Wu |
| collection | DOAJ |
| description | Abstract For finely representation of complex reservoir units, higher computing overburden and lower spatial resolution are limited to traditional stochastic simulation. Therefore, based on Generative Adversarial Networks (GANs), spatial distribution patterns of regional variables can be reproduced through high-order statistical fitting. However, parameters of GANs cannot be optimized under insufficient training samples. Also, a higher computing consumption and overfitting issue easily occurred by stacking Convolutional Neural Networks (CNNs). Therefore, a hybrid framework combined with octave convolution and multi-stage GAN (OctSinGAN) is proposed to perform reservoir simulation. Specifically, a pyramid structure is introduced to perform multiscale representation based on single Training Image (TI), with feature information being captured under different scales. Then, the octave convolution is used to perform multi-frequency feature representation on different feature maps. Finally, a joint loss function is defined to optimize network parameters to improve simulation qualities. Three different kinds of TIs are used to verify the simulation performance of OctSinGAN. Results show that different simulations are similar to the corresponding TIs in terms of spatial variability, channel connectivity and spatial structures, with a relatively high simulation performance overall. |
| format | Article |
| id | doaj-art-41b2dba1cdd0463f9d7a4a0dcbd47e20 |
| institution | Kabale University |
| issn | 2045-2322 |
| language | English |
| publishDate | 2024-12-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Scientific Reports |
| spelling | doaj-art-41b2dba1cdd0463f9d7a4a0dcbd47e202025-01-05T12:30:10ZengNature PortfolioScientific Reports2045-23222024-12-0114111910.1038/s41598-024-80317-1Reservoir Stochastic Simulation Based on Octave Convolution and Multistage Generative Adversarial NetworkXuechao Wu0Wenyao Fan1Shijie Peng2Bing Qin3Qing Wang4Mingjie Li5Yang Li6School of Computer and Information Engineering, Hubei Normal UniversitySchool of Computer Science, China University of GeosciencesHubei Post and Telecommunications Planning and Design Co., LtdChina Telecom Wuhan BranchHubei Post and Telecommunications Planning and Design Co., LtdHubei Post and Telecommunications Planning and Design Co., LtdSchool of Computer and Information Engineering, Hubei Normal UniversityAbstract For finely representation of complex reservoir units, higher computing overburden and lower spatial resolution are limited to traditional stochastic simulation. Therefore, based on Generative Adversarial Networks (GANs), spatial distribution patterns of regional variables can be reproduced through high-order statistical fitting. However, parameters of GANs cannot be optimized under insufficient training samples. Also, a higher computing consumption and overfitting issue easily occurred by stacking Convolutional Neural Networks (CNNs). Therefore, a hybrid framework combined with octave convolution and multi-stage GAN (OctSinGAN) is proposed to perform reservoir simulation. Specifically, a pyramid structure is introduced to perform multiscale representation based on single Training Image (TI), with feature information being captured under different scales. Then, the octave convolution is used to perform multi-frequency feature representation on different feature maps. Finally, a joint loss function is defined to optimize network parameters to improve simulation qualities. Three different kinds of TIs are used to verify the simulation performance of OctSinGAN. Results show that different simulations are similar to the corresponding TIs in terms of spatial variability, channel connectivity and spatial structures, with a relatively high simulation performance overall.https://doi.org/10.1038/s41598-024-80317-1Reservoir unitsGenerative adversarial networksOctave convolutionJoint loss functionMulti-scale spatial representation |
| spellingShingle | Xuechao Wu Wenyao Fan Shijie Peng Bing Qin Qing Wang Mingjie Li Yang Li Reservoir Stochastic Simulation Based on Octave Convolution and Multistage Generative Adversarial Network Scientific Reports Reservoir units Generative adversarial networks Octave convolution Joint loss function Multi-scale spatial representation |
| title | Reservoir Stochastic Simulation Based on Octave Convolution and Multistage Generative Adversarial Network |
| title_full | Reservoir Stochastic Simulation Based on Octave Convolution and Multistage Generative Adversarial Network |
| title_fullStr | Reservoir Stochastic Simulation Based on Octave Convolution and Multistage Generative Adversarial Network |
| title_full_unstemmed | Reservoir Stochastic Simulation Based on Octave Convolution and Multistage Generative Adversarial Network |
| title_short | Reservoir Stochastic Simulation Based on Octave Convolution and Multistage Generative Adversarial Network |
| title_sort | reservoir stochastic simulation based on octave convolution and multistage generative adversarial network |
| topic | Reservoir units Generative adversarial networks Octave convolution Joint loss function Multi-scale spatial representation |
| url | https://doi.org/10.1038/s41598-024-80317-1 |
| work_keys_str_mv | AT xuechaowu reservoirstochasticsimulationbasedonoctaveconvolutionandmultistagegenerativeadversarialnetwork AT wenyaofan reservoirstochasticsimulationbasedonoctaveconvolutionandmultistagegenerativeadversarialnetwork AT shijiepeng reservoirstochasticsimulationbasedonoctaveconvolutionandmultistagegenerativeadversarialnetwork AT bingqin reservoirstochasticsimulationbasedonoctaveconvolutionandmultistagegenerativeadversarialnetwork AT qingwang reservoirstochasticsimulationbasedonoctaveconvolutionandmultistagegenerativeadversarialnetwork AT mingjieli reservoirstochasticsimulationbasedonoctaveconvolutionandmultistagegenerativeadversarialnetwork AT yangli reservoirstochasticsimulationbasedonoctaveconvolutionandmultistagegenerativeadversarialnetwork |