Multitask learning for petrophysical attribute prediction using convolutional neural network and imbalance dataset

PreviousNext No AccessSEG Technical Program Expanded Abstracts 2020Multitask learning for petrophysical attribute prediction using convolutional neural network and imbalance datasetAuthors: Chaoshun HuBoqin SunChaoshun HuChevron Energy Technology CompanySearch for more papers by this author and Boqin SunChevron Energy Technology CompanySearch for more papers by this authorhttps://doi.org/10.1190/segam2020-w13-03.1 SectionsAboutPDF/ePub ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InRedditEmail AbstractPorosity, permeability and fluid saturations are three fundamental petrophysical attributes of reservoir systems that are directly related to the storage capacity, fluid flow capacity, and amount of hydrocarbon pore volume. Recently, neural networks have been widely utilized to analyze the data and predict the target. However, the traditional neural network based artificial intelligence method is implemented to train the input features to match these attributes, for example, porosity or permeability, respectively. Here we introduce a multitask learning which can learn to predict porosity and permeability at the same time using multitask learning neural networks. We also introduced a new workflow to resolve the imbalance issue existing in well log data which can enable us to estimate the porosity and permeability efficiently and accurately. We also demonstrated the effectiveness of the proposed algorithm and workflow using Equinor’s Volve field public dataset.Keywords: multi-task learning, convolutional neural network, petrophysics, data imbalance, samplingPermalink: https://doi.org/10.1190/segam2020-w13-03.1FiguresReferencesRelatedDetailsCited byLithofacies logging identification for strongly heterogeneous deep-buried reservoirs based on improved Bayesian inversion: The Lower Jurassic sandstone, Central Junggar Basin, China20 January 2023 | Frontiers in Earth Science, Vol. 11 SEG Technical Program Expanded Abstracts 2020ISSN (print):1052-3812 ISSN (online):1949-4645Copyright: 2020 Pages: 3887 publication data© 2020 Published in electronic format with permission by the Society of Exploration GeophysicistsPublisher:Society of Exploration Geophysicists HistoryPublished Online: 30 Sep 2020 CITATION INFORMATION Chaoshun Hu and Boqin Sun, (2020), "Multitask learning for petrophysical attribute prediction using convolutional neural network and imbalance dataset," SEG Technical Program Expanded Abstracts : 3857-3861. https://doi.org/10.1190/segam2020-w13-03.1 Plain-Language Summary Keywordsmulti-task learningconvolutional neural networkpetrophysicsdata imbalancesamplingPDF DownloadLoading ...