深度神经网络三维地震资料断层解释损失函数对比

doi:10.13810/j.cnki.issn.1000-7210.2023.06.001

Abstract
Figure/Table
References
Related Citation (12)

Download: PDF (6877 KB) HTML (1 KB)
Export: BibTeX | EndNote (RIS)

Abstract Fault interpretation is one of the key steps for seismic data interpretation.The rapid development of deep learning，represented by neural networks，has greatly improved the efficiency and accuracy of seismic fault interpretation.The neural networks are trained by stochastic gradient descent optimization.The parameters of the network model are updated iteratively by using the loss function to measure the error of the model.The selection of the loss function is crucial for the seismic fault interpretation.In this paper，to interpret 3D seismic fault，we use the 3D U-Net model as the network structure and Adam as the optimizer to train the network with 3D synthetic samples.In terms of fault interpretation effects，we compare 10 loss functions including Balanced Cross-Entropy （BCE），Dice，Focal，Cosine，Log-Cosh Dice，Tversky，Focal-Tversky，Wasserstein，BCE-Dice，and BCE-Cosine.Normalization and data augmentation are applied to the trained data to mitigate the discrepancy between synthetic and field data.With the same network model，training parameters，and stopping criteria，we compare the convergence speed，calculation efficiency，and anti-noise performance of the 10 loss functions on 3D U-Net and analyze the fault prediction effect by using actual seismic data of the F3 field from offshore Netherlands.The experimental results show that 3D U-Net trained by Tversky and focal-Tversky loss functions can predict fault with better continuity.When crossed or parallel faults are densely distributed，and adjacent fault features can influence each other，the 3D U-Net prediction faults trained by BCE，BCE-dice，and BCE-cosine loss functions are complete，clear，and rich in detail.The research can provide a reference for selecting appropriate loss functions in different scenarios for seismic fault interpretation.

Key words： fault interpretation loss functions 3D U-Net data augmentation

Received: 17 December 2022

PACS:

P631

	Service

	E-mail this article
	Add to my bookshelf
	Add to citation manager
	E-mail Alert
	RSS
	Articles by authors
	ZHANG Miaomiao
	WU Bangyu
	MA Debo
	WANG Zhiguo

Cite this article:

ZHANG Miaomiao,WU Bangyu,MA Debo等. Loss function comparison for fault interpretation of three-dimensional seismic data based on deep neural network[J]. OGP, 2023, 58(6): 1299-1312.

URL:

http://www.ogp-cn.com/EN/10.13810/j.cnki.issn.1000-7210.2023.06.001 OR http://www.ogp-cn.com/EN/Y2023/V58/I6/1299

[1]	DOU Y,LI K,ZHU J,et al.Attention-based 3-D seismic fault segmentation training by a few 2-D slice labels[J].IEEE Transactions on Geoscience and Remote Sensing,2022,60:1-15.
[2]	张政,严哲,顾汉明.基于残差网络与迁移学习的断层自动识别[J].石油地球物理勘探,2020,55(5):950-956.ZHANG Zheng,YAN Zhe,GU Hanming.Automatic fault recognition with residual network and transfer learning[J].Oil Geophysical Prospecting,2020,55(5):950-956.
[3]	BAHORICH M,FARMER S.3-D seismic discontinuity for faults and stratigraphic features:the coherence cube[J].The Leading Edge,1995,14(10):1053-1058.
[4]	MARFURT K J,KIRIN R L,FARMER S L,et al.3-D seismic attributes using a semblance-based coherency algorithm[J].Geophysics,1998,63(4):1150-1165.
[5]	GERSZTENKORN A,MARFURT K J.Eigenstructure-based coherence computations as an aid to 3-D structural and stratigraphic mapping[J].Geophysics,1999,64(5):1468-1479.
[6]	Van BEMMEL P P,PEPPER R E F.Seismic signal processing method and apparatus for generating a cube of variance values:US6151555[P].1999-08-19.
[7]	ROBERTS A.Curvature attributes and their application to 3D interpreted horizons[J].First Break,2001,19(2):85-100.
[8]	PEDERSEN S I,SKOV T,RANDEN T,et al.Automatic fault extraction using artificial Ants[M]//ISKE A,RANDEN T.Mathematical Methods and Modelling in Hydrocarbon Exploration and Production.Springer,Berlin,Heidelberg,2005,107-116.
[9]	WU X,FOMEL S.Automatic fault interpretation with optimal surface voting[J].Geophysics,2018,83(5):067-082.
[10]	BI Z,WU X.Improving fault surface construction with inversion-based methods[J].Geophysics,2021,86(1):IM1-IM14.
[11]	WANG Z,LI B,LIU N,et al.Distilling knowledge from an ensemble of convolutional neural networks for seismic fault detection[J].IEEE Geoscience and Remote Sensing Letters,2022,19:1-5.
[12]	LIU N,HE T,TIAN Y,et al.Common azimuth seismic data fault analysis using residual UNet[J].Interpretation,2020,8(3):SM25-SM37.
[13]	JING J,YAN Z,ZHANG Z,et al.Fault detection using a convolutional neural network trained with point-spread function-convolution-based samples[J].Geophy-sics,2023,88(1):IM1-IM14.
[14]	何涛,刘乃豪,吴帮玉,等.基于ResU-Net的三维断层识别方法及应用[J].工程数学学报,2023,40(1):1-19.HE Tao,LIU Naihao,WU Bangyu,et al.ResU-Net based three-dimensional fault identification method and application[J].Chinese Journal of Engineering Mathematics,2023,40(1):1-19.
[15]	YU J,WU B.Attention and hybrid loss guided deep learning for consecutively missing seismic data Reconstruction[J].IEEE Transactions on Geoscience and Remote Sensing,2022,60:1-8.
[16]	LI X,WU B,ZHU X,et al.Consecutively missing seismic data interpolation based on coordinate attention Unet[J].IEEE Geoscience and Remote Sensing Letters,2022,19:1-5.
[17]	HE T,WU B,ZHU X.Seismic data consecutively missing trace interpolation based on multistage neural network training process[J].IEEE Geoscience and Remote Sensing Letters,2022,19:1-5.
[18]	WU B,MENG D,WANG L,et al.Seismic impedance inversion using fully convolutional residual network and transfer learning[J].IEEE Geoscience and Remote Sensing Letters,2020,17(12):2140-2144.
[19]	ZHENG X,WU B,ZHU X.Multi-task deep learning seismic impedance inversion optimization based on homoscedastic uncertainty[J].Applied Sciences,2022,12(3):1200.
[20]	WU B,MENG D,ZHAO H.Semi-supervised learning for seismic impedance inversion using generative adversarial networks[J].Remote Sensing,2021,13(5):909.
[21]	伊小蝶,吴帮玉,孟德林,等.数据增广和主动学习在波阻抗反演中的应用[J].石油地球物理勘探,2021,56(4):707-715.YI Xiaodie,WU Bangyu,MENG Delin,et al.Application of data augmentation and active learning to seismic wave impedance inversion[J].Oil Geophysical Prospecting,2021,56(4):707-715.
[22]	QIU C,WU B,LIU N,et al.Deep learning prior model for unsupervised seismic data random noise attenuation[J].IEEE Geoscience and Remote Sensing Letters,2022,19:1-5.
[23]	QIU C,WU B,MENG D,et al.Attention neural network semblance velocity auto picking with reference velocity curve data augmentation[C].2021 IEEE International Geoscience and Remote Sensing Symposium IGARSS,Brussels,Belgium,2021,4596-4599.
[24]	路鹏飞,杜文龙,李丽,等.基于VNet深度学习架构的低序级断层智能识别方法[J].石油地球物理勘探,2022,57(6):1276-1286.LU Pengfei,DU Wenlong,LI Li,et al.Intelligent recognition method of low-grade faults based on VNet deep learning architecture[J].Oil Geophysical Prospecting,2022,57(6):1276-1286.
[25]	陈俊安,陈海东,龚伟,等.深度学习与边缘增强相结合的断裂综合检测技术——顺北地区超深走滑断裂检测应用实例[J].石油地球物理勘探,2022,57(6):1304-1316.CHEN Jun an,CHEN Haidong,GONG Wei,et al.Application of comprehensive fault detection technology combining deep learning with edge enhancement in detecting ultra-deep strike-slip faults in Shunbei block[J].Oil Geophysical Prospecting,2022,57(6):1304-1316.
[26]	杨平,宋强功,詹仕凡,等.基于深度学习的高效构造解释技术研发及工业化应用[J].石油地球物理勘探,2022,57(6):1265-1275.YANG Ping,SONG Qianggong,ZHAN Shifan,et al.Research and industrial application of efficient structural interpretation technology based on deep learning[J].Oil Geophysical Prospecting,2022,57(6):1265-1275.
[27]	WU X,LIANG L,SHI Y,et al.FaultSeg3D:using synthetic data sets to train an end-to-end convolutional neural network for 3D seismic fault segmentation[J].Geophysics,2019,84(3):IM35-IM45.
[28]	WEI X L,ZHANG C X,KIM S W,et al.Seismic fault detection using convolutional neural networks with focal loss[J].Computers & Geosciences,2022,158:104968.
[29]	DOU Y,LI K,ZHU J,et al.MD loss:efficient training of 3-D seismic fault segmentation network under sparse labels by weakening anomaly annotation[J].IEEE Transactions on Geoscience and Remote Sensing,2022,60:1-14.
[30]	何易龙,文晓涛,王锦涛,等.基于3D U-Net++L3卷积神经网络的断层识别[J].地球物理学进展,2022,37(2):607-616.HE Yilong,WEN Xiaotao,WANG Jintao,et al.Fault recognition based on 3D U-Net++L3 convolutional neural network[J].Progress in Geophysics,2022,37(2):607-616.
[31]	ARAYA-POLO M,DAHLKE T,FROGNER C,et al.Automated fault detection without seismic processing[J].The Leading Edge,2017,36(3):208-214.
[32]	SALEHI S S M,ERDOGMUS D,GHOLIPOUR A.Tversky loss function for image segmentation using 3D fully convolutional deep networks[C].Machine Learning in Medical Imaging,Cham,Switzer land,2017,379-387.
[33]	JADON S.A survey of loss functions for semantic segmentation[C].2020 IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology (CIBCB),Via del Mar,Chile,2020,1-7.
[34]	JAKHETIVA V,CHAUDHARY S,SUBUDHI B N,et al.Perceptually unimportant information reduction and cosine Similarity-Based quality assessment of 3D-Synthesized images[J].IEEE Transactions on Image Processing,2022,31:2027-2039.
[35]	RAJ A,TOLLENS F,HANSEN L,et al.Deep Learning-Based total kidney volume segmentation in autosomal dominant polycystic kidney disease using attention,cosine loss,and sharpness aware minimization[J].Diagnostics,2022,12(5):1159.
[36]	ZHU D,ZHENG C,WANG W,et al.Rapid fault extraction from seismic images via deep learning[J].Journal of Electronic Imaging,2022,31(5):051423.
[37]	JU M,DING C,GUO Y,et al.IDGCP:image dehazing based on gamma correction prior[J].IEEE Transactions on Image Processing,2020,29:3104-3118.
[38]	RONNEBERGER O,FISCHER P,BROX T.U-Net:convolutional networks for biomedical image segmentation[C].Medical Image Computing and Computer-Assisted Intervention-MICCAI 2015,Cham,Switzer- land,2015,234-241.
[39]	XIE S,TU Z.Holistically-Nested edge detection[C].2015 IEEE International Conference on Computer Vision (ICCV),Santiago,Chile,2015,1395-1403.
[40]	GAO K,HUANG L,ZHENG Y,et al.Automatic fault detection on seismic images using a multiscale attention convolutional neural network[J].Geophysics,2022,87(1):N13-N29.
[41]	MILLETARI F,NAVAB N,AHMADI S A.V-Net:fully convolutional neural networks for volumetric medical image segmentation[C].2016 Fourth International Conference on 3D Vision (3DV),Stanford,CA,USA,2016 565-571.
[42]	唐杰,孟涛,韩盛元,等.基于多分辨率U-Net网络的地震数据断层检测方法[J].石油地球物理勘探,2021,56(3):436-445.TANG Jie,MENG Tao,HAN Shengyuan,et al.A fault detection method of seismic data based on Multi ResU-Net[J].Oil Geophysical Prospecting,2021,56(3):436-445.
[43]	GAO K,HUANG L,ZHENG Y.Fault detection on seismic structural images using a nested residual U-Net[J].IEEE Transactions on Geoscience and Remote Sensing,2022,60:1-15.
[44]	LIN T Y,GOYAL P,GIRSHICK R,et al.Focal loss for dense object detection[C].2017 IEEE International Conference on Computer Vision (ICCV),Venice,Italy,2017,2999-3007.
[45]	邓建国,张素兰,张继福,等.监督学习中的损失函数及应用研究[J].大数据,2020,6(1):60-80.DENG Jianguo,ZHANG Sulan,ZHANG Jifu,et al.Loss function and application research in supervised learning[J].Big Data Research,2020,6(1):60-80.
[46]	NANNI L,CUZA D,LUMINI A,et al.Deep ensembles in bioimage segmentation [EB/OL].(2021-12-24)[2023-09-01].https://arxiv.org/abs/2112.12955.
[47]	ABRAHAM N,KHAN N M.A novel focal tversky loss function with improved attention U-Net for lesion segmentation[C].IEEE 16th International Symposium on Biomedical Imaging,Venice,Italy,2018,683-687.
[48]	FROGNER C,ZHANG C,MOBAHI H,et al.Learning with a Wasserstein loss[C].Proceedings of the 28th International Conference on Neural Information Processing Systems-Volume 2,Cambridge,MA,USA,2015,2053-2061.
[49]	RUBNER Y,TOMASI C,GUIBAS L J.The earth mover s distance as a metric for image retrieval[J].International Journal of Computer Vision,2000,40(2):99-121.
[50]	KINGMA D P,BA J.Adam:A method for stochastic optimization[EB/OL].(2017-01-30)[2023-09-01].https://arxiv.org/abs/1412.6980.
[51]	BI Z,WU X,GENG Z,et al.Deep relative geologic time:a deep learning method for simultaneously interpreting 3‐D seismic horizons and faults[J].Journal of Geophysical Research.Solid Earth,2021,126(9).DOI:10.1029/2021JB021882.
[52]	ALFARHAN M,DERICHE M,MAALEJ A,et al.Multiple events detection in seismic structures using a novel U-Net variant[C].2020 IEEE International Conference on Image Processing (ICIP),Abu Dhabi,United Arab Emirates,2020,2900-2904.