《计算机应用》唯一官方网站 ›› 2024, Vol. 44 ›› Issue (7): 2301-2309.DOI: 10.11772/j.issn.1001-9081.2023060808
收稿日期:
2023-06-26
修回日期:
2023-08-20
接受日期:
2023-08-24
发布日期:
2023-08-28
出版日期:
2024-07-10
通讯作者:
余瀚
作者简介:
孙明皓(1998—),男,江苏南京人,硕士研究生,主要研究方向:信号与图像处理;基金资助:
Minghao SUN1, Han YU1(), Yuqing CHEN2, Kai LU3
Received:
2023-06-26
Revised:
2023-08-20
Accepted:
2023-08-24
Online:
2023-08-28
Published:
2024-07-10
Contact:
Han YU
About author:
SUN Minghao, born in 1998, M. S. candidate. His research interests include signal and image processing.Supported by:
摘要:
针对传统勘探地震波初至拾取工作量大、抗噪性差和精度低所导致的低质量速度反演影响生产安全的问题,提出一种基于U形多层感知机(U-MLP)网络的地震波初至拾取与反演方法。首先,为解决传统U形网络(U-Net)中的交叉熵损失函数在数据类别不平衡时导致的性能变差问题,设计一种基于加权交叉熵Lovász归一化指数(WLS)的损失函数;然后,在特征融合阶段引入残差连接,缩小低级特征与高级特征间的差距,还原更多细节信息;最后,为使U-MLP网络更好学习图像局部特征,为高级语义引入标记化的多层感知机(MLP)模块,此模块降低了参数量和计算复杂度。实验结果表明,与U-Net相比,U-MLP网络在训练中收敛性更强,初至拾取最大误差降低了20%以上,交并比(IoU)值提升了约2%。可见,U-MLP网络在提取勘探地震波初至时不仅提高了拾取精度,而且拾取的初至在仿真数据和实际数据中的速度分布反演均达到了理想效果,具有更好的性能且适应性更强。
中图分类号:
孙明皓, 余瀚, 陈雨青, 陆恺. 基于U形多层感知机网络的地震波初至拾取与反演[J]. 计算机应用, 2024, 44(7): 2301-2309.
Minghao SUN, Han YU, Yuqing CHEN, Kai LU. First-arrival picking and inversion of seismic waveforms based on U-shaped multilayer perceptron network[J]. Journal of Computer Applications, 2024, 44(7): 2301-2309.
数据集 | 模型 | Max_diff_fb | Min_diff_fb | Mean_diff_fb |
---|---|---|---|---|
训练集 | U-Net | 20.45 | 0.500 0 | 6.397 |
U-MLP网络 | 15.60 | 0.350 0 | 5.213 | |
验证集 | U-Net | 20.23 | 0.416 7 | 9.458 |
U-MLP网络 | 15.14 | 0.364 7 | 8.438 | |
测试集 | U-Net | 20.16 | 0.784 3 | 10.180 |
U-MLP网络 | 15.84 | 0.571 1 | 8.279 |
表1 不同网络模型的初至拾取误差对比
Tab. 1 Error comparison of first arrivals picked by different network models
数据集 | 模型 | Max_diff_fb | Min_diff_fb | Mean_diff_fb |
---|---|---|---|---|
训练集 | U-Net | 20.45 | 0.500 0 | 6.397 |
U-MLP网络 | 15.60 | 0.350 0 | 5.213 | |
验证集 | U-Net | 20.23 | 0.416 7 | 9.458 |
U-MLP网络 | 15.14 | 0.364 7 | 8.438 | |
测试集 | U-Net | 20.16 | 0.784 3 | 10.180 |
U-MLP网络 | 15.84 | 0.571 1 | 8.279 |
数据集 | 模型 | IoU指标 |
---|---|---|
训练集 | U-Net | 0.975 6 |
U-MLP网络 | 0.994 5 | |
验证集 | U-Net | 0.973 2 |
U-MLP网络 | 0.994 1 | |
测试集 | U-Net | 0.975 4 |
U-MLP网络 | 0.993 9 |
表2 不同网络模型下的IoU结果对比
Tab. 2 IoU comparison between different network models
数据集 | 模型 | IoU指标 |
---|---|---|
训练集 | U-Net | 0.975 6 |
U-MLP网络 | 0.994 5 | |
验证集 | U-Net | 0.973 2 |
U-MLP网络 | 0.994 1 | |
测试集 | U-Net | 0.975 4 |
U-MLP网络 | 0.993 9 |
1 | ARAYA-POLO M, JENNINGS J, ADLER A, et al. Deep-learning tomography [J]. The Leading Edge, 2018, 37(1): 58-66. |
2 | BOSCHETTI F, DENTITH M D, LIST R D. A fractal-based algorithm for detecting first arrivals on seismic traces [J]. Geophysics, 1996, 61(4): 1095-1102. |
3 | ALLEN R V. Automatic earthquake recognition and timing from single traces [J]. Bulletin of the Seismological Society of America, 1978, 68(5): 1521-1532. |
4 | 潘树林,高磊,陈辉,等.可控震源地震记录初至拾取方法研究[J].石油物探, 2010, 49(2): 209-212. |
PAN S L, GAO L, CHEN H, et al. Study on methods for picking up first break of seismic record acquired by vibroseis [J]. Geophysical Prospecting for Petroleum, 2010, 49(2): 209-212. | |
5 | GACI S. The use of wavelet-based denoising techniques to enhance the first-arrival picking on seismic traces [J]. IEEE Transactions on Geoscience and Remote Sensing, 2014, 52(8): 4558-4563. |
6 | 许银坡,杨海申,杨剑,等.初至波能量比迭代拾取方法[J].地球物理学进展, 2016, 31(2): 845-850. |
XU Y P, YANG H S, YANG J, et al. Iteration pickup method of first break using energy ratio [J]. Progress in Geophysics, 2016, 31(2): 845-850. | |
7 | AKRAM J, EATON D W. A review and appraisal of arrival-time picking methods for downhole microseismic data [J]. Geophysics, 2016, 81(2): KS71-KS91. |
8 | HOLLANDER Y, MEROUANE A, YILMAZ O. Using a deep convolutional neural network to enhance the accuracy of first-break picking [C]// Proceedings of the 88th SEG International Exposition and Annual Meeting. Tulsa, Oklahoma: Society of Exploration Geophysicists, 2018: 4628-4632. |
9 | CHEN Y. Automatic microseismic event picking via unsupervised machine learning [J]. Geophysical Journal International, 2020, 222(3): 1750-1764. |
10 | KRIZHEVSKY A, SUTSKEVER I, HINTON G E. ImageNet classification with deep convolutional neural networks [J]. Communications of the ACM, 2017, 60(6): 84-90. |
11 | LeCUN Y, BENGIO Y, HINTON G. Deep learning [J]. Nature, 2015, 521(7553): 436-444. |
12 | LONG J, SHELHAMER E, DARRELL T. Fully convolutional networks for semantic segmentation [C]// Proceedings of the 2015 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2015: 3431-3440. |
13 | RONNEBERGER O, FISCHER P, BROX T. U-Net: convolutional networks for biomedical image segmentation [C]// Proceedings of the 2015 Medical Image Computing and Computer-Assisted Intervention. Cham: Springer, 2015: 234-241. |
14 | WU H, ZHANG B, LI F, et al. Semiautomatic first-arrival picking of microseismic events by using the pixel-wise convolutional image segmentation method [J]. Geophysics, 2019, 84(3): V143-V155. |
15 | 丁建群,何永清,郭锐,等.基于U-Net网络的地震数据初至自动拾取方法[C]// 2019油气地球物理学术年会论文集.南京:中国地球物理学会油气地球物理专业委员会, 2019: 132-136. |
DING J Q, HE Y Q, GUO R, et al. Automatic first break pickup method of seismic data based on U-Net network [C]// Proceedings of the 2019 Annual Oil and Gas Geophysics Conference. Nanjing: Chinese Geophysical Society, Committee on Oil and Gas Geophysics, 2019: 132-136. | |
16 | 李薇薇,龚仁彬,周相广,等.基于深度学习UNet++网络的初至波拾取方法[J].地球物理学进展, 2021, 36(1): 187-194. |
LI W W, GONG R B, ZHOU X G, et al. UNet++: a deep-neural-network-based seismic arrival time picking method [J]. Progress in Geophysics, 2021, 36(1): 187-194. | |
17 | 刘志文,王进强,王广鑫.基于UNet++的地震P波初至拾取研究[J].太原理工大学学报, 2023, 54(1): 65-72. |
LIU Z W, WANG J Q, WANG G X. Research on first-arrival picking of seismic P-wave based on UNet++ [J]. Journal of Taiyuan University of Technology, 2023, 54(1): 65-72. | |
18 | 胡婷,徐彬,王永发,等.基于U-Net的矿山微震初至拾取研究[J].科学技术与工程, 2023, 23(16): 6802-6809. |
HU T, XU B, WANG Y F, et al. Arrival picking of mine microseismic events using U-Net [J]. Science Technology and Engineering, 2023, 23(16): 6802-6809. | |
19 | BERMAN M, TRIKI A R, BLASCHKO M B. The Lovász-Softmax loss: a tractable surrogate for the optimization of the intersection-over-union measure in neural networks [C]// Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2018: 4413-4421. |
20 | HE K, ZHANG X, REN S, et al. Deep residual learning for image recognition [C]// Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2016: 770-778. |
21 | TOLSTIKHIN I O, HOULSBY N, KOLESNIKOV A, et al. MLP-Mixer: an all-MLP architecture for vision [J]. Advances in Neural Information Processing Systems, 2021, 34: 24261-24272. |
22 | TOUVRON H, BOJANOWSKI P, CARON M, et al. ResMLP: feedforward networks for image classification with data-efficient training [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023, 45(4): 5314-5321. |
23 | YU T, LI X, CAI Y, et al. S2-MLP: Spatial-shift MLP architecture for vision [C]// Proceedings of the 2022 IEEE/CVF Winter Conference on Applications of Computer Vision. Piscataway: IEEE, 2022: 3615-3624. |
24 | LIAN D, YU Z, SUN X, et al. AS-MLP: an axial shifted MLP architecture for vision [EB/OL]. (2022-05-17) [2023-05-21]. . |
25 | 南京邮电大学.一种基于U-MLP网络的勘探数据初至拾取方法: CN202210938581.8 [P]. 2022-11-08. |
Nanjing University of Posts and Telecommunications. A method for picking up first breaks in exploration data based on U-MLP network: CN202210938581.8 [P]. 2022-11-08. | |
26 | CAO H, WANG Y, CHEN J, et al. Swin-Unet: Unet-like pure Transformer for medical image segmentation [C]// Proceedings of the Computer Vision-ECCV 2022 Workshops. Cham: Springer, 2023: 205-218. |
27 | HOWARD A G, ZHU M, CHEN B, et al. MobileNets: efficient convolutional neural networks for mobile vision applications [EB/OL]. (2017-04-17) [2023-05-27]. . |
28 | XIE E, WANG W, YU Z, et al. SegFormer: simple and efficient design for semantic segmentation with Transformers [J]. Advances in Neural Information Processing Systems, 2021, 34: 12077-12090. |
29 | CHENG J, PAN S. U-Net first-arrival picking optimization method based on traces connectivity [C]// Proceedings of the SEG 4th International Workshop on Mathematical Geophysics: Traditional & Learning. Tulsa, Oklahoma: Society of Exploration Geophysicists, 2022: 119-121. |
[1] | 力尚龙, 刘建华, 贾鹤鸣. 融合多狩猎协调策略的爬行动物搜索算法[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2818-2828. |
[2] | 唐廷杰, 黄佳进, 秦进, 陆辉. 基于图共现增强多层感知机的会话推荐[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2357-2364. |
[3] | 沈嫣然, 温昕, 张瑾昊, 张帅, 曹锐, 高保禄. 轻量级多尺度卷积网络的功能磁共振成像脑龄预测模型[J]. 《计算机应用》唯一官方网站, 2024, 44(12): 3949-3957. |
[4] | 仇丽青, 苏小盼. 个性化多层兴趣提取点击率预测模型[J]. 《计算机应用》唯一官方网站, 2024, 44(11): 3411-3418. |
[5] | 周辉, 陈玉玲, 王学伟, 张洋文, 何建江. 基于生成对抗网络的联邦学习深度影子防御方案[J]. 《计算机应用》唯一官方网站, 2024, 44(1): 223-232. |
[6] | 陈蒙蒙, 乔志伟. 基于融合通道注意力的Uformer的CT图像稀疏重建[J]. 《计算机应用》唯一官方网站, 2023, 43(9): 2948-2954. |
[7] | 李淦, 牛洺第, 陈路, 杨静, 闫涛, 陈斌. 融合视觉特征增强机制的机器人弱光环境抓取检测[J]. 《计算机应用》唯一官方网站, 2023, 43(8): 2564-2571. |
[8] | 刘阳, 陆志扬, 王骏, 施俊. 基于自注意力连接UNet的磁共振成像去吉布斯伪影算法[J]. 《计算机应用》唯一官方网站, 2023, 43(5): 1606-1611. |
[9] | 罗开, 陈亮, 梁巍, 陈勇强. 基于稀疏表示的滤波反投影超声层析成像重建算法[J]. 《计算机应用》唯一官方网站, 2023, 43(3): 903-908. |
[10] | 韩泽芳, 张雄, 上官宏, 韩兴隆, 韩静, 奉刚, 崔学英. 用于低剂量CT降噪的伪影感知生成对抗网络[J]. 《计算机应用》唯一官方网站, 2022, 42(7): 2301-2310. |
[11] | 贾洪飞, 刘茜, 王瑜, 肖洪兵, 邢素霞. 3DPCANet在阿尔茨海默症功能磁共振成像图像分类中的应用[J]. 《计算机应用》唯一官方网站, 2022, 42(1): 310-315. |
[12] | 吴洁, 张师天, 谢海滨, 杨光. 基于多影像中心磁共振成像数据的半监督膝盖异常分类[J]. 《计算机应用》唯一官方网站, 2022, 42(1): 316-324. |
[13] | 张笑非, 杨阳, 黄佳进, 钟宁. 基于度中心性的认知特征选择方法[J]. 计算机应用, 2021, 41(9): 2767-2772. |
[14] | 祝承, 赵晓琦, 赵丽萍, 焦玉宏, 朱亚飞, 陈建英, 周伟, 谭颖. 基于谱聚类半监督特征选择的功能磁共振成像数据分类[J]. 计算机应用, 2021, 41(8): 2288-2293. |
[15] | 黄梨, 卢龙. 基于长距离依赖编码与深度残差U-Net的缺血性卒中病灶分割[J]. 计算机应用, 2021, 41(6): 1820-1827. |
阅读次数 | ||||||
全文 |
|
|||||
摘要 |
|
|||||