Data Reconstruction of Direct Current Resistivity Method Based on U-Net Convolutional Neural Network
-
摘要:
本文针对地球物理电阻率法实测数据采集中因人文噪声引起的数据突变问题，提出了一种基于卷积神经网络（CNN）的数据重构算法。鉴于电阻率法控制方程的非线性特征，传统的线性插值技术可能导致反演结果精度下降。本研究首先构建了CNN模型，并通过三维有限差分法进行电阻率法正演数值模拟，生成训练集和测试集。利用训练集对CNN模型进行训练，并基于损失函数结果优化U-Net网络参数。通过对比线性插值技术和CNN重构合成突变数据的反演结果，验证了CNN在数据重构中的有效性和优越性。研究结果表明，U-Net-CNN可以有效重构非线性直流电阻率法数据，为提高地球物理数据采集精度和反演结果的可靠性提供了新的技术途径。
Abstract:In this paper, a data reconstruction algorithm based on Convolutional Neural Networks (CNN) is proposed for the issue of data mutation caused by humanistic noise in the geophysical resistivity method data acquisition. Given the nonlinear characteristics of the resistivity method control equation, the traditional linear interpolation techniques may lead to a decline in the accuracy of the inversion results. In this study, a CNN model is first constructed and numerical simulation of the resistivity method inversion is carried out by the three-dimensional finite difference method to generate a training dataset and a test dataset. The CNN model is trained using the training set and the U-Net network parameters are optimized based on the loss function results. By comparing the inversion results of data reconstructed with linear interpolation technique and CNN reconstruction of synthetic mutant data, the effectiveness and superiority of CNN in data reconstruction are verified. The results demonstrate that U-Net-CNN can effectively reconstruct nonlinear direct current resistivity method data, providing a new technical approach to improve the accuracy of geophysical data acquisition and the reliability of inversion results.
-
-
表 1 U-Net模型的5次超参数调整设置
Table 1. 5 hyperparameter adjustment settings for the U-Net model
调整次数（次） 学习次数（次） 学习误差 步长 超参数β1 1 1000 0.001 10 0.9 2 800 0.001 10 0.9 3 1000 0.001 20 0.9 4 600 0.001 20 0.9 5 600 0.0001 20 0.9 
下载: 导出CSV
-
[1] 何 睿,王 润,徐 航,刘 帅,李彧磊,张 硕,陈 琨,蔡 宇,陈梦源.2024.基于U-Net的湖北省露天矿山土地损毁信息提取应用[J]. 资源环境与工程,38(1):100-110.
[2] 胡 阳,汪玲玲,陈海洋,杜子豪. 2022. 基于SE V-Net的三维地震数据断层识别方法[C]. 2022年中国石油物探学术年会论文集(下册), 514-517.
[3] 李 明,蒋委君,董佳慧,金少锋,张宸伟,牛瑞卿.2023.基于机器学习的滑坡灾害易发性评价——以三峡库区为例[J]. 华南地质,39(3):413-427.
[4] 齐 琪,苏英峰,王 晶.2024.基于U-net上气道流体力学分析研究[J]. 医用生物力学,39(S1):422.
[5] 邵才瑞,关 丽,张福明.2005.基于测井数据的地质曲面插值重构方法比较[J]. 测井技术,29(4):311-315+388.
[6] 唐云飞,但志平,洪郑天,陈永麟,程沛霖,成 果,刘芳婷.2024.CAMU-Net:基于Attention U-Net的视网膜血管分割改进模型[J]. 中国医学物理学杂志,41(8):960-968.
[7] 陶 涛,韩 鹏,马 欢,谭捍东.2024.时移电阻率法三维反演研究[J]. 地球物理学报,67(10):3973-3988. doi: 10.6038/cjg2023R0297
[8] 王 方,熊 杰,田慧潇,李思平,康佳帅.2024.基于深度学习的大地电磁二维反演方法[J]. 地质科技通报,43(2):344-354.
[9] 吴佳伟. 2023. 基于径向基函数的大地电磁数据重构[D]. 中国地质大学(北京)硕士学位论文.
[10] 邢 强,韩爱香.2013.基于压缩感知的过套管电阻率数据重构方法研究[J]. 石油仪器,27(1):46-48+9.
[11] 颜 钟. 2012. 基于BP人工神经网络的高密度电法反演理论应用研究[D]. 长江科学院硕士学位论文.
[12] 俞建宝,陈 高,李正文,张 杨.2009.基于双谱信号重构的物探数据插值方法[J]. 勘探地球物理进展,32(5):326-329+307.
[13] 张 岩,周一帆,宋利伟,董宏丽.2023.基于物理约束U-Net网络的地震数据低频延拓[J]. 石油地球物理勘探,58(1):31-45.
[14] 张 喆,马福恒,霍吉祥.2024.基于BP神经网络的高密度电法在水库清淤扩容坝后排泥区围堰探测中的应用[J]. 水电能源科学,42(5):174-178.
[15] 张瀚文,曹维娟,罗刚银,江 浩,邱 香,许 杰,史蓉蓉,郑 然.2024.基于改进型U-Net的变色油墨血浆判别模型[J]. 南京医科大学学报(自然科学版),44(9):1179-1189. doi: 10.7655/NYDXBNSN231104
[16] Du H K, Ren Z Y, Tang J T. 2016. A finite-volume approach for 2D magnetotellurics modeling with arbitrary topographies[J]. Studia Geophysica et Geodaetica, 60(2): 332-347. doi: 10.1007/s11200-014-1041-9
[17] Jiang F B, Dong L, Dai Q W. 2018. Electrical resistivity imaging inversion: An ISFLA trained kernel principal component wavelet neural network approach[J]. Neural Networks, 104: 114-123. doi: 10.1016/j.neunet.2018.04.012
[18] Lecun Y, Bottou L, Bengio Y, Haffner P. 1998. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE, 86(11): 2278-2324. doi: 10.1109/5.726791
[19] Loke M H, Chamvers J E, Rucker D F, Kuras O, Wilkinson P B. 2013. Recent developments in the direct-current geoelectrical imaging method[J]. Journal of Applied Geophysics, 95: 135-156. doi: 10.1016/j.jappgeo.2013.02.017
[20] Ren Z Y, Tang J T. 2010. 3D direct current resistivity modeling with unstructured mesh by adaptive finite-element method[J]. Geophysics: Journal of the Society of Exploration Geophysicists, 75(1): H7-H17.
[21] Wu X P, Xiao Y F, Qi C, Wang T T. 2003. Computations of secondary potential for 3D DC resistivity modelling using an incomplete Choleski conjugate - gradient method[J]. Geophysical Prospecting, 51(6): 567-577. doi: 10.1046/j.1365-2478.2003.00392.x
-
图(10)
表(1)
计量
- 文章访问数: 55
- PDF下载数: 11
- 施引文献: 0