Fault prediction method for coalfields based on an SVD–ResCBAM–U-Net framework

Abstract

Accurate fault prediction in coalfield seismic data is important for geological interpretation and the safe and efficient exploitation of coal resources. However, conventional fault interpretation methods and shallow machine-learning approaches usually rely on manually extracted seismic attributes. They often show limited robustness to noise and insufficient capability in characterizing fault continuity, boundary features, and small faults in structurally complex areas. To overcome these limitations, a fault prediction method based on the singular value decomposition–residual convolutional block attention module–U-Net (SVD–ResCBAM–U-Net) framework is proposed. First, SVD was used to denoise the seismic data and improve its quality. Then, residual blocks and a convolutional block attention module were incorporated into the U-Net architecture to enhance fault-related feature extraction and improve prediction performance. Experimental results show that the proposed SVD–ResCBAM–U-Net achieved the best performance among all compared models, with a global accuracy of 0.9556, a mean intersection over union of 0.6241, and a mean boundary F1-score of 0.6796. These results clearly demonstrate the proposed method’s advantages in fault continuity, boundary delineation, and small-fault prediction, underscoring its effectiveness for fault prediction in coalfield seismic data under complex geological conditions.

Keywords

Seismic interpretation / U-Net / Fault prediction / Singular value decomposition / Coal mine

References

[1] Wang H, Zhang D. Examining the interplay between fossil fuel mining, sustainable growth, and economic prosperity. Resour Policy . 2023; 87: 104324. doi: 10.1016/j.resourpol.2023.104324

[2] Odintsov E, Zhao Z, Gusev V, et al. Integrated physical and numerical assessment of the formation of water—conducting fracture zones in deep ore mines with structural faults. Mining. 2026; 6(1): 10. doi: 10.3390/mining6010010

[3] Zhu G, Wang S, Zhang W, et al. Research on the mechanism and evolution law of delayed water inrush caused by fault activation with mining. Water. 2023; 15(24): 4209. doi: 10.3390/w15244209

[4] Wang C, Ge C, Zhou B, et al. Dynamic damage law of coals under stress paths for disaster inoculation and its influence mechanism on outburst risk level. Nat Resour Res. 2026; 35(2): 1317-1347. doi: 10.1007/s11053—025—10585—9

[5] Yan J, Feng X, Guo Y, et al. Discussion on the main control effect of geological structures on coal and gas outburst. ACS Omega. 2023; 8(1): 835-845. doi: 10.1021/acsomega.2c06200

[6] Wang H, Wang L, Cheng Y, et al. Characteristics and dominant controlling factors of gas outburst in Huaibei coalfield and its countermeasures. Int J Min Sci Technol. 2013; 23(4): 591-596. doi: 10.1016/j.ijmst.2013.07.019

[7] Zhou B, Hatherly P, Sun W. Enhancing the detection of small coal structures by seismic diffraction imaging. Int J Coal Geol. 2017; 178: 1-12. doi: 10.1016/j.coal.2017.04.010

[8] Gou R, Song D, He X, et al. Microseismic response characteristics and stress anomaly zoning in deep outburst—prone coal seams with multi—concealed structures: A case study. J Appl Geophys. 2025; 242: 105932. doi: 10.1016/j.jappgeo.2025.105932

[9] Bahorich MS, Haskell NL, Nissen SE, et al. Stratigraphic and structural interpretation with 3—D coherence. In: SEG Technical Program Expanded Abstracts. Houston, TX: Society of Exploration Geophysicists; 1995: 97-100. doi: 10.1190/1.1887435

[10] Marfurt KJ, Kirlin RL, Farmer SL, et al. 3—D seismic attributes using a semblance—based coherency algorithm. Geophysics. 1998; 63(4): 1150-1165. doi: 10.1190/1.1444415

[11] Gao D. Integrating 3D seismic curvature and curvature gradient attributes for fracture characterization: Methodologies and interpretational implications. Geophysics. 2013; 78(2): O21-O31. doi: 10.1190/geo2012—0190.1

[12] Cao L, Yao Y, Liu D, et al. Application of seismic curvature attributes in the delineation of coal texture and deformation in Zhengzhuang field, southern Qinshui Basin. AAPG Bull. 2020; 104(5): 1143-1166. doi: 10.1306/12031918208

[13] Jingbin C, Zhonghong W, Ping C, et al. The application of seismic attribute analysis technique in coal field exploration. Interpretation. 2016; 4(1): SB13-SB21. doi: 10.1190/INT—2015—0090.1

[14] Khan M, Bery AA, Bashir Y, et al. Automated fault network extraction in complex tectonic regimes: A hybrid machine learning and structural attributes approach. Appl Comput Geosci. 2025; 27: 100264. doi: 10.1016/j.acags.2025.100264

[15] Gui Z, Zhang J, Zhang Y, et al. Characterization of fault—karst reservoirs based on deep learning and attribute fusion. Acta Geophys. 2024; 73(2): 1335-1347. doi: 10.1007/s11600—024—01420—5

[16] Yasin Q, Majdański M, Sohail GM, et al. Fault and fracture network characterization using seismic data: a study based on neural network models assessment. Geomech Geophys Geo—Energy Geo—Resour. 2022; 8(2): 41. doi: 10.1007/s40948—022—00352—y

[17] Zou G, Ren K, Sun Z, et al. Fault interpretation using a support vector machine: A study based on 3D seismic mapping of the Zhaozhuang coal mine in the Qinshui Basin, China. J Appl Geophys. 2019; 171: 103870. doi: 10.1016/j.jappgeo.2019.103870

[18] Ren K, Zou G, Peng S, et al. Fault identification based on the kernel principal component analysis—genetic particle swarm optimization—support vector machine algorithm for seismic attributes in the Sihe Coal Mine, Qinshui Basin, China. Interpretation. 2023; 11(1): T59-T73. doi: 10.1190/INT—2022—0039.1

[19] Han C, Zou G, Yeh HG, et al. Intelligent fault prediction with wavelet—SVM fusion in coal mine. Comput Geosci. 2025; 194: 105744. doi: 10.1016/j.cageo.2024.105744

[20] Li D, Peng S, Lu Y, et al. Seismic structure interpretation based on machine learning: A case study in coal mining. Interpretation. 2019; 7(3): SE69-SE79. doi: 10.1190/INT—2018—0208.1

[21] Tsinidis G, Pitilakis K. Improved R—F relations for the transversal seismic analysis of rectangular tunnels. Soil Dyn Earthq Eng. 2018; 107: 48-65. doi: 10.1016/j.soildyn.2018.01.004

[22] Jang J, So BD, Yuen DA. A machine learning algorithm with random forest for recognizing hidden control factors from seismic fault distribution. Geosci J. 2023; 27(1): 113-126. doi: 10.1007/s12303—022—0029—7

[23] Khan MK, Bashir Y, Dossary S, et al. Machine learning and seismic structural attributes hybrid approach to map complex fault system. In: Second EAGE Subsurface Intelligence Workshop. The Netherlands: European Association of Geoscientists & Engineers; 2022: 1-5. doi: 10.3997/2214—4609.2022616010

[24] Imran QS, Siddiqui NA, Latiff AA, et al. Automated fault detection and extraction under gas chimneys using hybrid discontinuity attributes. Appl Sci. 2021; 11(16): 7218. doi: 10.3390/app11167218

[25] Pham N, Fomel S, Dunlap D. Automatic channel detection using deep learning. Interpretation. 2019; 7(3): SE43-SE50. doi: 10.1190/INT—2018—0202.1

[26] Ross ZE, Meier M, Hauksson E, et al. Generalized seismic phase detection with deep learning. Bull Seismol Soc Am. 2018; 108(5A): 2894-2901. doi: 10.1785/0120180080

[27] Feng R. Estimation of reservoir porosity based on seismic inversion results using deep learning methods. J Nat Gas Sci Eng. 2020; 77: 103270. doi: 10.1016/j.jngse.2020.103270

[28] Azizzadeh Mehmandost Olya B, Mohebian R, Moradzadeh A. A new approach for seismic inversion with GAN algorithm. J Seism Explor. 2024; 33: 1-36. doi: 10.36922/JSE024450003.t

[29] Li S, Yang C, Sun H, et al. Seismic fault detection using an encoder—decoder convolutional neural network with a small training set. J Geophys Eng. 2019; 16(1): 175-189. doi: 10.1093/jge/gxy015

[30] Zou G, Liu H, Ren K, et al. Automatic recognition of faults in mining areas based on convolutional neural network. Energies. 2022; 15(10): 3758. doi: 10.3390/en15103758

[31] Wang J, Niu C, Wang Y, et al. Fault identification and enhancement using residual U—Net: Application to field seismic data. J Seism Explor. 2026; 35(1): 025360067. doi: 10.36922/JSE025360067

[32] Xiong W, Ji X, Ma Y, et al. Seismic fault detection with convolutional neural network. Geophysics. 2018; 83(5): O97-O103. doi: 10.1190/geo2017—0666.1

[33] 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. Geophysics. 2019; 84(3): IM35-IM45. doi: 10.1190/geo2018—0646.1

[34] Ronneberger O, Fischer P, Brox T. U—Net: convolutional networks for biomedical image segmentation. arXiv. Preprint posted online 2015. doi: 10.48550/arXiv.1505.04597

[35] Yang D, Cai Y, Hu G, et al. Seismic fault detection based on 3D Unet++ model. In: SEG International Exposition and Annual Meeting. OnePetro; 2020: 1631-1635. doi: 10.1190/segam2020—3426516.1

[36] Yan B, Qian L, Zhao J, et al. Fault Identification Based on W—Net in 3—D Seismic Images. IEEE Geosci Remote Sens Lett. 2024; 21: 1-5. doi: 10.1109/LGRS.2024.3404505

[37] Cui L, Huang Y, Niu Y, et al. MS—Unet: A multi—scale feature fusion U—Net for 3D seismic fault detection. Processes. 2025; 13(7): 1976. doi: 10.3390/pr13071976

[38] Gao K, Huang L, Zheng Y. Fault detection on seismic structural images using a Nested Residual U—Net. IEEE Trans Geosci Remote Sens. 2022; 60: 1-15. doi: 10.1109/TGRS.2021.3073840

[39] Sun Q, Wang X, Ni H, et al. Fault identification of U—Net based on enhanced feature fusion and attention mechanism. Electronics. 2023; 12(12): 2562. doi: 10.3390/electronics12122562

[40] Bekara M, Van Der Baan M. Local singular value decomposition for signal enhancement of seismic data. Geophysics . 2007; 72(2): V59-V65. doi: 10.1190/1.2435967

[41] Freire SLM, Ulrych TJ. Application of singular value decomposition to vertical seismic profiling. Geophysics. 1988; 53(6): 778-785. doi: 10.1190/1.1442513

[42] Eckart C, Young G. The approximation of one matrix by another of lower rank. Psychometrika . 1936; 1(3): 211-218. doi: 10.1007/BF02288367

[43] Siddique N, Sidike P, Elkin CP, et al. U—net and its variants for medical image segmentation: A review of theory and applications. IEEE Access. 2021; 9: 82031-82057. doi: 10.1109/ACCESS.2021.3086020

[44] Bengio Y, Simard P, Frasconi P. Learning long—term dependencies with gradient descent is difficult. IEEE Trans Neural Netw . 1994; 5(2): 157-166. doi: 10.1109/72.279181

[45] Hochreiter S. The vanishing gradient problem during learning recurrent neural nets and problem solutions. Int J Uncertain Fuzziness Knowl Based Syst . 1998; 6(2): 107-116. doi: 10.1142/S0218488598000094

[46] Chen M, Cao H, Jia L. AG—Net3D: embedding attention gate into U—Net for 3D seismic data fault detection. IEEE Access. 2025; 13: 139267-139274. doi: 10.1109/ACCESS.2025.3596339

[47] Perazzi F, Pont—Tuset J, McWilliams B, et al. A benchmark dataset and evaluation methodology for video object segmentation. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. IEEE; 2016: 724-732. doi: 10.1109/CVPR.2016.85