The Inversion Method Based on CNN-BiLSTM-Attention for SQUID TEM Data with IP Effect

Authors

  • Yanju Ji College of Instrumentation and Electrical Engineering Jilin University, Changchun 130026, China, State Key Laboratory of Deep Earth Exploration and Imaging Jilin University, Changchun 130026, China
  • Jinxiu Yuan College of Instrumentation and Electrical Engineering Jilin University, Changchun 130026, China
  • Hui Luan College of Instrumentation and Electrical Engineering Jilin University, Changchun 130026, China, State Key Laboratory of Deep Earth Exploration and Imaging Jilin University, Changchun 130026, China
  • Yuan Wang College of Instrumentation and Electrical Engineering Jilin University, Changchun 130026, China, State Key Laboratory of Deep Earth Exploration and Imaging Jilin University, Changchun 130026, China
  • Qiong Wu College of Instrumentation and Electrical Engineering Jilin University, Changchun 130026, China, State Key Laboratory of Deep Earth Exploration and Imaging Jilin University, Changchun 130026, China

DOI:

https://doi.org/10.13052/2026.ACES.J.410307

Keywords:

Attention mechanism, BiLSTM, CNN, inversion method, polarizable medium, SQUID TEM

Abstract

The superconducting quantum interference device time-domain electromagnetic (SQUID TEM) method has been widely used for the exploration of geological and mineral resources. Extracting resistivity and polarizability from TEM data aids in delineating subsurface metallic mineralization. However, traditional inversion methods are computationally intensive and slow. We propose an inversion method based on a convolutional neural network and bidirectional long short-term memory with attention (CNN-BiLSTM-Attention) to extract resistivity and polarizability of polarizable media from SQUID TEM data acquired with a magnetic source. The method combines the advantages of CNN for automatic feature extraction with the capabilities of BiLSTM for processing temporal data. Additionally, it incorporates an attention mechanism that emphasizes the extraction of key polarization features, thereby optimizing the parameters extraction process. The method can effectively extract resistivity and polarizability from SQUID TEM data. It is validated by the TEM data of theoretical models, and the errors of CNN-BiLSTM-Attention inversion results are smaller than that of the BiLSTM and CNN-LSTM methods.

Downloads

Download data is not yet available.

Author Biographies

Yanju Ji, College of Instrumentation and Electrical Engineering Jilin University, Changchun 130026, China, State Key Laboratory of Deep Earth Exploration and Imaging Jilin University, Changchun 130026, China

Yanju Ji received the M.S. degree in measurement technology and instruments and the Ph.D. degree in earth exploration and information techniques from Jilin University, Changchun, China, in 1998 and 2004, respectively. From 2004 to 2009, she was an Associate Professor with Jilin University. Since 2010, she has been with Jilin University, where she is currently a Professor of instrument science and technology. She has authored or co-authored more than 200 articles. Her research interests include computational electromagnetics, inverse problems, and electromagnetic detecting instrument.

Jinxiu Yuan, College of Instrumentation and Electrical Engineering Jilin University, Changchun 130026, China

Jinxiu Yuan is an undergraduate student at Jilin University, pursuing a degree in measurement and control technology and instrumentation. His research interests include, but are not limited to, machine learning, neural network inversion methods, and related fields. As an emerging researcher, he actively explores the application of advanced computational techniques to scientific problems.

Hui Luan, College of Instrumentation and Electrical Engineering Jilin University, Changchun 130026, China, State Key Laboratory of Deep Earth Exploration and Imaging Jilin University, Changchun 130026, China

Hui Luan received the Ph.D. degree in microwave remote sensing from the Chinese Academy of Science, Beijing, China, in 2007. Since 2007, she has been with the College of Instrumentation and Electrical Engineering, Jilin University, Changchun, where she is currently a Professor. Her research interests include the development of transient electromagnetic instruments and electromagnetic numerical simulation.

Yuan Wang, College of Instrumentation and Electrical Engineering Jilin University, Changchun 130026, China, State Key Laboratory of Deep Earth Exploration and Imaging Jilin University, Changchun 130026, China

Yuan Wang is a Professor with the College of Instrumentation and Electrical Engineering, Jilin University, Changchun, China. He specializes in the development of geophysical exploration instruments, EM instruments, and marine exploration instruments. His research interests include electrical signal detection.

Qiong Wu, College of Instrumentation and Electrical Engineering Jilin University, Changchun 130026, China, State Key Laboratory of Deep Earth Exploration and Imaging Jilin University, Changchun 130026, China

Qiong Wu received the B.S. degree in electrical engineering and automation from Jilin University, Changchun, Jilin Province, China, in 2013, and the Ph.D. degree in detection technology and automatic equipment from Jilin University, in 2019. From 2016 to 2017, she was an Exchange Student with Sustainable Resources Engineering, Faculty of Engineering, Hokkaido University, Sapporo, Japan. From 2019 to 2022, she was a Post-Doctoral Researcher with Jilin University. Since 2023, she is currently an Associate Professor with Jilin University. Her research interest includes the modeling and inversion method for electromagnetic method.

References

M. Radulescu, S. Dalal, U. K. Lilhore, and S. Saimiya, “Optimizing mineral identification for sustainable resource extraction through hybrid deep learning enabled FinTech model,” Resources Policy, vol. 89, Feb. 2024.

G. W. Hohmann, P. R. Kintzinger, G. D. Van Voorhis, and S. H. Ward, “Evaluation of the measurement of induced electrical polarization with an inductive system,” Geophysics, vol. 35, no. 5, pp. 901–915, Oct. 1970.

C. A. Moreira, K. Borssatto, L. M. Ilha, S. F. D. Santos, and F. T. G. Rosa, “Geophysical modeling in gold deposit through DC resistivity and induced polarization methods,” REM-International Engineering Journal, vol. 69, no. 3, pp. 293–299, Sep. 2016.

J. Clarke and A. I. Braginski, The SQUID Handbook: Applications of SQUIDs and SQUID Systems. Hoboken, NJ: John Wiley & Sons, 2006.

C. Flores and S. A. Peralta-Ortega, “Induced polarization with in-loop transient electromagnetic soundings: A case study of mineral discrimination at El Arco porphyry copper, Mexico,” Journal of Applied Geophysics, vol. 68, no. 3, pp. 423–436, July 2009.

I. Kumar, B. V. L. Kumar, S. N. S. Birua, J. K. Dash, and A. K. Chaturvedi, “Inductive induced polarization effect in transient electromagnetic surveys for mapping sulphide-rich zone: A case study from Gurulpada area, Singhbhum Shear Zone, Jharkhand,” Journal of Geophysics, vol. 38, no. 2, Apr. 2017.

T. Lee, “Sign reversals in the transient method of electrical prospecting (one-loop version),” Geophysical Prospecting, vol. 23, no. 4, pp. 653–662, Dec. 1975.

A. Viezzoli, V. Kaminski, and G. Fiandaca, “Modeling induced polarization effects in helicopter time domain electromagnetic data: Synthetic case studies,” Geophysics, vol. 82, no. 2, pp. E31–E50, Feb. 2017.

K. F. Man, C. C. Yin, Y. H. Liu, X. Y. Ren, S. Y. Sun, J. J. Miao, and B. Xiong, “Inversion of timedomain airborne EM data with IP effect based on Pearson correlation constraints,” Applied Geophysics, vol. 17, no. 4, pp. 589–600, 2021.

J. T. Lu, X. B. Wang, Z. W. Xu, M. S. Zhdanov, M. Guo, M. Q. Teng, and Z. Liu, “Quasi-2-D robust inversion of semi-airborne transient electromagnetic data with IP effects,” IEEE Trans. Geosci. Remote Sens., vol. 60, pp. 1–10, Nov. 2022.

X. Wang, J. Lu, M. Guo, S. Zhang, Q. Hu, and H. El-Kaliouby, “Inversion of induced polarizationaffected electrical-source transient electromagnetic data observed in the groundwater survey from eastern Tibet, China,” Geophysics, vol. 90, no. 1, pp. B17–B28, Jan. 2025.

S. Yu and J. Ma, “Deep learning for geophysics: Current and future trends,” Reviews of Geophysics, vol. 59, no. 3, e2021RG000742, Sep. 2021.

T. J. Zhao, S. Wang, C. J. Ouyang, M. Chen, C. Y. Liu, J. Zhang, L. Yu, F. Wang, and Z. Wang, “Artificial intelligence for geoscience: Progress, challenges, and perspectives,” The Innovation, vol. 5, no. 5, p. 100691, Sep. 2024.

Y. J. Ji, Y. Q. Wu, Y. H. Wu, and Y. Zhang, “Excitation process under the ramp-step waveform of inductive source-induced polarization method,” Geophysics, vol. 85, no. 2, pp. E57–E65, Feb. 2020.

T. Alyousuf and Y. Li, “Inversion using adaptive physics-based neural network: Application to magnetotelluric inversion,” Geophysical Prospecting, vol. 70, no. 7, pp. 1252–1272, Aug. 2022.

P. Li, M. L. Liu, M. Alfarraj, P. Tahmasebi, and D. Grana, “Probabilistic physics-informed neural network for seismic petrophysical inversion,” Geophysics, vol. 89, no. 2, pp. M17–M32, Mar. 2024.

K. S. Cole and R. H. Cole, “Dispersion and absorption in dielectrics: Alternating current characteristics,” The Journal of Chemical Physics, vol. 9, no. 4, pp. 341–351, Apr. 1941.

M. N. Nabighian, “Electromagnetic methods in applied geophysics,” in Investigations in Geophysics, no. 3. Tulsa, OK, USA: Society of Exploration Geophysics, pp. 1-2, 1988.

D. Guptasarma and B. Singh, “New digital linear filters for Hankel J0 and J1 transforms,” Geophys. Prospecting, vol. 45, no. 5, pp. 745–762, Sep. 1997.

M. Seidel and B. Tezkan, “1D Cole-Cole inversion of TEM transients influenced by induced polarization,” J. Appl. Geophys., vol. 138, pp. 220–232, Mar. 2017.

L. Alzubaidi, J. L. Zhang, A. J. Humaidi, and L. Farhan, “Review of deep learning: Concepts, CNN architectures, challenges, applications, future directions,” J. Big Data, vol. 8, no. 1, p. 53, Mar. 2021.

W. J. Lu, J. Z. Li, Y. F. Li, A. J. Sun, and J. Y. Wang, “A CNN-LSTM-based model to forecast stock prices,” Complexity, vol. 2020, no. 18, Nov. 2020.

L. Q. Shan, Y. C. Liu, M. Tang, M. Yang, and X. Y. Bai, “CNN-BiLSTM hybrid neural networks with attention mechanism for well log prediction,” J. Pet. Sci. Eng., vol. 205, p. 108838, Oct. 2021.

D. Soydaner, “Attention mechanism in neural networks: Where it comes and where it goes,” Neural Computing and Applications, vol. 34, no. 16, pp. 13371–13385, May 2022.

L. Huang, J. Qin, Y. Zhou, F. Zhu, L. Liu, and L. Shao, “Normalization techniques in training DNNs: Methodology, analysis and application,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 45, no. 8, pp. 10173–10196, Feb. 2023.

S. L. Zhou and W. Song, “Deep learning-based roadway crack classification using laser-scanned range images: A comparative study on hyperparameter selection,” Autom. Constr., vol. 114, p. 103171, June 2020.

P. Baldi and P. Sadowski, “The dropout learning algorithm,” Artificial Intelligence, vol. 210, pp. 78–122, May 2014.

M. Reyad, A. M. Sarhan, and M. Arafa, “A modified Adam algorithm for deep neural network optimization,” Neural Computing and Applications, vol. 35, no. 23, pp. 17095–17112, Apr. 2023.

H. Ren, D Lei, Q. Y. Di, and R. Wang, “Data discrepancy constraints for 3D airborne transient electromagnetic inversion for induced polarization parameters,” Geophysics, vol. 90, no. 3, pp. E105–E116, May 2025.

H. Z. Cai, M. H. Liu, J. J. Zhou, J. H. Li, and X. Y. Hu, “Effective 3D-transient electromagnetic inversion using finite-element method with a parallel direct solver,” Geophysics, vol. 87, no. 6, pp. E377–E392, Nov. 2022.

Downloads

Published

2026-03-30

How to Cite

[1]
Y. . Ji, J. . Yuan, H. . Luan, Y. . Wang, and Q. . Wu, “The Inversion Method Based on CNN-BiLSTM-Attention for SQUID TEM Data with IP Effect”, ACES Journal, vol. 41, no. 03, pp. 258–270, Mar. 2026.

Issue

2026: Vol 41 Iss 3

Section

General Submission

Categories