High-density resistivity method is widely used in engineering exploration because of its efficient and intuitive features. However, due to the high nonlinearity of the inversion problem, the traditional inversion method has some inaccuracy in describing the boundary of anomalous body. In order to achieve high precision two-dimensional nonlinear inversion imaging with high-density electrical method, to overcome the problem that a large number of saddle points in the parameter space of loss function of BP algorithm affect the calculation accuracy and that it is difficult to assign optimal weight threshold to BP network due to the precocious convergence of general genetic algorithm. In this paper,an Optimum Maintaining Adaptive Genetic Algorithm(OMAGA)is proposed to optimize the BP neural network for high density electrical two-dimensional inversion imaging. Good results have been obtained for the inversion calculation of simulation model data and measured data through this method, it shows that this method has strong generalization ability and high inversion accuracy. This study is helpful for the accurate inversion of high density resistivity method in the future,it is helpful to improve the accuracy of underground target identification.
刘湘浩, 刘四新, 胡铭奇, 孙中秋, 王千. 基于OMAGA-BP算法的高密度电阻率法反演研究[J]. 物探与化探, 2023, 47(6): 1519-1527.
LIU Xiang-Hao, LIU Si-Xin, HU Ming-Qi, SUN Zhong-Qiu, WANG Qian. Research on inversion of high-density resistivity method based on OMAGA-BP algorithm. Geophysical and Geochemical Exploration, 2023, 47(6): 1519-1527.
He M G, Wen Y H. The application of high density resistivity 2D inversion to engineering exploration[J]. Geophysical and Geochemical Exploration, 2002, 26(2):156-159.
Lyu H J, Liu S H, Liu B G. Application of resistivity tomography survey method in detecting ground subsidence[J]. Progress in Geophysics, 2005, 20(2):381-386.
[5]
Loke M H, Barker R D. Least-squares deconvolution of apparent resistivity pseudosections[J]. Geophysics, 1995, 60(6):1682-1690.
doi: 10.1190/1.1443900
Xu H L, Wu X P. 2-D resistivity inversion using the neural network method[J]. Chinese Journal of Geophysics, 2006, 49(2):584-589.
[7]
Raiche A. A pattern recognition approach to geophysical inversion using neural nets[J]. Geophysical Journal International, 1991, 105(3):629-648.
doi: 10.1111/j.1365-246X.1991.tb00801.x
[8]
Calderón-Macías C, Sen M K, Stoffa P L. Artificial neural networks for parameter estimation in geophysics[J]. Geophysical Prospecting, 2000, 48(1):21-47.
doi: 10.1046/j.1365-2478.2000.00171.x
[9]
El-Qady G, Ushijima K. Inversion of DC resistivity data using neural networks[J]. Geophysical Prospecting, 2001, 49(4):417-430.
doi: 10.1046/j.1365-2478.2001.00267.x
[10]
Stephen J, Manoj C, Singh S B. A direct inversion scheme for deep resistivity sounding data using artificial neural networks[J]. Journal of Earth System Science, 2004, 113(1):49-66.
doi: 10.1007/BF02701998
[11]
Mann C J H. Geophysical applications of artificial neural networks and fuzzy logic[J]. Kybernetes, 2006, 35(3/4):599-600.
[12]
Neyamadpour A, Taib S, Wan-Abdullah W A T W. Using artificial neural networks to invert 2D DC resistivity imaging data for high resistivity contrast regions: A MATLAB application[J]. Computers & Geosciences, 2009, 35(11):2268-2274.
doi: 10.1016/j.cageo.2009.04.004
[13]
Neyamadpour A, Wan-Abdullah W A T, Taib S. Inversion of quasi-3D DC resistivity imaging data using artificial neural networks[J]. Journal of Earth System Science, 2010, 119(1): 27-40.
doi: 10.1007/s12040-009-0061-2
[14]
Neyamadpour A, Wan-Abdullah W A T, Taib S, et al. 3D inversion of DC data using artificial neural networks[J]. Studia Geophysica et Geodaetica, 2010, 54(3):465-485.
doi: 10.1007/s11200-010-0027-5
[15]
Maiti S, Gupta G, Erram V C,et.al. Inversion of Schlumberger resistivity sounding data from the critically dynamic Koyna region using the Hybrid Monte Carlo-based neural network approach[J]. Nonlinear Processes in Geophysics, 2011, 18(2):179-192.
doi: 10.5194/npg-18-179-2011
[16]
张凌云. 高密度电阻率勘探反演的非线性方法研究[D]. 太原: 太原理工大学, 2011.
[16]
Zhang L Y. The study of nonlinear inversion method in high-density resistivity method inversion[D]. Taiyuan: Taiyuan University of Technology, 2011.
Dai Q W, Jiang F B. Nonlinear inversion for electrical resistivity tomography based on chaotic oscillation PSO-BP algorithm[J]. The Chinese Journal of Nonferrous Metals, 2013, 23(10):2897-2904.
[18]
Dai Q W, Jiang F B, Dong L. Nonlinear inversion for electrical resistivity tomography based on chaotic DE-BP algorithm[J]. Journal of Central South University, 2014, 21(5):2018-2025.
doi: 10.1007/s11771-014-2151-9
Gao M L, Yu S B, Zheng J B, et al. Research of resistivity imaging using neural network based on immune genetic algorithm[J]. Chinese Journal of Geophysics, 2016, 59(11):4372-4382.
[20]
Wu X X, Liu J G. A new early stopping algorithm for improving neural network generalization[C]// 2009 Second International Conference on Intelligent Computation Technology and Automation.Changsha,China,IEEE, 2009:15-18.
[21]
Dauphin Y N, Pascanu R, Gulcehre C, et al. Identifying and attacking the saddle point problem in high-dimensional non-convex optimization[C]// Proceedings of the 27th International Conference on Neural Information Processing Systems, 2014.
Hun W M, Xi Y G. The analysis of global convergence and computational efficiency for genetic algorithm[J]. Control Theory & Applications, 1996, 13(4):455-460.
[23]
Chen L, Gallet A, Huang S S, et al. Probabilistic cracking prediction via deep learned electrical tomography[J]. Structural Health Monitoring, 2022, 21(4): 1574-1589.
doi: 10.1177/14759217211037236
[24]
Bishop C M. Training with noise is equivalent to tikhonov regularization[J]. Neural Computation, 1995, 7(1):108-116.
doi: 10.1162/neco.1995.7.1.108
[25]
An Z. The effects of adding noise during backpropagation training on a generalization performance[J]. Neural Computation, 1996, 8(3):643-674.
doi: 10.1162/neco.1996.8.3.643