Magnetotelluric deep neural network inversion method based on space mapping technology

A deep neural network and magnetotelluric technology, applied in neural learning methods, biological neural network models, neural architectures, etc., can solve problems such as sharp changes in resistivity values ​​that do not conform to actual geological significance, increased calculations, and increased learning costs

Active Publication Date: 2020-04-07
CHONGQING UNIV +1
View PDF11 Cites 23 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] Although the existing research on using artificial intelligence algorithms to predict the geoelectric structure has achieved some results, it can only satisfy the prediction of the layered geoelectric model with a small number of layers, and the resistivity range is limited, so it cannot be applied to more complex geoelectric structures. Geoelectric structure
The main reasons for this problem are as follows: For the layered geoelectric model, although the parameter inversion imaging established by the artificial intelligence algorithm can obtain layer thickness and resistivity information at the same time, the learning cost increases sharply as the number of layers increases. The amount of calculation in the model also increases sharply; when the number of formation layers is small, the sharp change of the resistivity value does not conform to the actual geological significance
Not only that, the inaccuracy of the electrical parameters of a certain layer has a great influence on the overall geoelectric structure

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Magnetotelluric deep neural network inversion method based on space mapping technology
  • Magnetotelluric deep neural network inversion method based on space mapping technology
  • Magnetotelluric deep neural network inversion method based on space mapping technology

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0090] see Figure 1 to Figure 3 , a magnetotelluric deep neural network inversion method based on space mapping technology, mainly includes the following steps:

[0091] 1) Determine the detection area, that is, the layered geoelectric section.

[0092] 2) Establish geoelectric model sample set A 2 , the main steps are as follows:

[0093] 2.1) Establish a training sample set G based on the resistivity of each layer of geoelectric section, namely:

[0094]

[0095] In the formula, σ 0 and σ 1 Represent the minimum conductivity and maximum conductivity in the training sample set G, respectively. n is the set G capacity. When the number of ground section layers is M, the number of subsets of the training sample set is n M . i represents any sample.

[0096] 2.2) Use the conductivity constrained sampling strategy to simplify the training sample set G, and obtain the simplified training sample set A 1 . Simplify the training sample set A 1 The number of subsets is ...

Embodiment 2

[0170] A method for inversion of magnetotelluric deep neural network based on space mapping technology, mainly comprising the following steps:

[0171] 1) Determine the detection area, that is, the layered geoelectric section.

[0172] 2) Establish geoelectric model sample set A 2 , the main steps are as follows:

[0173] 2.1) Establish a training sample set G based on the resistivity of each layer of geoelectric section, namely:

[0174]

[0175] In the formula, σ 0 and σ 1 Represent the minimum conductivity and maximum conductivity in the training sample set G; n is the capacity of the set G; when the number of local electric section layers is M, the number of subsets of the training sample set is n M .

[0176] 2.2) Use the conductivity constrained sampling strategy to simplify the training sample set G, and obtain the simplified training sample set A 1 ; Simplify the training sample set A 1 The number of subsets is n×3 M-1 ;

[0177] The conductivity-constraine...

Embodiment 3

[0189] A method for inversion of magnetotelluric deep neural network based on space mapping technology, the main steps are shown in embodiment 2, wherein, the establishment of magnetotelluric forward modeling response data set A 3 The main steps are as follows:

[0190] 1) Using the magnetotelluric sounding method to calculate the orthogonal components of the electric field E and magnetic field H on the earth's surface.

[0191] 2) Based on the electric field E and magnetic field H, establish the magnetotelluric detection data set Z, namely:

[0192]

[0193] In the formula, Z is the impedance tensor used to characterize the electromagnetic field relationship. x and y represent two-dimensional coordinate directions. Among them, Z xx = 0, Z yy = 0, Z xy =-Z yx .

[0194] 3) Calculate the top surface impedance Z of the mth layer m ,which is:

[0195]

[0196] Among them, k m is the wavenumber of the mth layer. h m is the layer thickness of the mth layer. Z om...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses a magnetotelluric deep neural network inversion method based on a space mapping technology. The magnetotelluric deep neural network inversion method mainly comprises the following steps: 1) determining a detection area; 2) establishing a geoelectric model sample set A2; 3) establishing a magnetotelluric forward modeling response data set A3; 4) performing normalization processing; 5) establishing a deep learning neural network model; 6) obtaining a trained deep learning neural network model;7) obtaining a layered geoelectric section electromagnetic prediction data set;8) establishing a layered geoelectric section electromagnetic verification data set; 9) judging whether the fitting degree error of the layered geoelectric section electromagnetic prediction data setand the layered geoelectric section electromagnetic verification data set meets a convergence condition or not, if so, ending inversion, and outputting the layered geoelectric section electromagneticverification data set. The method can be widely applied to the field of magnetotelluric inversion imaging, and has good practical value and application prospect for quickly and accurately predicting the underground electrical structure.

Description

technical field [0001] The invention relates to the field of geophysical magnetotelluric neural network inversion, in particular to a magnetotelluric deep neural network inversion method based on space mapping technology. Background technique [0002] Magnetotelluric (MT) is a branch method of electromagnetic sounding by changing the frequency of the electromagnetic field. Usually, the field source is a vertically incident magnetic field, and the propagation of the underground electromagnetic field satisfies Maxwell's equations. The propagation problem is the modeling problem of magnetotelluric imaging. Magnetotelluric imaging uses the inversion method to obtain the geoelectric model of the underground structure, and infers the underground structure, that is, electromagnetic imaging. Inversion is an extremely critical step in the processing and interpretation of MT data. At present, MT has entered the stage of two-dimensional or even three-dimensional inversion from the ini...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06F17/10G06N3/04G06N3/08
CPCG06F17/10G06N3/08G06N3/045Y02A90/30
Inventor 余年蔡志坤李睿恒葛垚刘洋高磊
Owner CHONGQING UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap