Electromagnetic method initial model building method based on sub-domain and reverse gradient absorption

Abstract

The application discloses an electromagnetic method initial model establishment method based on sub-domain and reverse gradient absorption, and comprises the following steps: acquiring electromagnetic detection data of a detection area; dividing the detection area into a high-frequency sub-domain, a medium-frequency sub-domain and a low-frequency sub-domain along a horizontal direction and a vertical direction of detection according to corresponding frequencies of the electromagnetic detection data; performing dynamic sub-domain division on the high-frequency sub-domain, the medium-frequency sub-domain and the low-frequency sub-domain respectively, and performing dynamic sub-domain control to obtain a plurality of dynamic sub-domains; establishing an inversion target function of the dynamic sub-domains, and sequentially inverting the plurality of dynamic sub-domains to obtain a sub-domain inversion model; presetting a depth constraint condition; establishing a regional inversion function, and performing local boundary gradient mutation processing; and combining the depth constraint condition to perform depth inversion on the sub-domain inversion model to obtain a global full-depth inversion initial model. Through the above scheme, the application has the advantages of simple specific logic, high accuracy and reliability and the like.

Description

Technical Field

[0001] This invention relates to the field of geophysical electromagnetic exploration technology, and in particular to a method for establishing an initial electromagnetic model based on domain division and reverse gradient absorption. Background Technology

[0002] Electromagnetic methods face a series of technical challenges in current engineering geophysical exploration, mineral resource exploration, and hydrogeological investigation. First, the propagation characteristics of signals differ significantly between different depths and geological layers, leading to discrepancies in detection results between shallow and deep layers. Signal attenuation is rapid in shallow areas, limiting the accuracy of deep detection; while the complex geological structures and anomalous signal interference in deep areas often make it difficult for conventional detection methods to accurately distinguish responses at different levels. How to effectively improve detection depth and accuracy in multiple depths and geological environments has become a major challenge. Second, current electromagnetic inversion algorithms struggle to effectively separate response signals from different strata under complex deep geological conditions, significantly limiting the accuracy and reliability of inversion results. Therefore, there is an urgent need to develop new technological systems to overcome the bottleneck of inversion accuracy under complex geological conditions, thereby improving the overall effectiveness of deep engineering geophysical exploration, mineral resource exploration, and hydrogeological investigation, and enhancing the synergistic ability of detection depth and accuracy. These technological breakthroughs will provide more efficient solutions for the precise detection of deep mineral resources, groundwater, and environmental issues.

[0003] To address the problem of weak and difficult-to-detect deep anomaly signals, traditional geophysical electromagnetic methods generally suffer from drawbacks such as high construction difficulty, low accuracy, and weak effective information. For example, the technology disclosed in "Publication No. CN118604896A, entitled A Geophysical Instrument Based on Natural Electromagnetic Field and Its Mapping Method" determines the frequency value corresponding to each specific geological depth in the geological profile map to be generated based on the conversion relationship between the geological depth and frequency of the area to be measured, and marks it as the target frequency value; then it acquires the electromagnetic field data detected by each acquisition module at different target frequency values, thereby reflecting the depth of different specific geological conditions based on different electromagnetic field data. This patent may have the following shortcomings: the detection method based on natural electromagnetic fields is easily affected by the geological environment and electromagnetic interference, reducing the detection accuracy; the adaptability and stability of the geophysical instrument under complex geological structures are limited; the mapping method does not consider the complexity of the medium and the sensitivity differences under actual conditions, so the reliability of its final results is poor.

[0004] Furthermore, the technology disclosed in "Publication No. CN119337638A, entitled 'Regional Decomposition Magnetotelluric Three-Dimensional Forward Modeling Simulation Method and System'" decomposes a large-scale MT problem into multiple smaller subdomains by introducing overlapping DDM, using a direct solver to solve each small linear system within the subdomain, and reusing the LU decomposition results during the Schwarz iteration process. This patent may have the following shortcomings: the regional decomposition method may not fully consider the mutual influence of different regions in complex geological environments, resulting in limited simulation accuracy, and it is not integrated into the inversion system; moreover, this technology cannot improve the actual three-dimensional inversion imaging accuracy, its regional decomposition mode does not consider dividing the spatial regions of sensitivity at different frequencies, and its decomposition technique is only for improving computational efficiency, failing to contribute to the final required improvement in three-dimensional structural resolution.

[0005] In practical applications, due to the complex and variable underground environment, the changes in natural electromagnetic fields and geoelectric fields may be affected by geological structures and groundwater distribution, leading to instability in measurement results. Especially under complex underground conditions, traditional electromagnetic detection methods are difficult to meet the needs of accurate detection.

[0006] Therefore, there is an urgent need to propose a logically simple, accurate and reliable method for establishing an initial model of the electromagnetic method based on domain division and reverse gradient absorption. Summary of the Invention

[0007] To address the aforementioned problems, the present invention aims to provide a method for establishing an initial model for electromagnetic methods based on domain division and reverse gradient absorption. The technical solution adopted by the present invention is as follows: The method for establishing the initial model of the electromagnetic method based on domain division and reverse gradient absorption includes the following steps: Electromagnetic detection data of the detection area is acquired, and the detection area is divided into high-frequency domain, mid-frequency domain and low-frequency domain along the horizontal and vertical directions of the detection based on the frequency corresponding to the electromagnetic detection data. The high-frequency domain, mid-frequency domain, and low-frequency domain are divided into dynamic subdomains, and dynamic subdomain control is performed to obtain several dynamic subdomains. Establish the inversion objective function for the dynamic subdomain, and invert several dynamic subdomains in sequence to obtain the domain-specific inversion model; Preset depth constraints; Establish a regional inversion function and perform gradient mutation processing on the local boundary; combine the depth constraint condition to perform a fixed-depth inversion on the regional inversion model to obtain the initial model of the global full-depth inversion.

[0008] Furthermore, dynamic subdomains are dynamically divided into high-frequency, mid-frequency, and low-frequency domains, and dynamic subdomain control is performed; the dynamic subdomain control matrix is ​​used for the dynamic subdomain control. The expression is: in, This represents the block matrix corresponding to the electromagnetic detection data of the nth dynamic subdomain; This represents the block matrix corresponding to the electromagnetic detection data of the i-th dynamic subdomain; The number of matrix units corresponding to the electromagnetic detection data of the detection area; n is a natural number greater than 1; i is a natural number less than or equal to n; The expression is: in, This represents the number of matrix units for all electromagnetic detection data in the i-th dynamic subdomain.

[0009] Furthermore, the expression for the inversion objective function of the dynamic subdomain is: in, Represents a data weighting matrix; Represents the observation data matrix; Represents a forward-modeled data matrix; Represents the model covariance matrix; denoted by ; T denotes the prior model; represents the regularization factor; m represents the model vector.

[0010] Furthermore, the depth inversion function The expression is:

[0011] in, This represents the transpose of the sensitivity matrix. This represents the transpose of the data weighting matrix; This represents the conjugate residual between the observed data and the forward modeling data; Represents the prior model; denoted by , where m represents the regularization factor and m represents the model vector.

[0012] Furthermore, the expression for the local boundary gradient abrupt change processing is: in, Indicates the j-th frequency. The number of grids corresponding to the grid layer from the local boundary to the ground; This represents the gradient matrix parameters at the j-th frequency; This indicates the number of grid layers.

[0013] Compared with the prior art, the present invention has the following beneficial effects: (1) This invention improves the traditional solid-state domain division by using a dynamic domain division method to make the domain division more reasonable. In addition, by improving the gradient calculation method, the dynamic domain division and the depth division can be jointly inverted. Furthermore, this invention uses the final result of the joint inversion as the initial model to perform full-domain and full-depth inversion.

[0014] (2) This invention, by combining a reverse gradient absorption strategy for domain-specific processing within each depth range, can clearly extract electromagnetic anomaly signals at different depths and enhance the spatial resolution of the signals, thereby achieving a more accurate and reliable inversion effect. It effectively solves the problem of mutual interference between various signals. This invention not only improves the independence of signals and avoids the complexity of multiple factors superimposed, but also significantly enhances the detection capability of complex underground media, especially showing obvious advantages when dealing with signal interference in complex environments.

[0015] (3) This invention divides the detection area into high-frequency, mid-frequency, and low-frequency domains along the transverse and longitudinal directions of the detection based on the frequency corresponding to the electromagnetic detection data. It then performs dynamic sub-domain division and dynamic sub-domain control on each of the high-frequency, mid-frequency, and low-frequency domains to obtain several dynamic sub-domains. Compared to traditional domain division methods, dynamic sub-domains can more rationally utilize the spatial sensitivity characteristics of electromagnetic waves for domain imaging.

[0016] In summary, this invention, based on a domain-specific and reverse gradient absorption model, possesses advantages such as logical simplicity, accuracy, and reliability. It provides strong technical support for the precise assessment of resource potential and the optimization of development plans, thus advancing deep resource exploration technology. This invention has high practical and promotional value in the field of geophysical electromagnetic exploration technology. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope of protection. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a logic flowchart of the present invention.

[0019] Figure 2 This is a schematic diagram of the high-frequency domain division method in this invention.

[0020] Figure 3 This is a schematic diagram of the intermediate frequency domain division method in this invention.

[0021] Figure 4 This is a schematic diagram of the low-frequency domain division method in this invention.

[0022] Figure 5 This is a schematic diagram of the high-frequency domain matrix in this invention.

[0023] Figure 6 This is a schematic diagram of the intermediate frequency domain matrix in this invention.

[0024] Figure 7 This is a schematic diagram of the low-frequency domain matrix in this invention.

[0025] Figure 8 This is a schematic diagram of the depth division in this invention.

[0026] Figure 9 This is a schematic diagram of the frequency-1 localization matrix in this invention.

[0027] Figure 10 This is a schematic diagram of the frequency-2 localized matrix in this invention.

[0028] Figure 11 This is a schematic diagram of the frequency n-local matrix in this invention.

[0029] Figure 12 This is a schematic diagram of the observation system in this invention.

[0030] Figure 13 This is a diagram of the three-dimensional geoelectric model in this invention.

[0031] Figure 14 This is a diagram showing the full-frequency inversion results in this invention.

[0032] Figure 15 This is a diagram showing the frequency division inversion results in this invention. Detailed Implementation

[0033] To make the objectives, technical solutions, and advantages of this application clearer, the present invention will be further described below with reference to the accompanying drawings and embodiments. The embodiments of the present invention include, but are not limited to, the following embodiments. All other embodiments obtained by those skilled in the art based on the embodiments in this application without inventive effort are within the scope of protection of this application.

[0034] In this embodiment, the term "and / or" is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone.

[0035] The terms "first" and "second," etc., used in the specification and claims of this embodiment are used to distinguish different objects, not to describe a specific order of objects. For example, "first target object" and "second target object," etc., are used to distinguish different target objects, not to describe a specific order of target objects.

[0036] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0037] In the description of the embodiments in this application, unless otherwise stated, "multiple" means two or more. For example, multiple processing units means two or more processing units; multiple systems means two or more systems.

[0038] like Figures 1 to 15 As shown, this embodiment provides a method for establishing an initial electromagnetic model based on domain-specific and reverse gradient absorption. It performs domain-specific and depth-specific inversions on engineering geophysical exploration, mineral resource, and hydrogeological data, significantly reducing the mutual influence between multiple regions and levels. This method uses a reverse gradient absorption strategy to process signals from different sources and levels independently, avoiding the complexity of multiple factors superimposed, thereby improving data processing accuracy and inversion results. Independent inversion at each level not only extracts anomalous signals more clearly but also effectively improves spatial resolution, providing more accurate and reliable inversion results. This method demonstrates excellent detection capabilities when dealing with signal interference in complex environments, providing strong technical support for resource potential assessment and development scheme optimization.

[0039] To achieve efficient inversion of engineering geophysical exploration, mineral resources, and hydrogeological data, this embodiment first performs domain-based processing on different regions, independently inverting their respective data to avoid mutual interference between multiple domains. Next, depth-based inversion is performed within different depth regions to ensure that data at each level can be extracted separately, reducing the impact of depth differences. This step-by-step separation and independent inversion method not only extracts clearer anomaly signals but also improves the spatial and temporal resolution of the data, providing more accurate and reliable support for resource exploration and hydrogeological investigation.

[0040] Specifically, the method for establishing the initial model of the electromagnetic method based on domain division and reverse gradient absorption includes the following steps: The first step is to acquire electromagnetic detection data of the detection area, and then divide the detection area into high-frequency, mid-frequency, and low-frequency domains along the horizontal and vertical directions of the detection based on the frequency corresponding to the electromagnetic detection data.

[0041] The second step involves dynamically dividing the high-frequency, mid-frequency, and low-frequency domains into subdomains and then controlling these subdomains to obtain several dynamic subdomains. For example... Figures 2 to 4As shown, the domain is divided according to the different detection areas in the horizontal and vertical directions at different frequencies; the lower the frequency, the larger the range. The gradient matrix diagram is shown below. Figures 5 to 7 As shown.

[0042] The third step is to establish the inversion objective function for the dynamic subdomains, and then sequentially invert several dynamic subdomains to obtain the domain-specific inversion model. Here, the expression for the inversion objective function for the dynamic subdomains is: in, Represents a data weighting matrix; Represents the observation data matrix; Represents a forward-modeled data matrix; Represents the model covariance matrix; denoted by ; T denotes the prior model; represents the regularization factor; m represents the model vector.

[0043] Among them, the dynamic subdomain control matrix of dynamic subdomain control The expression is: in, This represents the block matrix corresponding to the electromagnetic detection data of the nth dynamic subdomain; This represents the block matrix corresponding to the electromagnetic detection data of the i-th dynamic subdomain; The number of matrix units corresponding to the electromagnetic detection data of the detection area; n is a natural number greater than 1; i is a natural number less than or equal to n.

[0044] The The expression is: in, This represents the number of matrix units for all electromagnetic detection data in the i-th dynamic subdomain.

[0045] The fourth step is to perform depth-based inversion based on the previous domain segmentation. A diagram illustrating the depth-based inversion is shown below. Figure 8 As shown, frequency 1 is inverted first. Then, 1 / n of the frequency domain inversion depth is controlled by a fixed depth to prevent it from participating in the inversion of frequency 2, continuing until all frequencies have been inverted. Here, during the depth determination, we divide different effective depths according to different underground resistivity. Figures 9 to 11 In the diagram, ①, ②, ③, ④, and ⑤ represent the five subdomains in the front, back, left, right, and top directions of the target layer, respectively.

[0046] The fifth step involves establishing a depth inversion function and performing gradient abrupt changes at the local boundary. Inversion is then performed sequentially on several effective depths to obtain the initial model for full-domain, full-depth inversion. Here, the depth inversion function... The expression is: in, This represents the transpose of the sensitivity matrix. This represents the transpose of the data weighting matrix; This represents the conjugate residual between the observed data and the forward modeling data; Represents the prior model; denoted by , where m represents the regularization factor and m represents the model vector.

[0047] In this embodiment, to better address the problem of abrupt gradient changes at the local boundary, the expression for handling abrupt gradient changes at the local boundary is as follows: like Figure 4 As shown, this embodiment divides the space where the gradient needs to be controlled into 5 regions. In formula (5), we use... We can then perform gradient control on it. It is a parameter less than 1 that we assign to it. It represents the number of grid layers from the fixed depth boundary to the ground, indicated by the subscript. It represents Figure 4 Which region in the map? The closer to the surface, the better. The smaller the value, the smaller the gradient value, and the smaller the impact on the inversion.

[0048] The following is a brief example: The first step is to establish a three-dimensional geoelectric model: To verify the feasibility of this embodiment, multiple anomaly models were established in a uniform half-space. Figure 13 Preliminary verification was conducted; the background resistivity was set to 500 Ωm, and the resistivity of the black anomaly in the figure was 10 Ωm.

[0049] The second step is to establish the observation system. The observation system in this embodiment varies depending on the different data collected and the different measurement methods used. For example, in magnetotelluric methods, the system mainly measures data such as Ex, Ey, Hx, Hy, and Hz, which represent the components of electric and magnetic fields in different directions. To measure these data, the system deploys electrodes and magnetic rods in different directions on the ground. Specifically, electrodes and magnetic rods parallel to the x-axis are used to measure the Ex and Hx components, those parallel to the y-axis are used to measure the Ey and Hy components, and those parallel to the z-axis are specifically used to measure the Hz component. This arrangement of electrodes and magnetic rods in different directions effectively acquires data on different electric and magnetic field components, thus helping researchers to comprehensively understand the underground electrical characteristics. Different measurement methods and arrangements not only affect the types of data collected but also determine the application effectiveness of the observation system in different scenarios. Therefore, in practical use, selecting appropriate measurement methods and data acquisition arrangements based on specific needs and objectives is crucial.

[0050] The third step is data preprocessing: For the multiple datasets obtained from forward modeling, a domain- and depth-based processing method is employed. This effectively separates various signals independently, avoiding mutual interference between different domains and depths. This separation process not only reduces interference caused by signal overlap but also allows each type of data to exhibit more accurate characteristics during independent analysis, thereby improving the accuracy and reliability of data inversion. This method provides a clearer and more systematic processing solution for data analysis in complex environments, effectively enhancing the quality of the final inversion results.

[0051] The fourth step is to establish an initial model: Since inversion problems often involve significant ambiguity, multiple different models may produce the same observational data. Therefore, before performing forward and inversion calculations, a reasonable and reliable initial model must be established to provide a suitable starting point for the inversion process, thereby reducing solution uncertainty and improving computational accuracy. In this case, the initial model is set as a uniform half-space with a resistivity of 500 Ω·m. This simple assumption helps avoid overfitting in complex geological environments and provides a basic reference framework for subsequent inversion. By setting these initial models, the inversion process can be effectively guided, unnecessary computational biases can be avoided, and the final inversion results can be ensured to be closer to the actual geological conditions.

[0052] like Figure 13 As shown in the example, the model provides a uniform half-space three-dimensional geoelectric model with a background resistivity of 500 Ωm. The black anomaly in the figure represents a low-resistivity anomaly with a resistivity of 10 Ωm. Impedance data Zxx, Zxy, Zyx, and Zyy are obtained through forward modeling, with 2% random noise added as inversion data. The global and full-depth joint inversion results of this example are shown below. Figure 14 As shown in the figure, it has some effect on low-resistivity recovery, but it is not very significant. Furthermore, this embodiment uses an electromagnetic method based on domain division and reverse gradient absorption to establish the initial model, and the final inversion result is as follows. Figure 15 As shown, a comparison with the results of the joint inversion of the whole domain and the whole depth shows that the inversion of the domain and the depth significantly improves the inversion effect, which proves the correctness of the frequency division inversion technique in the uniform half space in this embodiment.

[0053] The above embodiments are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any changes made based on the design principles of the present invention, or any non-creative modifications made thereon, shall fall within the scope of protection of the present invention.

Claims

1. A method for establishing an initial model for electromagnetic methods based on domain division and reverse gradient absorption, characterized in that, Includes the following steps: Electromagnetic detection data of the detection area is acquired, and the detection area is divided into high-frequency domain, mid-frequency domain and low-frequency domain along the horizontal and vertical directions of the detection based on the frequency corresponding to the electromagnetic detection data. The high-frequency domain, mid-frequency domain, and low-frequency domain are divided into dynamic subdomains, and dynamic subdomain control is performed to obtain several dynamic subdomains. The dynamic subdomain is divided into domains based on the different detection areas in the horizontal and vertical directions at different frequencies; the dynamic subdomain control matrix controls the dynamic subdomain. The expression is: ; in, This represents the block matrix corresponding to the electromagnetic detection data of the nth dynamic subdomain; This represents the block matrix corresponding to the electromagnetic detection data of the i-th dynamic subdomain; This represents the number of matrix units corresponding to the electromagnetic detection data of the detection area; where n is a natural number greater than 1; and i is a natural number less than or equal to n. The The expression is: ; in, This represents the number of matrix units for all electromagnetic detection data in the i-th dynamic subdomain; An inversion objective function for the dynamic subdomain is established, and the inversion is performed sequentially on several dynamic subdomains to obtain a domain-specific inversion model; the expression for the inversion objective function of the dynamic subdomain is: ； in, Represents a data weighting matrix; Represents the observation data matrix; Represents a forward-modeled data matrix; Represents the model covariance matrix; denoted by ; T denotes the prior model; The regularization factor is represented by m; the model vector is represented by m. Preset depth constraints; Establish a regional inversion function and perform gradient mutation processing on the local boundary; combine the depth constraint condition to perform a fixed-depth inversion on the regional inversion model to obtain the initial model of the global full-depth inversion.

2. The method for establishing an initial model for the electromagnetic method based on domain division and reverse gradient absorption according to claim 1, characterized in that, Also includes: The gradient operator of the inversion objective function of the dynamic subdomain is obtained, and the deep inversion function is obtained. Its expression is: ； in, The transpose of the sensitivity matrix; This represents the conjugate residual between the observed data and the forward modeling data; Represents the prior model; denoted by , where m represents the regularization factor and m represents the model vector.

3. The method for establishing an initial model for the electromagnetic method based on domain division and reverse gradient absorption according to claim 2, characterized in that, The expression for the gradient mutation treatment at the local boundary is: ； in, Indicates the j-th frequency. The number of grids corresponding to the grid layer from the local boundary to the ground; This represents the gradient matrix parameters at the j-th frequency; This indicates the number of vertical grid layers that require gradient matrix control.

Images