Magnetization vector inversion method and system based on down-sampling structure reconstruction

Abstract

The application is suitable for the field of geophysical exploration technology, and provides a magnetization intensity vector inversion method and system based on down-sampling structure reconstruction, which comprises the following steps: acquiring magnetic anomaly data of a target area, and performing boundary classification to identify the boundary of an anomaly body; calculating a down-sampling ratio of each scale, performing spatial down-sampling reconstruction to obtain a down-sampling reconstruction data set; constructing a target function of the magnetization intensity vector inversion based on the down-sampling structure reconstruction; and iteratively inverting the down-sampling reconstruction data set of each scale by using the target function of the magnetization intensity vector inversion based on the down-sampling structure reconstruction. The application constructs double structure constraints by relying on the spatial down-sampling multi-scale strategy and combining normalized source intensity, optimizes the target function on the basis of a mature MMM inversion framework, and matches the characteristics of different scale geological bodies by using data of different down-sampling scales, so that the geometric shape and spatial position of a magnetic source can be accurately described.

Description

Technical Field

[0001] This invention belongs to the field of geophysical exploration technology, and in particular relates to a method and system for magnetization intensity vector inversion based on downsampling structure reconstruction. Background Technology

[0002] Magnetic exploration is a core geophysical method in mineral resource exploration, underground geological body detection, and structural analysis. Three-dimensional magnetic anomaly inversion can obtain the spatial distribution of magnetization intensity in underground media, thereby inferring the occurrence state of magnetic sources and geological structural characteristics. It is currently widely used in the detection of metallic minerals and concealed rock masses. Remanent magnetization is commonly found within natural rocks and ore bodies, causing a significant deviation between the total magnetization direction and the geomagnetic field direction. Conventional scalar magnetic inversion assumes the magnetization direction is consistent with the geomagnetic field, resulting in substantial errors in the final inversion results. To mitigate the interference from remanent magnetization, the industry has gradually developed magnetization intensity vector inversion technology. This technology can simultaneously recover the three components of magnetization intensity and is adaptable to complex magnetic sources containing remanent magnetization, and has now become the mainstream technology in the field of magnetic inversion.

[0003] Existing magnetization vector inversion methods are mainly divided into two types. The first type sets the magnetization amplitude, dip angle, and deflection angle as unknowns and relies on the initial magnetization direction as a priori constraint. The overall inversion process is complex, and the calculation results require extremely high accuracy of the initial model. The second type is the three-component magnetization vector direct inversion method (MMM inversion method for short), which directly solves the orthogonal three components of the magnetization vector without setting a precise initial direction. The calculation process is simple and more suitable for complex geological environments. It is also the most widely used technical solution at present.

[0004] However, the commonly used MMM-type vector inversion methods still have many shortcomings. In actual exploration work, the magnetic anomalies observed are mostly superimposed responses formed by magnetic sources with different burial depths, spatial scales, and magnetization directions. Single-scale inversion models cannot simultaneously take into account the characteristics of large-scale geological bodies and small local magnets, which easily leads to problems such as blurred model boundaries and ineffective separation of magnetic sources. Conventional local optimization strategies can also cause the inversion model to be overly smoothed, making it difficult to accurately characterize the fine geometry of the magnetic sources. Shallow magnetic sources can also seriously interfere with the identification and reconstruction of deep small magnets. Although data transformation methods such as analytical signal amplitude, magnetic amplitude transformation, and normalized source strength (NSS) can suppress residual magnetism interference, they lose key magnetization direction information. Conventional multi-scale processing methods such as wavelet transform and spectral filtering only focus on frequency decomposition and do not combine the actual spatial scale characteristics of geological bodies. Therefore, developing a new structural reconstruction magnetic intensity vector inversion method has important engineering application value. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for magnetization vector inversion based on downsampling structure reconstruction, aiming to solve the problems mentioned in the background art.

[0006] This invention is implemented as follows: a magnetization vector inversion method based on downsampling structure reconstruction, comprising the following steps:

[0007] Magnetic anomaly data of the target area is acquired, and boundary classification is performed based on the normalized source intensity of the magnetic anomaly data to identify the boundary of the anomaly body.

[0008] Based on the identified anomalous body boundaries, the downsampling ratio at each scale is calculated, and the magnetic anomaly data is spatially downsampled and reconstructed according to the downsampling ratio to generate magnetic anomaly datasets at different scales and their corresponding normalized source intensity datasets, thus obtaining the downsampled reconstructed dataset.

[0009] Based on the three-component magnetization vector direct inversion method, a downsampled multi-scale structure constraint and a multi-scale normalized source intensity constraint are introduced to construct the objective function for magnetization vector inversion of downsampled structure reconstruction.

[0010] Following the order from coarse to fine scales, the objective function of magnetization vector inversion reconstructed by the downsampling structure is used to iteratively invert the downsampling reconstructed datasets at each scale: the inversion result of the current scale is passed to the next finer scale as the initial model, and the optimization is performed step by step until the preset stopping condition is met, so as to obtain the final magnetization vector inversion result.

[0011] Furthermore, the method for identifying the boundaries of anomalous bodies is as follows: the boundary range of different magnetic source bodies is identified by using the normalized total horizontal derivative of the source intensity, thus completing the classification of anomalous bodies.

[0012] Furthermore, the method for calculating the downsampling ratio at each scale is as follows: calculate the characteristic length l of each type of geological body. i Width w i Combined with the total length of the target area l s Total width w s Calculate the downsampling ratio f at the i-th scale. i :

[0013] ;

[0014] In the formula, [·] is the floor function; n min This represents the minimum number of consecutive sampling points for a single anomaly on a sparse grid.

[0015] Furthermore, the method for constructing the objective function for magnetization vector inversion in downsampled structure reconstruction specifically includes: based on the three-component magnetization vector direct inversion method, firstly, introducing downsampled multi-scale structural constraints to construct the objective function of downsampled multi-scale structural constraints; then, based on the objective function of downsampled multi-scale structural constraints, superimposing multi-scale normalized source intensity constraints to obtain the objective function for magnetization vector inversion in downsampled structure reconstruction, and using the conjugate gradient method to complete the minimization solution of the objective function.

[0016] Furthermore, the objective function of the downsampling multi-scale structural constraint The expression is:

[0017] ;

[0018] In the formula, To ultimately observe the magnetic anomaly at the nth scale; The magnetization vector to be inverted at the d-th scale; To finally solve for the magnetization vector; The final combined kernel matrix is ​​obtained; This is the model weight matrix; The constraint coefficients are the total magnetization results at the d-th scale Gram and the j-th scale. For regularization parameters; The Gramian constraint terms are as follows:

[0019] ;

[0020] In the formula, The x, y, and z axes represent the magnetization component in the η direction at the d-th scale, with the x, y, and z axes pointing north, east, and vertically upward, respectively. This represents the total magnetization at the j-th scale.

[0021] Furthermore, the objective function for inverting the magnetization vector reconstructed by the downsampling structure is... The expression is:

[0022] ;

[0023] In the formula, For multi-scale normalized source intensity constraint coefficients; The Gram constraint coefficient; The Gramian constraint terms are as follows:

[0024] ;

[0025] In the formula, The equivalent magnetization is derived from the normalized source intensity anomaly at the d-th scale. This represents the magnetization component in the η direction at the d-th scale.

[0026] Another object of the present invention is to provide a magnetization vector inversion system based on downsampled structure reconstruction, for implementing the above-mentioned magnetization vector inversion method based on downsampled structure reconstruction, comprising:

[0027] The data acquisition and classification unit is used to acquire magnetic anomaly data of the target area and perform boundary classification based on the normalized source intensity of the magnetic anomaly data to identify the boundary of the anomaly body.

[0028] The dataset construction unit is used to calculate the downsampling ratio at each scale based on the identified anomalous body boundary, and to perform spatial downsampling reconstruction on the magnetic anomaly data according to the downsampling ratio, generating magnetic anomaly datasets at different scales and their corresponding normalized source intensity datasets, thus obtaining the downsampling reconstruction dataset.

[0029] The objective function construction unit is used to construct the objective function for magnetization vector inversion based on the three-component magnetization vector direct inversion method, by introducing downsampled multi-scale structural constraints and multi-scale normalized source intensity constraints;

[0030] The iterative inversion unit is used to iteratively invert the downsampled reconstructed dataset at each scale in order from coarse to fine scale, using the objective function of the magnetization intensity vector inversion reconstructed by the downsampled structure: the inversion result of the current scale is passed to the next finer scale as the initial model, and the optimization is performed step by step until the preset stopping condition is met, so as to obtain the final magnetization intensity vector inversion result.

[0031] This invention provides a magnetization vector inversion method based on downsampling structure reconstruction. It leverages a multi-scale spatial downsampling strategy combined with normalized source intensity to construct a dual structural constraint. Building upon the mature MMM inversion framework, it optimizes the objective function and utilizes data from different downsampling scales to match the characteristics of geological bodies of different sizes. This enhances the decoupling capability and spatial resolution of superimposed magnetic anomalies, thereby accurately characterizing the geometry and spatial location of magnetic sources. Furthermore, this invention combines the advantages of normalized source intensity—its insensitivity to magnetization direction and its strong ability to identify lateral boundaries—to further improve the accuracy and stability of the inversion, contributing to the prediction of concealed targets such as hidden rock masses in mining areas. Attached Figure Description

[0032] Figure 1 This is a schematic flowchart of the magnetization vector inversion method based on downsampling structure reconstruction provided in an embodiment of the present invention.

[0033] Figure 2The following are examples of magnetic anomalies in bicubic synthetic models with different magnetization directions obtained by the method provided in this embodiment of the invention: (a) a three-dimensional view of the model location; (b) the magnetic anomaly of the model; (c) the inversion result of the MMM method; (d) the inversion result of the method of this invention; (e) the three-dimensional truncated volume of the MMM method (threshold = 0.07); and (f) the three-dimensional truncated volume of the method of this invention (threshold = 0.154). Detailed Implementation

[0034] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0035] like Figure 1 As shown, in one embodiment of the present invention, a magnetization vector inversion method based on downsampling structure reconstruction is provided, which solves the problems of low resolution, insufficient inversion accuracy, and large magnetization direction recovery deviation in magnetic anomaly magnetization vector inversion technology under superimposed magnetic anomaly scenarios; specifically, the method includes the following steps:

[0036] S1. Data Acquisition and Normalized Source Intensity (NSS) Boundary Classification: Acquire magnetic anomaly data of the target area and perform boundary classification based on the normalized source intensity of the magnetic anomaly data to identify the boundary of the anomaly body;

[0037] S2. Calculate the downsampling ratio and construct the downsampling reconstruction dataset: Based on the identified anomalous body boundary, calculate the downsampling ratio at each scale, and perform spatial downsampling reconstruction on the magnetic anomaly data according to the downsampling ratio to generate magnetic anomaly datasets at different scales and their corresponding normalized source intensity datasets, thus obtaining the downsampling reconstruction dataset.

[0038] S3. Constructing the objective function for high-precision magnetization vector inversion of downsampled structure reconstruction: Based on the three-component magnetization vector direct inversion method (MMM), we introduce downsampled multi-scale structure constraints and multi-scale normalized source intensity constraints to construct the objective function for magnetization vector inversion of downsampled structure reconstruction.

[0039] S4. Iterative inversion with coarse-to-fine downsampling and NSS structure constraints: Following the order from coarse to fine scale, the objective function of the magnetization intensity vector inversion reconstructed by the downsampling structure is used to iteratively invert the downsampling reconstruction dataset at each scale in turn: The inversion result of the current scale is passed to the next finer scale as the initial model, and the optimization is carried out step by step until the preset stopping condition is met, and the final magnetization intensity vector inversion result is obtained.

[0040] The theoretical basis of the method provided in this embodiment of the invention is as follows:

[0041] This invention's embodiments are based on the MMM inversion method of directly solving the three-component magnetization vector, and introduce Gram constraint to enhance the three-component magnetization (MMM inversion method). x M y M z The structural correlation of the MMM method reduces the non-uniqueness of the inversion. The original objective function of the MMM method is... The expression is as follows:

[0042] ;

[0043] in, To observe magnetic anomalies; This is the combined kernel matrix corresponding to the magnetization components in the three directions (x, y, z); This is the model weight matrix, used to compensate for the decay of the kernel function with depth; The three-component magnetization vector to be recovered, where the x, y, and z axes point north, east, and vertically upward, respectively; regularization parameter and Gram constraint coefficient Determined through an adaptive method.

[0044] Gramian constraints for:

[0045] ;

[0046] in, is the total magnetization; is the magnetization component corresponding to the η direction.

[0047] Furthermore, the normalized source intensity is extremely insensitive to the magnetization direction. It possesses excellent lateral boundary recognition capabilities and can effectively suppress residual magnetic interference. The original magnetic anomaly data is transformed to calculate global normalized source intensity anomaly data, which serves as an auxiliary structural constraint term. Simultaneously, the normalized source intensity, combined with the horizontal derivative, completes anomaly boundary classification, providing a basis for subsequent downsampling structure reconstruction.

[0048] In a preferred embodiment of the present invention, the method for identifying the boundary of anomalies is as follows: calculate the normalized source intensity of the magnetic anomaly data to obtain global normalized source intensity anomaly data; use the total horizontal derivative of the normalized source intensity to identify the boundary range of different magnetic sources and complete the classification of anomalies.

[0049] In a preferred embodiment of the present invention, the method for calculating the downsampling ratio at each scale is as follows: based on the identified anomalous body boundaries, the characteristic length l of each type of geological body is calculated. i Width w i Combined with the total length of the target area l s Total width w s Calculate the downsampling ratio f at the i-th scale.i :

[0050] ;

[0051] In the formula, [·] is the floor function; n min This represents the minimum number of consecutive sampling points (3-4) required for a single anomaly on a sparse grid.

[0052] Generate magnetic anomaly datasets {ΔT} at different downsampling scales according to the downsampling ratio. d The dataset is a normalized source intensity dataset with corresponding scales (d is the scale number, d=1,2...n, scales arranged from coarse to fine). This embodiment of the invention uses downsampling reconstruction instead of conventional spectral decomposition to preserve the spatial physical properties of geological bodies and avoid aberrant distortions caused by frequency domain decomposition.

[0053] In a preferred embodiment of the present invention, the method for constructing the objective function for magnetization vector inversion of downsampled structure reconstruction specifically includes: based on the inversion framework of the three-component magnetization vector direct inversion method (MMM), firstly, introducing downsampled multi-scale structural constraints to construct the objective function of downsampled multi-scale structural constraints; then, based on the objective function of downsampled multi-scale structural constraints, superimposing multi-scale normalized source intensity constraints to obtain the objective function for magnetization vector inversion of downsampled structure reconstruction, and using the conjugate gradient method to complete the minimization solution of the objective function.

[0054] In a preferred embodiment of the present invention, the objective function of the downsampled multi-scale structural constraint described above is... The expression is:

[0055] ;

[0056] In the formula, To ultimately observe the magnetic anomaly at the nth scale; The magnetization vector to be inverted at the d-th scale; To finally solve for the magnetization vector; The final combined kernel matrix is ​​obtained; This is the model weight matrix; The constraint coefficients for the total magnetization results at the d-th scale Gram and the j-th scale are adaptively selected. For regularization parameters; The Gramian constraint terms are as follows:

[0057] ;

[0058] In the formula, The x, y, and z axes represent the magnetization component in the η direction at the d-th scale, with the x, y, and z axes pointing north, east, and vertically upward, respectively. This represents the total magnetization at the j-th scale.

[0059] The objective function for magnetization vector inversion reconstructed from the above downsampled structure The expression is:

[0060] ;

[0061] In the formula, For multi-scale normalized source intensity constraint coefficients; The Gram constraint coefficient; The Gramian constraint terms are as follows:

[0062] ;

[0063] In the formula, The equivalent magnetization is derived from the normalized source intensity anomaly at the d-th scale. This represents the magnetization component in the η direction at the d-th scale.

[0064] The objective function for magnetization vector inversion based on this downsampling structure The high-precision magnetization vector of the downsampled structure reconstruction is iteratively inverted in order of scale from coarse to fine. Coarse-scale data corresponds to large-scale geological bodies, prioritizing the constraint of the overall regional magnetic structure. Fine-scale data corresponds to small local magnets, gradually supplementing local fine features. After the inversion work at each scale is completed, the inversion result of the current scale is used as the initial model for the next level of fine-scale inversion, and the calculation results are iteratively optimized step by step. In this embodiment of the invention, a dual iteration stopping criterion can be set, namely a preset maximum number of iterations and a data root mean square deviation threshold. The conventional maximum number of iterations is set to 25. The iteration is terminated when either criterion is met during the calculation.

[0065] Example 1: The method provided in this embodiment of the invention (i.e., the method of this invention) was used to conduct testing using a dual magnetic source synthesis model with different magnetization directions. Two cubic magnetic sources were constructed. Cube 1, with dimensions of 8×6×4 km, had a magnetization tilt angle of -40° and a deflection angle of 25°. Cube 2, with dimensions of 3×1.8×3 km, had a magnetization tilt angle of 50° and a deflection angle of 5°. The two magnetic sources had different burial depths and spatial scales to simulate a superimposed magnetic anomaly environment with significant differences in magnetization directions in the field. The geomagnetic field tilt angle was set to 60° and the deflection angle to 10°. The magnetization intensity was uniformly set to 0.5 A / m, the observation grid was set to 72×72, and the grid spacing was 5 km. Applying the method of this invention, anomaly classification is completed by normalizing the source intensity horizontal derivative, generating a downsampled sequence {9,3,1} and constructing a three-level downsampled structure to reconstruct the dataset. The inversion grid is set to 36×36×20, with a horizontal spacing of 1 km and a vertical spacing of 0.5 km. The maximum number of iterations is set to 25. Simultaneously, the conventional MMM method is used to conduct synchronous comparative inversion. The comparison results are shown in Table 1 and... Figure 2 As shown.

[0066] Table 1

[0067]

[0068] Test results show that the conventional MMM method calculates a tilt angle error of 10.6° and an eccentricity error of 12.2° for cube 1, and an orientation error of over 10.0° for cube 2. The global model similarity is only 84.72%, and there is a severe fusion phenomenon at the boundaries of the two magnetic sources. The method of this invention calculates a tilt angle error of 2.4° and an eccentricity error of 0.3° for cube 1, and a tilt angle error of 0.9° and an eccentricity error of 1.2° for cube 2. The global model similarity reaches 98.10%, which is 19.38% higher than the conventional MMM method. The shape and position of the two magnetic sources are highly consistent with the real model, and there are no obvious artifacts.

[0069] In summary, the method provided by this invention utilizes data from different downsampling scales to match the characteristics of geological bodies of different scales in practical applications, thereby improving the decoupling capability and spatial resolution of superimposed magnetic anomalies and accurately characterizing the geometry and spatial location of magnetic sources. Furthermore, by combining the advantages of normalized source intensity being insensitive to magnetization direction and possessing outstanding lateral boundary identification capabilities, the accuracy and stability of the inversion are further improved. This invention addresses the shortcomings of existing methods. Simulation experiments verify that, when facing complex magnetic sources with strong remanence and large differences in magnetization direction, the conventional MMM method can achieve magnetization tilt and deflection angle errors exceeding 10°. The method provided by this invention can control the direction error within 3°. The global model similarity of this invention is improved by nearly 20% compared to the conventional MMM method, accurately separating multiple superimposed magnetic sources with clear model boundaries and significantly reduced artifacts.

[0070] In another embodiment of the present invention, a magnetization vector inversion system based on downsampling structure reconstruction is also provided, for implementing the above-described magnetization vector inversion method based on downsampling structure reconstruction, comprising:

[0071] The data acquisition and classification unit is used to acquire magnetic anomaly data of the target area and perform boundary classification based on the normalized source intensity of the magnetic anomaly data to identify the boundary of the anomaly body.

[0072] The dataset construction unit is used to calculate the downsampling ratio at each scale based on the identified anomalous body boundary, and to perform spatial downsampling reconstruction on the magnetic anomaly data according to the downsampling ratio, generating magnetic anomaly datasets at different scales and their corresponding normalized source intensity datasets, thus obtaining the downsampling reconstruction dataset.

[0073] The objective function construction unit is used to construct the objective function for magnetization vector inversion based on the three-component magnetization vector direct inversion method, by introducing downsampled multi-scale structural constraints and multi-scale normalized source intensity constraints;

[0074] The iterative inversion unit is used to iteratively invert the downsampled reconstructed dataset at each scale in order from coarse to fine scale, using the objective function of the magnetization intensity vector inversion reconstructed by the downsampled structure: the inversion result of the current scale is passed to the next finer scale as the initial model, and the optimization is performed step by step until the preset stopping condition is met, so as to obtain the final magnetization intensity vector inversion result.

[0075] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods.

[0076] The above embodiments merely illustrate several implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this patent should be determined by the appended claims.

Claims

1. A magnetization vector inversion method based on downsampling structure reconstruction, characterized in that, Includes the following steps: Magnetic anomaly data of the target area is acquired, and boundary classification is performed based on the normalized source intensity of the magnetic anomaly data to identify the boundary of the anomaly body. Based on the identified anomalous body boundaries, the downsampling ratio at each scale is calculated, and the magnetic anomaly data is spatially downsampled and reconstructed according to the downsampling ratio to generate magnetic anomaly datasets at different scales and their corresponding normalized source intensity datasets, thus obtaining the downsampled reconstructed dataset. Based on the three-component magnetization vector direct inversion method, a downsampled multi-scale structure constraint and a multi-scale normalized source intensity constraint are introduced to construct the objective function for magnetization vector inversion of downsampled structure reconstruction. Following the order from coarse to fine scales, the objective function of magnetization vector inversion reconstructed by the downsampling structure is used to iteratively invert the downsampling reconstructed datasets at each scale: the inversion result of the current scale is passed to the next finer scale as the initial model, and the optimization is performed step by step until the preset stopping condition is met, so as to obtain the final magnetization vector inversion result.

2. The magnetization vector inversion method based on downsampling structure reconstruction according to claim 1, characterized in that, The method for identifying the boundaries of anomalous bodies is as follows: the boundary range of different magnetic sources is identified by using the normalized total horizontal derivative of the source intensity, thus completing the classification of anomalous bodies.

3. The magnetization vector inversion method based on downsampling structure reconstruction according to claim 1, characterized in that, The method for calculating the downsampling ratio at each scale is as follows: calculate the characteristic length l of each type of geological body. i Width w i Combined with the total length of the target area l s Total width w s Calculate the downsampling ratio f at the i-th scale. i : ； In the formula, [·] is the floor function; n min This represents the minimum number of consecutive sampling points for a single anomaly on a sparse grid.

4. The magnetization vector inversion method based on downsampling structure reconstruction according to claim 1, characterized in that, The method for constructing the objective function for magnetization vector inversion in downsampled structure reconstruction specifically includes: based on the three-component magnetization vector direct inversion method, firstly, introducing downsampled multi-scale structural constraints to construct the objective function of downsampled multi-scale structural constraints; then, based on the objective function of downsampled multi-scale structural constraints, superimposing multi-scale normalized source intensity constraints to obtain the objective function for magnetization vector inversion in downsampled structure reconstruction, and using the conjugate gradient method to complete the minimization solution of the objective function.

5. The magnetization vector inversion method based on downsampling structure reconstruction according to claim 4, characterized in that, The objective function of the downsampled multi-scale structural constraint The expression is: ； In the formula, To ultimately observe the magnetic anomaly at the nth scale; The magnetization vector to be inverted at the d-th scale; To finally solve for the magnetization vector; The final combined kernel matrix is ​​obtained; This is the model weight matrix; The constraint coefficients are the total magnetization results at the d-th scale Gram and the j-th scale. For regularization parameters; The Gramian constraint terms are as follows: ； In the formula, The x, y, and z axes represent the magnetization component in the η direction at the d-th scale, with the x, y, and z axes pointing north, east, and vertically upward, respectively. This represents the total magnetization at the j-th scale.

6. The magnetization vector inversion method based on downsampling structure reconstruction according to claim 5, characterized in that, The objective function for magnetization vector inversion reconstructed by the downsampling structure is... The expression is: ； In the formula, For multi-scale normalized source intensity constraint coefficients; The Gram constraint coefficient; The Gramian constraint terms are as follows: ； In the formula, The equivalent magnetization is derived from the normalized source intensity anomaly at the d-th scale. This represents the magnetization component in the η direction at the d-th scale.

7. A magnetization vector inversion system based on downsampling structure reconstruction, used to implement the magnetization vector inversion method based on downsampling structure reconstruction as described in any one of claims 1-6, characterized in that, include: The data acquisition and classification unit is used to acquire magnetic anomaly data of the target area and perform boundary classification based on the normalized source intensity of the magnetic anomaly data to identify the boundary of the anomaly body. The dataset construction unit is used to calculate the downsampling ratio at each scale based on the identified anomalous body boundary, and to perform spatial downsampling reconstruction on the magnetic anomaly data according to the downsampling ratio, generating magnetic anomaly datasets at different scales and their corresponding normalized source intensity datasets, thus obtaining the downsampling reconstruction dataset. The objective function construction unit is used to construct the objective function for magnetization vector inversion based on the three-component magnetization vector direct inversion method, by introducing downsampled multi-scale structural constraints and multi-scale normalized source intensity constraints; The iterative inversion unit is used to iteratively invert the downsampled reconstructed dataset at each scale in order from coarse to fine scale, using the objective function of the magnetization intensity vector inversion reconstructed by the downsampled structure: the inversion result of the current scale is passed to the next finer scale as the initial model, and the optimization is performed step by step until the preset stopping condition is met, so as to obtain the final magnetization intensity vector inversion result.

Images