Curie high-precision inversion method based on aeromagnetic data

Abstract

The application provides a high-precision inversion method for the crustal interior based on aeromagnetic data, which comprises the following steps: preprocessing the aeromagnetic data of a target area, eliminating noise and interference, and extracting regional magnetic anomaly information related to the deep crustal interior; constructing a residual magnetization strength model based on a depth attenuation function, and obtaining the residual magnetization strength of each grid point in the target area according to the target area and the residual magnetization strength model; establishing a nonlinear decay model of the residual magnetization strength with depth based on the residual magnetization strength model, and obtaining the magnetic anomaly strength generated on the ground by each depth layer in the target area by using the nonlinear decay model; and inputting the regional magnetic anomaly information, residual magnetization probability parameters, geological constraints and magnetic anomaly strength into a crustal interior inversion model to obtain the depth prediction value of the target area. The application considers the influence of residual magnetization on the magnetization direction, and improves the inversion precision of the crustal interior.

Description

Technical Field

[0001] This invention relates to the field of geophysical technology, and in particular to a high-precision Curie inversion method based on aeromagnetic data. Background Technology

[0002] The Curie point is the temperature interface in the Earth's crust where ferromagnetic minerals transform into paramagnetic substances when the temperature rises to the Curie point. As the bottom interface of the magnetic layer in magnetic exploration, the Curie point effectively characterizes the distribution of underground temperature fields, possessing significant scientific and practical value for geothermal energy development, oil and gas resource prediction, earthquake and volcanic hazard prevention, and primary hydrothermal mineral exploration. By using aeromagnetic data, regional magnetic anomaly information caused by the magnetic basement can be extracted, and the depth distribution of the Curie point can be accurately calculated using inversion methods, thus providing crucial evidence for related geological research and resource exploration.

[0003] Existing Curie inversion methods are typically based on the assumption that the study area has no remanent magnetization, meaning that parameters such as effective magnetization dip and deflection are directly set to the current values ​​of the geomagnetic field during the inversion process. However, in areas with complex geological histories and significant remanent magnetization effects, this assumption ignores the significant contribution of remanent magnetization to the magnetization direction, leading to large deviations in the Curie inversion results. This reduces the accuracy and reliability of the inversion, making it difficult to meet the needs of high-precision resource exploration and geological research.

[0004] To address the above problems, this invention provides a high-precision Curie inversion method based on aeromagnetic data, applicable to cases where residual magnetism exists. Summary of the Invention

[0005] This invention provides a high-precision Curie inversion method based on aeromagnetic data. Its main purpose is to consider the influence of residual magnetism on Curie inversion and effectively improve the inversion accuracy of Curie.

[0006] In a first aspect, embodiments of the present invention provide a Curie high-precision inversion method based on aeromagnetic data, comprising:

[0007] Preprocess the aeromagnetic data of the target area to eliminate noise and interference, and extract regional magnetic anomaly information related to the deep Curie region;

[0008] A remanence model based on a depth decay function is constructed, and the remanence intensity of each grid point in the target region is obtained according to the target region and the remanence model.

[0009] Based on the remanent magnetization model, a nonlinear decay model of remanent magnetization with depth is established. Using the nonlinear decay model, the magnetic anomaly intensity generated at the surface of each depth layer in the target area is obtained.

[0010] The regional magnetic anomaly information, the remanent magnetic probability parameter, geological constraints, and the magnetic anomaly intensity are input into the Curie inversion model to obtain the predicted Curie depth value of the target area. The remanent magnetic probability parameter is the mean and variance of the total distribution field, and the total distribution field is composed of the remanent magnetic intensity of each grid point in the target area.

[0011] The Curie inversion model uses a residual convolution module to extract the spatial pattern of magnetic anomalies, and then uses a parallel fully connected branch to fuse the remanent magnetization probability parameters and geological constraints. The geological constraints include the spatial distribution of different rock types, the distribution of fault zones (referring to fault locations and fracture zone ranges).

[0012] Furthermore, the Curie inversion model is trained through the following steps:

[0013] Based on the remanence intensity model, magnetic anomaly intensity simulation data are generated by randomly sampling remanence intensity and direction combinations.

[0014] The simulated magnetic anomaly intensity data and the collected real magnetic anomaly intensity data are added to the training dataset, and the Curie inversion model is trained using the training dataset.

[0015] The Curie depth prediction value output by the Curie inversion model is converted into a theoretical magnetic anomaly through a differentiable physical layer. A joint loss function is constructed with the measured Curie depth value to force the magnetic anomaly-depth physical relationship, thus obtaining the trained Curie inversion model.

[0016] Furthermore, the formula for calculating the joint loss function is as follows:

[0017] L = L data +λL laplace ;

[0018]

[0019]

[0020] Where L represents the joint loss function, λ represents the weight parameter, and N is the number of measurement points. This represents the predicted Curie depth value at the i-th measurement point. Let represent the theoretical Curie depth of the i-th measurement point, and M represent the total number of magnetopotential points involved in the joint loss function calculation. U represents the directional derivative, i.e., the gradient. j Let i represent the magnetic potential at point j, where i and j are both positive integers.

[0021] Furthermore, the preprocessing of the aeromagnetic data of the target area to eliminate noise and interference, and to extract regional magnetic anomaly information related to the deep Curie region, includes the following steps:

[0022] External magnetic field interference in the aeromagnetic data is reduced by using geomagnetic diurnal variation correction and altitude correction.

[0023] Wavelet transform and principal component analysis are used to eliminate high-frequency noise in the aeromagnetic data, while retaining the low-frequency magnetic anomaly component dominated by the deep Curie region, thus obtaining the regional magnetic anomaly information.

[0024] Furthermore, the calculation formula for the remanence intensity model is as follows:

[0025] R(z i )=R0·f(z i );

[0026]

[0027] Among them, R(z) i ) represents the depth value z i Remanence at z i Let represent the depth value of the i-th group, R0 represent the remanent magnetization at z=0 on the surface, α be the attenuation coefficient, β be the nonlinear exponent, and i be a positive integer. This formula represents the subsurface depth z. i Remanence at point R(z) i It can be obtained by multiplying the remanent magnetization R0 at the surface with the depth decay function.

[0028] Furthermore, the calculation formula for the nonlinear attenuation model is as follows:

[0029]

[0030] Wherein, ΔB(z) i ) represents the depth value z i The ratio of the intensity of the magnetic anomaly at the Earth's surface to the remanence, z i R(z) represents the depth value of the i-th group, k is the scaling factor, and R(z) represents the depth value of the i-th group. i ) represents the depth value z i The remanence at depth z is given by the formula, where i is a positive integer. i The ratio ΔB(z) of the surface magnetic anomaly intensity to the remanent magnetic intensity at the corresponding location. i ), determined by the proportionality coefficient k and the remanence R(z) at that depth. i ) and depth z i The squares of z together determine ΔB(z). i ) and remanence R(z) i It is directly proportional to the depth z. i The ratio is inversely proportional to the square of the value, representing the quantitative influence of depth and remanent magnetization on the surface magnetic anomaly intensity and the ratio of remanent magnetization.

[0031] Further, the step of inputting the regional magnetic anomaly information, the remanent magnetization probability parameter, geological constraints, and the magnetic anomaly intensity into the Curie inversion model to obtain the predicted Curie depth value of the target region includes:

[0032] The regional magnetic anomaly information and the remanence probability parameter are used as additional feature vectors;

[0033] The geological constraints are converted into a two-dimensional network matrix;

[0034] The additional feature vector, the two-dimensional network matrix, and the magnetic anomaly intensity are input into the Curie inversion model to obtain the predicted Curie depth value.

[0035] Secondly, embodiments of the present invention provide a Curie high-precision inversion system based on aeromagnetic data, comprising:

[0036] The preprocessing module is used to preprocess the aeromagnetic data of the target area, eliminate noise and interference, and extract regional magnetic anomaly information related to the deep Curie region.

[0037] The remanence module is used to construct a remanence intensity model based on a depth decay function, and to obtain the remanence intensity of each grid point in the target region according to the target region and the remanence intensity model.

[0038] The magnetic anomaly module is used to establish a nonlinear decay model of remanent magnetic intensity with depth based on the remanent magnetic intensity model, and to obtain the magnetic anomaly intensity generated at the surface of each depth layer in the target area using the nonlinear decay model.

[0039] The inversion module is used to input the regional magnetic anomaly information, the remanent magnetic probability parameter, geological constraints and the magnetic anomaly intensity into the Curie inversion model to obtain the predicted Curie depth value of the target area. The remanent magnetic probability parameter is the mean and variance of the total distribution field, and the total distribution field is composed of the remanent magnetic intensity of each grid point in the target area.

[0040] The Curie inversion model uses a residual convolution module to extract the spatial pattern of magnetic anomalies, and then uses a parallel fully connected branch to fuse the remanent magnetization probability parameters and geological constraints. The geological constraints include the spatial distribution of different rock types, the distribution of fault zones (referring to fault locations and fracture zone ranges).

[0041] Thirdly, embodiments of the present invention provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described Curie high-precision inversion method based on aeromagnetic data.

[0042] Fourthly, embodiments of the present invention provide a computer storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described high-precision Curie inversion method based on aeromagnetic data.

[0043] This invention proposes a high-precision Curie inversion method based on aeromagnetic data. In the presence of remanent magnetization, preprocessing addresses the signal mixing problem caused by the superposition of remanent and induced magnetization. Furthermore, it introduces spatially correlated random remanent magnetization perturbations and establishes a nonlinear decay model of remanent magnetization intensity with depth based on a depth decay function. This simulates the coupling effect of remanent magnetization at different depths on surface magnetic anomaly signals, solving the problem of inaccurate estimation of effective magnetization dip / deflection angle caused by neglecting the remanent magnetization component in traditional methods. In addition, the above steps can be used to construct a synthetic magnetic anomaly dataset containing the mixed effects of remanent and induced magnetization, effectively enhancing the generalization ability of the Curie inversion model to complex remanent magnetization interference patterns. Moreover, through a spatial-depth dual-constraint mechanism, remanent magnetization distortions that are difficult to separate in traditional methods are transformed into learnable physical features, providing the Curie inversion model with magnetic field inputs closer to real geological conditions and improving the accuracy of Curie inversion. Attached Figure Description

[0044] Figure 1 A flowchart of a high-precision Curie inversion method based on aeromagnetic data provided in this embodiment of the invention;

[0045] Figure 2 A schematic diagram of the structure of a Curie high-precision inversion system based on aeromagnetic data provided in an embodiment of the present invention;

[0046] Figure 3 This is a schematic diagram of the structure of a computer device provided in an embodiment of the present invention.

[0047] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0048] The embodiments of this application are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application.

[0049] To enable those skilled in the art to better understand the solutions of this application, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0050] In the embodiments of this application, "at least one" refers to one or more; "multiple" refers to two or more. In the description of this application, terms such as "first," "second," and "third" are used only for descriptive purposes and should not be construed as indicating or implying relative importance or order. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0051] References such as “one embodiment” or “some embodiments” as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the terms “comprising,” “including,” “having,” and variations thereof, as used in this specification, mean “including, but not limited to,” unless otherwise specifically emphasized.

[0052] Figure 1 A flowchart of a high-precision Curie inversion method based on aeromagnetic data provided in this embodiment of the invention is shown below. Figure 1 As shown, the method includes:

[0053] S10, preprocess the aeromagnetic data of the target area to eliminate noise and interference, and extract regional magnetic anomaly information related to the deep Curie region;

[0054] Step S10 includes:

[0055] S11, using geomagnetic diurnal variation correction and altitude correction to reduce external magnetic field interference in the aeromagnetic data;

[0056] S12, wavelet transform and principal component analysis are used to eliminate high-frequency noise in the aeromagnetic data, while retaining the low-frequency magnetic anomaly component dominated by the deep Curie region, and finally obtaining the regional magnetic anomaly information.

[0057] In the presence of remanent magnetization, the preprocessing steps for aeromagnetic data need to focus on addressing the signal contamination caused by the superposition of remanent and induced magnetization. First, geomagnetic diurnal variation correction and altitude correction are used to reduce external magnetic field interference in the aeromagnetic data. Then, wavelet transform and principal component analysis are employed to eliminate high-frequency noise such as aircraft interference and surface magnetic noise, while retaining the low-frequency magnetic anomaly component dominated by the deep Curie region. This ensures that the obtained regional magnetic anomaly information reflects both the deep Curie morphology and the contribution characteristics of the remanent magnetization effect, providing a high-quality dataset for the subsequent training of the Curie inversion model.

[0058] S20, construct a remanence intensity model based on the depth decay function, and obtain the remanence intensity of each grid point in the target region according to the target region and the remanence intensity model;

[0059] Traditional methods assume that the magnetization direction is determined solely by the current geomagnetic field, and neglecting the remanent magnetization component leads to errors in the estimation of the effective magnetization dip / deflection angle. In this invention, spatially correlated random remanent magnetization disturbances are introduced into the magnetosphere, and a nonlinear decay model of remanent magnetization intensity with depth is established based on a depth decay function, thereby simulating the coupling effect of remanent magnetization at different depths on surface magnetic anomaly signals.

[0060] The specific inputs for constructing a spatially correlated stochastic remanence model for the study area are as follows:

[0061] The study area is divided into spatial grids, where the grid coordinates are (x... i ,y j The grid consists of M×N points with point spacing of Δx and Δy; spatial correlation parameters are set, including a horizontal correlation length of L. x and L y The spatial correlation range can be controlled by setting the horizontal correlation length; statistical parameters are also set, where the mean disturbance is μ and the variance is σ. 2 .

[0062] The output of the remanent magnetization model is as follows: the remanent magnetization perturbation value ΔM(x) at each grid point. i ,y j ), which satisfies the random distribution of spatial correlation.

[0063] The calculation steps for the remanence model are as follows:

[0064] 1. Generate white noise, generating a white noise field W(x) that follows a normal distribution. i ,y j )~N(μ,σ 2 There is no spatial correlation.

[0065] 2. Introduce spatial correlation:

[0066] First calculate the covariance between any two points.

[0067] in σ represents the covariance between two points. 2 d represents the variance, reflecting the overall degree of fluctuation. x =|x1-x2|、d y =|y1-y2| is the distance between two points in the x and y directions, L x L y The characteristic length determines the degree to which distance affects covariance. This formula shows that the covariance of two points decreases exponentially with increasing distance between them in the x and y directions; that is, points that are spatially closer are more correlated, reflecting the influence of spatial location on the correlation of magnetic properties—points that are close to each other have more similar magnetic properties. For each grid point (x... i ,y j Its remanent magnetic perturbation value can be obtained by weighted averaging of neighboring points:

[0068]

[0069] Finally, the calculation formula for the remanence intensity model is as follows:

[0070] R(z i )=R0·f(z i );

[0071]

[0072] Among them, R(z) i ) represents the depth value z i Remanence at z i Let represent the depth value of the i-th group, R0 represent the remanent magnetization at z=0 on the surface, α be the attenuation coefficient, β be the nonlinear exponent, and i be a positive integer. This formula represents the subsurface depth z. i Remanence at point R(z) i It can be obtained by multiplying the remanent magnetization R0 at the surface with the depth decay function.

[0073] S30. Based on the remanent magnetization model, a nonlinear decay model of remanent magnetization with depth is established. Using the nonlinear decay model, the magnetic anomaly intensity generated at the surface of each depth layer in the target area is obtained.

[0074] In this embodiment of the invention, the specific input for establishing a nonlinear decay model of remanent magnetization with depth based on the depth decay function is as follows:

[0075] Depth layer: a set of depth values ​​z1, z2, ..., z N Where z = 0 represents the Earth's surface, and downward is positive;

[0076] Remanent magnetization at the Earth's surface: R0, the remanent magnetization at the Earth's surface at z=0;

[0077] Attenuation function parameters: nonlinear attenuation function Where α is the attenuation coefficient and β is the nonlinear exponent.

[0078] The intensity of the magnetic anomaly generated at the Earth's surface at each depth layer is directly proportional to the remanence and inversely proportional to the square of the depth.

[0079]

[0080] Wherein, ΔB(z) i ) represents the depth value z i The ratio of the intensity of the magnetic anomaly at the Earth's surface to the remanence, z i R(z) represents the depth value of the i-th group, k is the scaling factor, and R(z) represents the depth value of the i-th group. i ) represents the depth value z i The remanence at depth z. This formula shows that the remanence at depth z... i The ratio ΔB(z) of the surface magnetic anomaly intensity to the remanent magnetic intensity at the corresponding location. i ), determined by the proportionality coefficient k and the remanence R(z) at that depth. i ) and depth z i The squares of z together determine ΔB(z). i ) and remanence R(z) i It is directly proportional to the depth z. i The ratio is inversely proportional to the square of the value, reflecting the quantitative influence of depth and remanent magnetization on the surface magnetic anomaly intensity and the ratio of remanent magnetization.

[0081] The above steps can be used to construct a synthetic dataset of magnetic anomalies containing a hybrid effect of remanent magnetization and inductive magnetization, effectively enhancing the generalization ability of neural networks to complex remanent magnetization disturbance patterns. This design, through a spatial-depth dual-constraint mechanism, transforms remanent magnetization distortions, which are difficult to separate in traditional methods, into learnable physical features, providing neural networks with magnetic field inputs that more closely resemble real geological conditions.

[0082] S40, input the regional magnetic anomaly information, the remanent magnetic probability parameter, geological constraints and the magnetic anomaly intensity into the Curie inversion model to obtain the predicted value of the Curie depth of the target area, wherein the remanent magnetic probability parameter is the mean and variance of the total distribution field, and the total distribution field is composed of the remanent magnetic intensity of each grid point in the target area;

[0083] The Curie inversion model uses a residual convolution module to extract the spatial pattern of magnetic anomalies, and then uses a parallel fully connected branch to fuse the remanent magnetization probability parameters and geological constraints. The geological constraints include the spatial distribution of different rock types, the distribution of fault zones (referring to fault locations and fracture zone ranges).

[0084] Step S40 includes:

[0085] S41, the regional magnetic anomaly information and the remanence probability parameter are used as additional feature vectors;

[0086] To address the challenge of nonlinear mapping caused by the coupling of remanence and magnetic induction, a hybrid neural network architecture embedding physical information is designed as the Curie inversion model. At the input end, in addition to regional magnetic anomaly information, a remanence probability parameter is introduced as an additional feature vector. The remanence distribution field mainly consists of the remanence intensity R(z) at each grid point generated in the above steps. i The total distribution field is composed of the remanent magnetization. The remanent magnetization probability parameter is a global statistic of the distribution field (mean, variance), which is used as the input of the average remanent magnetization intensity and deviation in the post-Curie inversion model.

[0087] S42, convert the geological constraints into a two-dimensional network matrix;

[0088] The Curie inversion model uses a residual convolution module to extract the spatial pattern of magnetic anomalies, and a parallel fully connected branch to fuse the remanent magnetic probability parameters (mean and variance of remanent magnetic properties) and geological constraints. Among them, the geological constraints refer to the spatial distribution of different rock types (such as granite and basalt), and the fault zone distribution refers to the location of faults and the range of fracture zones, which reflect the influence of geological structure on remanent magnetic properties.

[0089] Before inputting the geological constraints into the Curie inversion model, the above geological constraints are converted into a two-dimensional network matrix as a spatial mask. For example, in the fault constraint, fault zones are represented by 1 and no fault zones by 2; in the rock type constraint, granite is represented by 1 and basalt by 2, etc. The converted mask matrix is ​​then used as a separate channel and input into the Curie inversion model along with the regional magnetic anomaly information. For example, the regional magnetic anomaly information is channel one and the mask matrix is ​​channel two.

[0090] S43, input the additional feature vector, the two-dimensional network matrix and the magnetic anomaly intensity into the Curie inversion model to obtain the predicted value of the Curie depth.

[0091] By inputting the aforementioned additional feature vectors, the two-dimensional network orange, and the magnetic anomaly intensity into the Curie inversion model, the predicted Curie depth value can be obtained. A differentiable physical layer (using the Laplace equation, etc.) is used to convert the predicted depth into a theoretical magnetic anomaly. A joint loss function is then constructed with the measured data, forcing the network to satisfy the magnetic anomaly-depth physical relationship, thereby improving the accuracy of the predicted Curie depth value.

[0092] The Curie inversion model in this embodiment is a neural network model, which needs to be trained before application. Specifically, the Curie inversion model is trained through the following steps:

[0093] Based on the remanence intensity model, magnetic anomaly intensity simulation data are generated by randomly sampling remanence intensity and direction combinations.

[0094] The simulated magnetic anomaly intensity data and the collected real magnetic anomaly intensity data are added to the training dataset, and the Curie inversion model is trained using the training dataset.

[0095] The Curie depth prediction value output by the Curie inversion model is converted into a theoretical magnetic anomaly through a differentiable physical layer. A joint loss function is constructed with the measured Curie depth value to force the magnetic anomaly-depth physical relationship, thus obtaining the trained Curie inversion model.

[0096] In this embodiment of the invention, a training dataset is constructed using simulated magnetic anomaly intensity data generated by a remanence intensity model and real magnetic anomaly intensity data collected by measurement. This solves the generalization problem caused by the spatial heterogeneity of remanence parameters. Furthermore, a Monte Carlo Dropout strategy is introduced during training. During inference, the standard deviation of the Curie depth prediction value is calculated through multiple random forward propagations, quantifying the impact of remanence parameter uncertainty on the inversion results. Simultaneously, an adaptive weighted loss function is employed to apply higher weights to the measured data region, balancing the theoretical completeness of the synthetic data with the local realism of the measured data, thereby improving the training accuracy of the Curie inversion model.

[0097] (1) First, calculate the ordinary loss function:

[0098]

[0099] Among them, L data The average loss value (MSE) is used. This represents the predicted Curie depth value at the i-th measurement point. This represents the theoretical value of the Curie depth at the i-th measurement point, where N is the number of measurement points and i is a positive integer.

[0100] (2) Then, the loss is constrained using physical information:

[0101] Forced predictions to satisfy the Laplace equation Now calculate the sum of squared Laplace equation residuals for all grid points within the region:

[0102]

[0103] Where Llaplace represents the physical information constraint loss, and M represents the total number of magnetic potential points participating in the joint loss function calculation. U represents the directional derivative, i.e., the gradient. j Let j represent the magnetic potential at point j, where j is a positive integer.

[0104] (3) Combining (1) and (2) above, we obtain the joint loss function.

[0105] By combining the two in a weighted manner, the network is driven to simultaneously fit the data and satisfy the physical law L=L data +λL laplace λ is a weighting parameter used to adjust the strength of physical constraints. This design deeply couples the remanence model with data-driven learning, overcoming the limitation of traditional pure data-driven models lacking physical interpretability.

[0106] This invention proposes a high-precision Curie inversion method based on aeromagnetic data. In the presence of remanent magnetization, preprocessing addresses the signal mixing problem caused by the superposition of remanent and induced magnetization. Furthermore, it introduces spatially correlated random remanent magnetization perturbations and establishes a nonlinear decay model of remanent magnetization intensity with depth based on a depth decay function. This simulates the coupling effect of remanent magnetization at different depths on surface magnetic anomaly signals, solving the problem of inaccurate estimation of effective magnetization dip / deflection angle caused by neglecting the remanent magnetization component in traditional methods. In addition, the above steps can be used to construct a synthetic magnetic anomaly dataset containing the mixed effects of remanent and induced magnetization, effectively enhancing the generalization ability of the Curie inversion model to complex remanent magnetization interference patterns. Moreover, through a spatial-depth dual-constraint mechanism, remanent magnetization distortions that are difficult to separate in traditional methods are transformed into learnable physical features, providing the Curie inversion model with magnetic field inputs closer to real geological conditions and improving the accuracy of Curie inversion.

[0107] Figure 2 A schematic diagram of a Curie high-precision inversion system based on aeromagnetic data provided in an embodiment of the present invention is shown below. Figure 2 As shown, the system includes a preprocessing module 210, a remanence module 220, a magnetic anomaly module 230, and an inversion module 240, wherein:

[0108] Preprocessing module 210 is used to preprocess the aeromagnetic data of the target area, eliminate noise and interference, and extract regional magnetic anomaly information related to the deep Curie region.

[0109] The remanence module 220 is used to construct a remanence intensity model based on the depth decay function, and to obtain the remanence intensity of each grid point in the target region according to the target region and the remanence intensity model;

[0110] The magnetic anomaly module 230 is used to establish a nonlinear decay model of remanent magnetic intensity with depth based on the remanent magnetic intensity model, and to obtain the magnetic anomaly intensity generated at the surface of each depth layer in the target area using the nonlinear decay model.

[0111] The inversion module 240 is used to input the regional magnetic anomaly information, the remanent magnetic probability parameter, geological constraints and the magnetic anomaly intensity into the Curie inversion model to obtain the predicted value of the Curie depth of the target area. The remanent magnetic probability parameter is the mean and variance of the total distribution field, and the total distribution field is composed of the remanent magnetic intensity of each grid point in the target area.

[0112] The Curie inversion model uses a residual convolution module to extract the spatial pattern of magnetic anomalies, and then uses a parallel fully connected branch to fuse the remanent magnetization probability parameters and geological constraints. The geological constraints include the spatial distribution of different rock types, the distribution of fault zones (referring to fault locations and fracture zone ranges).

[0113] This embodiment is a system embodiment corresponding to the above method embodiment. Its specific implementation process is the same as that of the above method embodiment. For details, please refer to the above method embodiment. This system embodiment will not repeat the details.

[0114] The modules in the Curie high-precision inversion system based on aeromagnetic data described above can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the computer device's memory as software, so that the processor can call and execute the corresponding operations of each module.

[0115] Figure 3 This is a schematic diagram of the structure of a computer device provided in an embodiment of the present invention. The computer device may be a server, and its internal structure diagram may be as follows: Figure 3 As shown, the computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computational and control capabilities. The memory includes a computer storage medium and internal memory. The computer storage medium stores the operating system, computer programs, and the database. The internal memory provides the environment for the operation of the operating system and computer programs in the computer storage medium. The database stores data generated or acquired during the execution of a Curie high-precision inversion method based on aeromagnetic data, such as aeromagnetic data and remanence models. The network interface is used for communication with external terminals via a network connection. The computer program is executed by the processor to implement a Curie high-precision inversion method based on aeromagnetic data.

[0116] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of a Curie high-precision inversion method based on aeromagnetic data as described in the above embodiment. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in this embodiment of a Curie high-precision inversion system based on aeromagnetic data.

[0117] In one embodiment, a computer storage medium is provided, on which a computer program is stored. When executed by a processor, the computer program implements the steps of the Curie high-precision inversion method based on aeromagnetic data described in the above embodiment. Alternatively, when executed by a processor, the computer program implements the functions of each module / unit in the above embodiment of the Curie high-precision inversion system based on aeromagnetic data.

[0118] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer 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. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0119] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0120] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for high-precision inversion of the earth's interior based on aeromagnetic data, characterized in that, The method comprises the following steps: Preprocessing the aeromagnetic data of the target area to eliminate noise and interference and extract regional magnetic anomaly information related to deep crustal intrusions; Constructing a remanent magnetization intensity model based on a depth decay function, and obtaining the remanent magnetization intensity of each grid point in the target area according to the target area and the remanent magnetization intensity model; Based on the remanent magnetization intensity model, a nonlinear decay model of remanent magnetization intensity with depth is established, and the magnetic anomaly intensity generated by each depth layer in the target area on the ground is obtained by using the nonlinear decay model; The regional magnetic anomaly information, the remanent magnetization probability parameter, the geological constraint and the magnetic anomaly intensity are input into a crustal intrusion inversion model to obtain the crustal intrusion depth prediction value of the target area, wherein the remanent magnetization probability parameter is the mean and variance of the total distribution field composed of the remanent magnetization intensity of each grid point in the target area; The crustal intrusion inversion model uses a residual convolution module to extract the spatial pattern of magnetic anomalies, and a parallel full connection branch to fuse the remanent magnetization probability parameter and the geological constraint, wherein the geological constraint includes the spatial distribution of different rock types, the distribution of fault zones, the location of faults and the range of fracture zones.

2. The method according to claim 1, wherein, The crustal intrusion inversion model is trained by the following steps: Based on the remanent magnetization intensity model, randomly sample remanent magnetization intensity, direction combination to generate magnetic anomaly intensity simulation data; The magnetic anomaly intensity simulation data and the collected magnetic isomerization intensity real data are added to the training data set, and the crustal intrusion inversion model is trained by using the training data set; The crustal intrusion depth prediction value output by the crustal intrusion inversion model is converted into a theoretical magnetic anomaly by a differentiable physical layer, and a joint loss function is constructed with the crustal intrusion depth measured value to force the satisfaction of the magnetic anomaly-depth physical relationship, thereby obtaining the trained crustal intrusion inversion model.

3. The method according to claim 2, wherein, The calculation formula of the joint loss function is as follows: L = L data + λL laplace ; wherein L represents the joint loss function, λ represents a weight parameter, N represents the number of measurement points, represents the predicted value of the internal potential depth of the i-th measurement point, represents the theoretical value of the internal potential depth of the i-th measurement point, and M represents the total number of magnetic potential points participating in the calculation of the joint loss function, represents the directional derivative, i.e., the gradient, U j represents the magnetic potential of the j-th point, and i and j are both positive integers.

4. The method according to claim 1, wherein, The preprocessing of the aeromagnetic data of the target area to eliminate noise and interference and extract regional magnetic anomaly information related to deep crustal intrusions comprises the following steps: Using geomagnetic diurnal variation correction and height correction to weaken the external magnetic field interference of the aeromagnetic data; Using wavelet transform and principal component analysis to eliminate high-frequency noise of the aeromagnetic data while retaining low-frequency magnetic anomaly components dominated by deep crustal intrusions, and finally obtaining the regional magnetic anomaly information.

5. The method according to claim 1, wherein, The calculation formula of the remanent magnetization intensity model is as follows: R(z i ) = R0·f(z i ) Among them, R(z) i ) represents the depth value z i Remanence at z i Let R0 represent the depth value of the i-th group, R0 represent the remanent magnetization at z=0 on the Earth's surface, α be the attenuation coefficient, β be the nonlinear exponent, and i be a positive integer.

6. The method according to claim 1, wherein, The calculation formula of the nonlinear decay model is as follows: where ΔB(z i ) represents the ratio of the magnetic anomaly intensity generated at the depth value z i to the residual intensity, z i represents the i-th group of depth values, k is a proportional coefficient, R(z i ) represents the residual intensity at the depth value z i , and i is a positive integer. where ΔB(z i ) represents the ratio of the magnetic anomaly intensity generated at the depth value z i to the residual intensity, z i represents the i-th group of depth values, k is a proportional coefficient, R(z i ) represents the residual intensity at the depth value z i , and i is a positive integer.

7. The method according to claim 1, wherein, The steps of inputting the regional magnetic anomaly information, the remanent magnetization probability parameter, the geological constraint and the magnetic anomaly intensity into the crustal intrusion inversion model to obtain the crustal intrusion depth prediction value of the target area comprise the following steps: The regional magnetic anomaly information and the remanent magnetization probability parameter are used as additional feature vectors; The geological constraint is converted into a two-dimensional network matrix; The additional feature vectors, the two-dimensional network matrix and the magnetic anomaly intensity are input into the crustal intrusion inversion model to obtain the crustal intrusion depth prediction value.

8. A high-precision inversion system for the earth's interior based on aeromagnetic data, characterized in that, The method comprises the following steps: A preprocessing module is configured to preprocess the aeromagnetic data of the target area to eliminate noise and interference and extract regional magnetic anomaly information related to deep crustal intrusions; A residual magnetism module is configured to construct a residual magnetism intensity model based on a deep decay function, and obtain residual magnetism intensities of grid points in the target region according to the target region and the residual magnetism intensity model; A magnetic anomaly module is configured to construct a nonlinear decay model of residual magnetism intensity with depth based on the residual magnetism intensity model, and obtain magnetic anomaly intensities generated by each depth layer in the target region on the ground surface by using the nonlinear decay model; An inversion module is configured to input the regional magnetic anomaly information, the residual magnetism probability parameter, the geological constraint, and the magnetic anomaly intensity into an intracrust inversion model to obtain a predicted value of the target region in the intracrust, wherein the residual magnetism probability parameter is a mean value and a variance of a total distribution field composed of residual magnetism intensities of grid points in the target region. The intracrust inversion model uses a residual convolution module to extract a spatial pattern of magnetic anomalies, and a parallel full connection branch to fuse the residual magnetism probability parameter and the geological constraint, wherein the geological constraint includes spatial distribution of different rock types, distribution of fault zones, fault positions, and ranges of fracture zones.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the steps of the high-precision intracrust inversion method based on aeromagnetic data according to any one of claims 1 to 7.

10. A computer storage medium storing a computer program, the computer program comprising instructions, which, when executed by a computer, cause the computer to perform the method according to any one of claims 1 to 9. The computer program is executed by the processor to implement the steps of the high-precision intracrust inversion method based on aeromagnetic data according to any one of claims 1 to 7.

Images