Self-feedback type regularization inversion method

Abstract

This invention discloses a self-feedback regularized inversion method, relating to the field of electromagnetic wave imaging technology. Its key technical points include the following steps: S1, acquiring imaging data; S2, preprocessing the imaging data; S3, establishing an inversion model, setting regularization parameters and threshold parameters, and performing initial inversion to obtain preliminary inversion results; S4, calculating error feedback coefficients based on the preliminary inversion results; S5, updating the inversion model using the error feedback coefficients; S6, iteratively inverting the updated inversion model until convergence conditions are met; S7, performing post-processing operations to obtain the final imaging result. This invention combines traditional regularized inversion methods with self-feedback control technology, effectively improving imaging quality while ensuring the stability of the inversion results. Furthermore, the method proposed in this invention has significant application value and practical significance in electromagnetic wave imaging technology.

Description

Technical Field

[0001] This invention relates to the field of electromagnetic wave imaging technology, and specifically to a self-feedback regularized inversion method. Background Technology

[0002] Electromagnetic wave imaging technology is a crucial imaging technique with wide applications in military, medical, geological, and engineering fields. It identifies and locates objects by analyzing the reflection, refraction, and scattering properties of electromagnetic waves. In practical applications, electromagnetic wave imaging technology is affected by factors such as noise and signal attenuation; therefore, effectively improving image quality is a key research focus in this field.

[0003] Currently, many inversion methods have been applied to electromagnetic wave imaging technology, such as least squares method and total variational regularization. Among them, regularized inversion method is an important method, which can effectively overcome noise and attenuation problems in the inversion process. A search revealed Chinese patent CN113466954B, which discloses a self-feedback regularized inversion method. The technical problem it addresses is that existing inversion methods are easily affected by data noise and prone to getting trapped in local minima, leading to overfitting. The disclosed method for solving these problems includes the following steps:

[0004] Step S1: Polarize and plot the planar magnetic anomaly data; Step S2: Delineate the magnetic anomaly plane based on the polarized map and interpolate the delineated data; Step S3: Find the plane location of the maximum value and use this plane location as the initial position for inversion; Step S4: Start the inversion iteration process, first assign values ​​to the initial model, then iteratively update the model parameters, and calculate the fitting difference; after iteration, obtain the output inversion result; Step S5: Obtain the inversion model parameters and the degree of fit; if the degree of fit meets the requirements, output the result; if the degree of fit does not meet the requirements, return to step S2, re-delineate the data, until the output degree of fit meets the requirements.

[0005] However, the methods in the aforementioned patents and existing regularized inversion methods often suffer from problems such as over-smoothing and distortion, which affect image quality.

[0006] Therefore, the present invention aims to provide a self-feedback regularized inversion method to solve the above problems. Summary of the Invention

[0007] The purpose of this invention is to solve the above-mentioned problems and provide a self-feedback regularized inversion method.

[0008] To achieve the above objectives, the technical solution of the present invention is as follows:

[0009] This invention provides a self-feedback-type regularized inversion method, comprising the following steps:

[0010] S1. Acquire imaging data;

[0011] S2, Preprocessing imaging data;

[0012] S3. Establish an inversion model, set regularization parameters and threshold parameters, and perform initial inversion to obtain preliminary inversion results;

[0013] S4. Calculate the error feedback coefficient based on the preliminary inversion results;

[0014] S5. Update the inversion model using the error feedback coefficients;

[0015] S6. Iterate the updated inversion model again until the convergence condition is met.

[0016] S7. Perform post-processing operations to obtain the final imaging result.

[0017] Furthermore, the formula for calculating the error feedback coefficient in step S4 is as follows:

[0018] β=γ×(1+ε),

[0019] Where γ is the regularization parameter and ε is the error term.

[0020] Furthermore, the update formula for the inversion model described in step S5 is:

[0021] m k+1 =m k +Δm,

[0022] Where, m k and m k+1 These are the inversion models before and after the iteration, respectively. Δm is the correction amount for the inversion model, calculated using the following formula:

[0023]

[0024] Where λ is the threshold parameter, and J is the inversion objective function. This is the gradient of the objective function with respect to the inversion model.

[0025] Furthermore, the post-processing operations described in step S7 include denoising and smoothing operations.

[0026] Furthermore, the imaging data mentioned in step S1 is electromagnetic wave imaging data.

[0027] Furthermore, the imaging object of the imaging data is underground rock strata.

[0028] Furthermore, the regularization parameter and threshold parameter are parameters set according to the actual application requirements.

[0029] In the above-described scheme of this invention, the method combines traditional regularized inversion methods and self-feedback control techniques, which can effectively improve imaging quality while ensuring the stability of the inversion results. The method proposed in this invention has significant application value and practical significance in electromagnetic wave imaging technology.

[0030] Compared with existing technologies, the beneficial effects of this solution are:

[0031] 1. In the method of the present invention, the self-feedback control technology is used to dynamically adjust the inversion model according to the error generated during the inversion process, which can effectively avoid problems such as over-smoothing and distortion.

[0032] 2. In the method of the present invention, by dynamically adjusting the inversion model during the inversion process, the stability and robustness of the inversion can be improved, and the oscillation and divergence of the inversion results can be avoided.

[0033] 3. The self-feedback regularized inversion method proposed in this invention can be applied to different imaging data types and imaging objects, and has strong universality and applicability. Attached Figure Description

[0034] Figure 1 This is a flowchart of the method in an embodiment of the present invention. Detailed Implementation

[0035] To enable those skilled in the art to better understand the present invention, the technical solution of the present invention will be described in further detail below with reference to the embodiments and accompanying drawings. Obviously, the described embodiments are merely some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort should fall within the scope of protection of the present invention.

[0036] It should be noted that, unless otherwise specified, the embodiments and features described in the embodiments of this invention can be combined with each other. The invention will now be described in detail with reference to the embodiments.

[0037] Example:

[0038] The solution provided by the embodiments of the present invention is as described above, specifically as follows:

[0039] A self-feedback-based regularized inversion method, see [link to relevant documentation]. Figure 1 As shown, it includes the following steps:

[0040] Step 1: Acquire imaging data;

[0041] Step 2: Preprocess the imaging data;

[0042] Step 3: Establish the inversion model, set the regularization parameters and threshold parameters, and perform the initial inversion to obtain preliminary inversion results;

[0043] Step 4: Calculate the error feedback coefficient based on the preliminary inversion results;

[0044] Step 5: Update the inversion model using the error feedback coefficients;

[0045] Step 6: Iterate the updated inversion model again until the convergence condition is met;

[0046] Step 7: Perform post-processing to obtain the final imaging result.

[0047] In this embodiment, the formula for calculating the error feedback coefficient in step four is:

[0048] β=γ×(1+ε),

[0049] Where γ is the regularization parameter and ε is the error term.

[0050] In this embodiment, the update formula for the inversion model in step five is:

[0051] m k+1 =m k +Δm,

[0052] Where, m k and m k+1 These are the inversion models before and after the iteration, respectively. Δm is the correction amount for the inversion model, calculated using the following formula:

[0053]

[0054] Where λ is the threshold parameter, and J is the inversion objective function. This is the gradient of the objective function with respect to the inversion model.

[0055] In this embodiment, the post-processing operations in step seven include noise reduction and smoothing operations.

[0056] In this embodiment, the imaging data in step one is electromagnetic wave imaging data.

[0057] In this embodiment, the imaging object of the imaging data is underground rock strata.

[0058] In this embodiment, the regularization parameter and the threshold parameter are set according to the actual application requirements.

[0059] The following are specific application examples of the solutions in the embodiments of the present invention:

[0060] Taking geophysical exploration as an example, suppose we need to image underground rock strata using electromagnetic wave imaging technology. First, we acquire the imaging data, then perform preprocessing, including noise reduction, filtering, and normalization. Next, we establish an inversion model, setting regularization and threshold parameters according to the actual exploration needs, and perform an initial inversion to obtain preliminary inversion results.

[0061] Based on the preliminary inversion results, error feedback coefficients are calculated, and then the inversion model is updated using these coefficients. The updated inversion model is then iteratively inverted until the convergence condition is met. Finally, post-processing operations, including denoising and smoothing, are performed based on the final inversion results to obtain the final imaging result.

[0062] Through the above examples of this invention, compared with the traditional regularized inversion method in the prior art, the self-feedback regularized inversion method of this invention can effectively improve the imaging quality while ensuring the stability of the inversion results. Furthermore, the practical application results show that the self-feedback regularized inversion method proposed in this invention has significant application value and practical significance in electromagnetic wave imaging technology.

[0063] In summary, the above embodiments of the present invention propose a novel self-feedback regularized inversion method, which combines traditional regularized inversion methods with self-feedback control technology. This method can effectively improve imaging quality while ensuring the stability of the inversion results. In addition, this method has broad application value in electromagnetic wave imaging technology and has important practical significance.

[0064] The above specific embodiments are merely explanations of the present invention and are not intended to limit the present invention. After reading this specification, those skilled in the art can make modifications to these embodiments without contributing any inventive step, but as long as they are within the scope of the claims of the present invention, they are protected by patent law.

Claims

1. A self-feedback type regularized inversion method, characterized by: Includes the following steps: S1. Acquire imaging data; S2, Preprocessing imaging data; S3. Establish an inversion model, set regularization parameters and threshold parameters, and perform initial inversion to obtain preliminary inversion results; S4. Calculate the error feedback coefficient based on the preliminary inversion results; S5. Update the inversion model using the error feedback coefficients; S6. Iterate the updated inversion model again until the convergence condition is met. S7. Perform post-processing operations to obtain the final imaging result; The formula for calculating the error feedback coefficient in step S4 is as follows: ， wherein, is a regularization parameter, is an error term.

2. The self-feedback regularized inversion method as described in claim 1, characterized in that: The update formula for the inversion model mentioned in step S5 is: ， in, and These are the inversion models before and after the iteration, respectively. The correction amount for the inversion model is calculated using the following formula: ， in, Let J be the threshold parameter and J be the inversion objective function. This is the gradient of the objective function with respect to the inversion model.

3. The self-feedback regularized inversion method as described in claim 1, characterized in that: The post-processing operations described in step S7 include denoising and smoothing operations.

4. The self-feedback regularized inversion method as described in claim 1, characterized in that: The imaging data mentioned in step S1 is electromagnetic wave imaging data.

5. The self-feedback regularized inversion method as described in claim 1, characterized in that: The imaging object of the imaging data is underground rock strata.

6. The self-feedback regularized inversion method as described in claim 1, characterized in that: The regularization parameter and threshold parameter are set according to the actual application requirements.

Images