Coal mine multi-source geophysical data fusion method and system based on principal component analysis

Abstract

The application discloses a kind of coal mine multi-source geophysical data fusion method and system based on principal component analysis, method includes: step 1, multi-source data acquisition: attenuation voltage data is obtained using transient electromagnetic method, apparent resistivity data is obtained using audio electric perspective method, wave signal is obtained using slot wave method;Step 2, the result of step 1 is uniformly processed, data gridding, interpolation, resampling, data standardization and normalization, to obtain preprocessed data, correspond to a matrix respectively;Step 3, for each matrix in three matrices, the correlation coefficient matrix of the matrix is calculated, and the component contribution rate and cumulative contribution rate are calculated, then the principal component is obtained according to the cumulative contribution rate, and imaging is carried out.The application gives full play to the synergistic advantages of the three kinds of underground detection methods, effectively solves the problem of single method in complex geological conditions, and improves the abnormal body identification precision.

Description

Technical Field

[0001] This invention belongs to the field of geophysical exploration and relates to a method and system for fusing multi-source geophysical data in coal mines based on principal component analysis. By integrating the unique advantages of three downhole detection methods, this invention solves the detection limitations of single methods under complex geological conditions and is suitable for the accurate exploration of anomalies such as coal mine goaf, water-rich bodies, and collapse columns. Background Technology

[0002] In geophysical exploration practice, single geophysical detection methods have significant limitations. For example, while transient electromagnetic methods are sensitive to low-resistivity media, they are easily affected by surface electromagnetic interference, and the layer resolution decreases with increasing detection depth. Audio-frequency electro-optics can penetrate high-resistivity layers to identify low-resistivity anomalies; however, phase data is greatly affected by the anisotropy of the medium, and the detection range at a single frequency is limited. The channel wave method uses coal seams as waveguide layers and accurately identifies internal coal seam structures (faults, collapse columns) through guided wave propagation characteristics, with high lateral resolution, but its response to non-coal seam geological bodies is weak.

[0003] Current multi-source data fusion technology faces many challenges, including: (1) After the fusion of multi-source data, the dimensionality increases significantly, and the computational complexity increases dramatically; (2) The data scales of the three downhole methods differ greatly, making spatial registration difficult; (3) Data obtained by different methods contain a lot of redundant information in characterization of features such as low resistance anomalies, which leads to unstable results; (4) Method-specific noise superposition will mask the effective signal; (5) The lack of a weight allocation mechanism based on method characteristics means that simple superposition will dilute the advantageous features. Summary of the Invention

[0004] This invention aims to provide a method and system for fusing multi-source geophysical data in coal mines based on principal component analysis. Through an innovative data-driven fusion strategy, it highlights the geological response characteristics of the three methods, effectively improves the identification accuracy of complex geological bodies, and provides a more efficient and accurate technical means for geophysical exploration.

[0005] To achieve the above objectives, the present invention employs the following technical solution: On the one hand, this invention provides a method for fusing multi-source geophysical data from coal mines based on principal component analysis, specifically including the following steps: Step 1, Multi-source data acquisition: Simultaneously deploy transient electromagnetic method survey lines, audio-frequency electro-optical imaging observation points, and slot wave method survey network; use transient electromagnetic method to acquire attenuated voltage data, use audio-frequency electro-optical imaging to acquire apparent resistivity data at frequencies of 1Hz-2kHz, and use slot wave method to acquire wave signals of 50-500Hz and extract propagation velocity and amplitude attenuation coefficient from them; integrate the raw data obtained from transient electromagnetic method, audio-frequency electro-optical imaging method, and slot wave method into a multi-source heterogeneous dataset, and perform denoising and imaging processing respectively, storing the processing results in the form of a multi-dimensional data matrix; Step 2, Preprocessing: The multidimensional data matrix obtained in Step 1 is subjected to spatial coordinate unification, data gridding, interpolation, resampling, data standardization and normalization to obtain three sets of preprocessed geophysical data, each corresponding to a matrix; Step 3, Data Fusion: For each of the three preprocessed matrices obtained in Step 2, calculate the correlation coefficient matrix of the matrix, and calculate the component contribution rate and cumulative contribution rate based on the correlation coefficient matrix. Then, obtain the principal components based on the cumulative contribution rate and use the principal components for imaging. Specifically, the process includes the following steps: Step 31: Expand each of the three preprocessed matrices obtained in Step 2 by columns to obtain a column vector for each type of geophysical data, resulting in a total of three column vectors; calculate the correlation coefficients between each pair of these three column vectors to obtain the correlation coefficient matrix of the three column vectors. Step 32: Calculate the eigenvalues ​​and eigenvectors of the correlation coefficient matrix; Step 33: Calculate the principal component contribution rate and cumulative contribution rate based on the eigenvalues ​​of the correlation coefficient matrix obtained in Step 32, and select the principal components based on the cumulative contribution rate; Step 34: Multiply the feature vector obtained in step 32 by the three column vectors obtained in step 31 to obtain the three principal components respectively; Step 35: Use principal components to perform imaging and obtain the imaging results.

[0006] On the other hand, the present invention provides a coal mine multi-source geophysical data fusion system based on principal component analysis, specifically including the following modules: Multi-source data acquisition module: Used to complete the following processes: synchronously deploy transient electromagnetic method survey lines, audio-frequency electro-optical observation points, and slot wave method survey networks; acquire attenuated voltage data using transient electromagnetic method, acquire apparent resistivity data at frequencies of 1Hz-2kHz using audio-frequency electro-optical method, acquire wave signals of 50-500Hz using slot wave method, and extract propagation velocity and amplitude attenuation coefficient from them; integrate the raw data obtained from transient electromagnetic method, audio-frequency electro-optical method, and slot wave method into a multi-source heterogeneous dataset, and perform denoising and imaging processing respectively, storing the processing results in the form of a multi-dimensional data matrix; The preprocessing module is used to perform spatial coordinate unification, data gridding, interpolation, resampling, data standardization, and normalization on the multidimensional data matrix obtained by the multi-source data acquisition module, resulting in three preprocessed sets of geophysical data, each corresponding to a matrix. The data fusion module performs the following operations: For each of the three preprocessed matrices output by the preprocessing module, it calculates the correlation coefficient matrix, and then calculates the component contribution rate and cumulative contribution rate based on the correlation coefficient matrix. Finally, it obtains the principal components based on the cumulative contribution rate and uses these principal components for imaging. Specifically, the process includes the following steps: S1. Expand each of the three preprocessed matrices obtained from the preprocessing module by columns to obtain a column vector for each type of geophysical data, resulting in a total of three column vectors; calculate the correlation coefficients between each pair of these three column vectors to obtain the correlation coefficient matrix of the three column vectors; S2, calculate the eigenvalues ​​and eigenvectors of the correlation coefficient matrix; S3, calculate the principal component contribution rate and cumulative contribution rate based on the eigenvalues ​​of the correlation coefficient matrix obtained in step 32, and select the principal component based on the cumulative contribution rate; S4, Multiply the feature vector obtained in step 32 by the three column vectors obtained in step 31 to obtain the three principal components respectively; S5 uses principal components for imaging to obtain the imaging results.

[0007] Compared with existing technologies, this invention fully utilizes the characteristics of transient electromagnetic method, audio-visual radiography method and channel wave method to achieve complementary advantages, so that the low resistance sensitivity of transient electromagnetic method, the high resistance penetration of audio-visual radiography method and the coal seam structural response of channel wave method are fully preserved. It not only solves the problem of data scale difference among the three methods, but also has a fast calculation speed. At the same time, it effectively reduces noise variance and improves the stability and reliability of data. The fusion feature has a significantly improved sensitivity to low resistance anomaly identification compared with the single method, making the features of geological anomalies more prominent and facilitating accurate identification and location. The fusion method based on principal component analysis of this invention has the following main features: (1) After the original variables are combined into several factors, the factors can represent the structural relationship between the original variables and respond to the functional relationship of the original variables, and the logic of the functional relationship will be strengthened. (2) Principal components can recombine information while preserving most of the information, and can effectively represent the original data. (3) The principal components are uncorrelated with each other, which can effectively solve the problem of overlapping variable information. Attached Figure Description

[0008] Figure 1 A flowchart for multi-source geophysical fusion; Figure 2 This is a schematic diagram of a single coal seam model; Figure 3 This is a forward modeling profile of the trough wave seismic method; Figure 4 This is a cross-sectional view of the forward modeling using the transient electromagnetic method. Figure 5 This is a cross-sectional view created using the audio-visual imaging method. Figure 6 This is the result of single coal seam fusion; Figure 7 This is a schematic diagram of a collapse column model; Figure 8 This is a forward modeling profile of the trough wave seismic method; Figure 9 This is a cross-sectional view of the transient electromagnetic method forward modeling; Figure 10 This is a cross-sectional view created using the audio-visual imaging method. Figure 11 This is the fusion result of the collapsed column model. Detailed Implementation

[0009] like Figure 2 As shown, the coal mine multi-source geophysical data fusion method based on principal component analysis (PCA) provided by this invention specifically includes the following steps: Step 1, Multi-source Data Acquisition: Transient electromagnetic (TEM) survey lines, audio-frequency electro-optical (AFE) observation points, and slot wave method survey networks are deployed simultaneously, ensuring an overlap of over 80% in the detection ranges of the three methods. Attenuated voltage data is acquired using TEM, apparent resistivity data at frequencies of 1Hz-2kHz is acquired using AFE, and wave signals at 50-500Hz are acquired using slot wave methods, from which propagation velocity and amplitude attenuation coefficients are extracted. The raw data obtained from TEM, AFE, and slot wave methods are integrated into a multi-source heterogeneous dataset, and denoising and imaging processing are performed separately. The processing results are stored in the form of a multi-dimensional data matrix.

[0010] The specific steps are as follows: Transient electromagnetic method: A small loop device (side length 1~3m) is used to collect the secondary field attenuation signal in the time period of 0.01ms-10ms (corresponding to the induced voltage data of the target in the depth range of 0-80m). Audio electrofluorometry: Using a bipolar power supply mode, specifically an electrode distance of 1~10m, frequency data of 1Hz-2kHz are collected (resistivity data within the range of 0-80m are collected) to highlight the low resistivity anomaly in high resistivity coal seams. The channel wave method employs a hammer-driven seismic source and a geophone array, specifically using a grid spacing of 1–10 m, a sampling interval of 0.1 ms, and a 21-channel geophone array (channel spacing 10 m) to acquire wave signals of 50–500 Hz (primarily acquiring wave signals within the 0–80 m range inside the coal seam), and extracting propagation velocity and amplitude attenuation coefficients. This allows for the identification of coal seam faults, collapse columns, and other structural features.

[0011] After collecting the above three types of data, denoising processing is required to improve data quality. Then, imaging or inversion processing is performed on the three types of data. The transient electromagnetic method uses a full-area apparent resistivity algorithm to convert the attenuated voltage, obtaining a two-dimensional apparent resistivity profile; the audio-visual imaging method uses two-dimensional inversion processing to obtain a two-dimensional resistivity profile; the slot wave method uses cross-correlation imaging conditions to obtain reverse-time migration imaging results; finally, the processing results are stored in the form of a multi-dimensional data matrix.

[0012] Among the aforementioned technical solutions, the transient electromagnetic method uses a small loop device to detect targets in the well. It is sensitive to low-resistivity bodies but is susceptible to electromagnetic interference and has poor sensitivity to high-resistivity bodies. The audio-visual electron microscopy method is sensitive to changes in the lithology of the top and bottom strata of coal seams and can detect changes in water-bearing properties within the strata of the working face, but its detection range is limited. The channel wave method has high resolution, a large detection range, and strong anti-interference ability, but it is strictly dependent on coal seam conditions. To address these limitations, the geophysical data from the three methods are fused to achieve complementary advantages, overcome the limitations of a single method under complex geological conditions, and improve the accuracy of anomaly identification. Step 2, Preprocessing: The multidimensional data matrix obtained in Step 1 is subjected to spatial coordinate unification, data gridding, interpolation, resampling, data standardization and normalization to obtain three sets of preprocessed geophysical data, each corresponding to a matrix.

[0013] Specifically, the coordinates of the multi-source geophysical data in the multi-dimensional data matrix obtained in step 1 are transformed to the same coordinate system; further processing is performed using Kriging interpolation or natural nearest neighbor interpolation to register their spatial coordinates; and then... z-score standardization is performed on the interpolated data. μ For column mean, σ The standard deviation is used to eliminate the influence of dimensions, unify the data distribution, and make heterogeneous data fusionable. The standardized data is then subjected to noise suppression to obtain preprocessed three-source geophysical data, each corresponding to a matrix, for a total of three matrices.

[0014] Step 3, Data Fusion: For each of the three preprocessed matrices obtained in Step 2, calculate the correlation coefficient matrix of the matrix, and calculate the component contribution rate and cumulative contribution rate based on the correlation coefficient matrix. Then, obtain the principal components based on the cumulative contribution rate and use the principal components for imaging.

[0015] Specifically, the process includes the following steps: Step 31: Expand each of the three preprocessed matrices obtained in Step 2 by columns to obtain a column vector for each type of geophysical data, resulting in a total of three column vectors; calculate the correlation coefficients between each pair of these three column vectors to obtain the correlation coefficient matrix of the three column vectors.

[0016] Specifically, it refers to the correlation coefficient matrix R between the two column vectors:

[0017] in,

[0018] In the formula, For each of two column vectors, one column vector and another column vector The correlation coefficient, =3; This represents the first element of a column vector. , These represent the means of the two column vectors, respectively. n The number of elements in a column vector.

[0019] Step 32: Calculate the eigenvalues ​​and eigenvectors of the correlation coefficient matrix.

[0020] Specifically: Calculate the eigenvalues ​​of the correlation coefficient matrix R. And arrange them in order of size, that is Then, calculate the corresponding eigenvalues ​​respectively. eigenvectors This requires ,Right now ,in, Representing vectors The j-th component.

[0021] Step 33: Calculate the principal component contribution rate and cumulative contribution rate based on the eigenvalues ​​of the correlation coefficient matrix obtained in step 32, and select the principal components based on the cumulative contribution rate.

[0022] Specifically: principal components The contribution rate The cumulative contribution rate is Select the first k principal components such that the cumulative contribution rate k is the smallest positive integer that satisfies this condition.

[0023] Step 34, Generate Principal Components: Multiply the feature vector obtained in Step 32 by the three column vectors obtained in Step 31 to obtain the three principal components: , , ; The order of principal components is determined by the magnitude of their eigenvalues; the larger the eigenvalue, the more information the corresponding principal component contains from the original data. In this way, principal component 1, principal component 2, and principal component 3 sequentially extract the most important, second most important, and third most important information dimensions from the original data, providing a basis for subsequent data fusion and geological interpretation.

[0024] Step 35: Use the principal components (such as principal component 1, principal component 2) selected in step 34 to perform imaging and obtain the imaging results.

[0025] Based on the imaging results from step 35, further analysis can be performed. Therefore, step 3 may also include the following steps: Step 36 involves geological interpretation of the imaging results from Step 35. This includes analyzing low-resistivity / high-resistivity anomalies in the imaging and, in conjunction with the mine's geological background (such as the distribution of goafs and faults), determining the location, scale, and nature of adverse geological bodies. The operations in this step are all well-known techniques in the field and will not be described in detail here.

[0026] The coal mine multi-source geophysical data fusion system based on principal component analysis provided in this invention specifically includes the following modules: Multi-source data acquisition module: Used to complete the following processes: synchronously deploy transient electromagnetic method survey lines, audio-frequency electro-optical observation points, and slot wave method survey networks; acquire attenuated voltage data using transient electromagnetic method, acquire apparent resistivity data at frequencies of 1Hz-2kHz using audio-frequency electro-optical method, acquire wave signals of 50-500Hz using slot wave method, and extract propagation velocity and amplitude attenuation coefficient from them; integrate the raw data obtained from transient electromagnetic method, audio-frequency electro-optical method, and slot wave method into a multi-source heterogeneous dataset, and perform denoising and imaging processing respectively, storing the processing results in the form of a multi-dimensional data matrix; The preprocessing module is used to perform spatial coordinate unification, data gridding, interpolation, resampling, data standardization, and normalization on the multidimensional data matrix obtained by the multi-source data acquisition module, resulting in three preprocessed sets of geophysical data, each corresponding to a matrix. The data fusion module performs the following operations: For each of the three preprocessed matrices output by the preprocessing module, it calculates the correlation coefficient matrix, and then calculates the component contribution rate and cumulative contribution rate based on the correlation coefficient matrix. Finally, it obtains the principal components based on the cumulative contribution rate and uses these principal components for imaging. Specifically, the process includes the following steps: S1. Expand each of the three preprocessed matrices obtained from the preprocessing module by columns to obtain a column vector for each type of geophysical data, resulting in a total of three column vectors; calculate the correlation coefficients between each pair of these three column vectors to obtain the correlation coefficient matrix of the three column vectors; S2, calculate the eigenvalues ​​and eigenvectors of the correlation coefficient matrix; S3, calculate the principal component contribution rate and cumulative contribution rate based on the eigenvalues ​​of the correlation coefficient matrix obtained in step 32, and select the principal component based on the cumulative contribution rate; S4, Multiply the feature vector obtained in step 32 by the three column vectors obtained in step 31 to obtain the three principal components respectively; S5 uses principal components for imaging to obtain the imaging results.

[0027] This invention fully utilizes the characteristics of transient electromagnetic methods, audio-visual imaging, and channel wave methods, achieving complementary advantages. It fully preserves the low-resistivity sensitivity of transient electromagnetic methods, the high-resistivity penetration of audio-visual imaging, and the coal seam structural response of channel wave methods. This not only solves the data scale difference problem among the three methods but also significantly improves computational efficiency. Simultaneously, it effectively reduces noise variance, improving data stability and reliability. The fused features significantly enhance the sensitivity of low-resistivity anomaly identification compared to individual methods, making the characteristics of geological anomalies more prominent and facilitating accurate identification and location. This invention can be flexibly adjusted according to different geological conditions and exploration needs, adapting to complex and changing geological environments. It has broad application prospects in multiple fields such as coal mine water hazard early warning, mineral resource exploration, and geological structure research, and is particularly suitable for mines with complex geological conditions such as multiple coal seams, high gas content, and abundant water.

[0028] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments, but the implementation of the present invention is not limited thereto.

[0029] Application Example 1: Fusion of Single Coal Seam Models First, set as follows: Figure 2 The geological model shown is of a single high-resistivity coal seam, with a burial depth of 45m and a resistivity set to 500. The coal seam is 10m deep, and the resistivity of the overlying and underlying strata is 100 Ω·cm. The model was forward modeled using the channel wave seismic method, transient electromagnetic method, and audio-frequency electron microscopy method, respectively. The corresponding forward modeling profile results are as follows: Figure 3 , Figure 4 , Figure 5 As shown. The forward modeling results show that the reflection coefficient profile obtained from the channel wave seismic model clearly shows the upper and lower interfaces of the coal seam, the anomaly position in the transient electromagnetic forward modeling results is biased upwards, while the anomaly position in the audio-visual fluoroscopy forward modeling is biased downwards. Data fusion of the spatially registered and normalized data yields the following principal component contribution table and correlation coefficient table: Table 1 Principal Component Contribution Table

[0030] Table 2 Correlation Coefficient Table

[0031] Based on the principal component contributions, the cumulative contribution rates of principal component one and principal component two are close to 1, which is sufficient to represent the content of the three original data graphs. Among them, principal component one accounts for a larger proportion, about 65%, and principal component one summarizes most of the original data.

[0032] As shown in the correlation coefficient table, Principal Component 1 and Principal Component 3 have extremely high correlations with the two sets of data from the transient electromagnetic method and the audio-frequency electro-optic method, but Principal Component 3 has a low correlation with the channel wave seismic method, while Principal Component 1 has a high correlation with the channel wave method; and Principal Component 2 has a low correlation with the three sets of data.

[0033] In summary, Principal Component Analysis (PCA) can be considered essentially the result of PCA data fusion processing, such as... Figure 6 As shown in the diagram, the analysis of the fusion results reveals that the fusion and identification of coal seam interface and relative physical property data are excellent. Compared with the original data, this data fusion completely identifies the location of the coal seam and its high-value characteristics.

[0034] Application Example 2: Fusion of Falling Column Models Settings such as Figure 7 The geological model of the coal seam with collapse column shown has a high resistivity coal seam buried at a depth of 45m and a resistivity set at 500. The coal seam depth is 10m; the resistivity of the overlying and underlying strata is 200 Ω·m. The collapse column has a burial depth of 35m at its apex, a horizontal coordinate of 150m, and a total thickness of 30m. Forward modeling of the collapse column was performed using the channel wave seismic method, transient electromagnetic method, and audio-frequency electron microscopy method, respectively. The forward modeling profiles obtained by each method are shown below. Figure 8 , Figure 9 , Figure 10 As shown. The forward modeling results show that the reflection coefficient profile obtained from the channel wave seismic model clearly shows the upper and lower interfaces of the coal seam and the collapse column structure. The transient electromagnetic forward modeling results show that the collapse column anomaly is shifted to the left, while the audio-visual fluoroscopy forward modeling shows that the collapse column anomaly is shifted to the lower part. After spatial registration and normalization, the data is fused to obtain the following principal component contribution table and correlation coefficient table: Table 3 Principal Component Contribution Table

[0035] Table 4 Correlation Coefficient Table

[0036] According to the principal component contribution rate, the cumulative contribution rate of principal component 1 and principal component 2 is 0.83, which can basically represent the content of the three original data graphs. Among them, principal component 1 accounts for a larger proportion, at 48%, and principal component 2 accounts for 35%. Principal component 1 and principal component 2 summarize the vast majority of the original data.

[0037] As shown in the correlation coefficient table, Principal Component 1 has a high correlation with transient electromagnetic and channel wave seismic data; while Principal Component 2 has a low correlation with all three sets of data.

[0038] In summary, principal component analysis can be considered as the result of data fusion processing, such as... Figure 11 As shown.

[0039] Therefore, a multi-source geophysical data fusion method based on transient electromagnetic method, audio-frequency electro-optical imaging method and channel wave method was adopted to realize the fusion processing of geophysical data obtained by multiple methods. The calculation results of the model show that this method can be applied to actual exploration and is an effective data fusion method with great development potential.

[0040] The two embodiments have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the specific details in the above embodiments. Within the scope of the technical concept of the present invention, various simple modifications can be made to the technical solution, and these simple modifications all fall within the protection scope of the present invention.

[0041] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable way without contradiction. In order to avoid unnecessary repetition, the various possible combinations will not be described separately here.

[0042] Furthermore, various implementation methods of this embodiment can be combined arbitrarily, as long as they do not violate the spirit of the technical solution of this invention, they should also be considered within the scope of protection of this invention.

Claims

1. A method for fusing multi-source geophysical data from coal mines based on principal component analysis, characterized in that, Specifically, the steps include the following: Step 1, Multi-source data acquisition: Simultaneously deploy transient electromagnetic method survey lines, audio-frequency electro-optical imaging observation points, and slot wave method survey network; use transient electromagnetic method to acquire attenuated voltage data, use audio-frequency electro-optical imaging to acquire apparent resistivity data at frequencies of 1Hz-2kHz, and use slot wave method to acquire wave signals of 50-500Hz and extract propagation velocity and amplitude attenuation coefficient from them; integrate the raw data obtained from transient electromagnetic method, audio-frequency electro-optical imaging method, and slot wave method into a multi-source heterogeneous dataset, and perform denoising and imaging processing respectively, storing the processing results in the form of a multi-dimensional data matrix; Step 2, Preprocessing: The multidimensional data matrix obtained in Step 1 is subjected to spatial coordinate unification, data gridding, interpolation, resampling, data standardization and normalization to obtain three sets of preprocessed geophysical data, each corresponding to a matrix; Step 3, Data Fusion: For each of the three preprocessed matrices obtained in Step 2, calculate the correlation coefficient matrix of the matrix, and calculate the component contribution rate and cumulative contribution rate based on the correlation coefficient matrix. Then, obtain the principal components based on the cumulative contribution rate and use the principal components for imaging. Specifically, the process includes the following steps: Step 31: Expand each of the three preprocessed matrices obtained in Step 2 by columns to obtain a column vector for each type of geophysical data, resulting in a total of three column vectors; calculate the correlation coefficients between each pair of these three column vectors to obtain the correlation coefficient matrix of the three column vectors. Step 32: Calculate the eigenvalues ​​and eigenvectors of the correlation coefficient matrix; Step 33: Calculate the principal component contribution rate and cumulative contribution rate based on the eigenvalues ​​of the correlation coefficient matrix obtained in Step 32, and select the principal components based on the cumulative contribution rate; Step 34: Multiply the eigenvectors obtained in step 32 by the three column vectors obtained in step 31 to obtain the three principal components; Step 35: Use principal components to perform imaging and obtain the imaging results.

2. The coal mine multi-source geophysical data fusion method based on principal component analysis as described in claim 1, characterized in that, In step 1, ensure that the detection range overlap rate of the electromagnetic method, audio electro-optical imaging method, and slot wave method remains above 80%.

3. The coal mine multi-source geophysical data fusion method based on principal component analysis as described in claim 1, characterized in that, In step 2, specifically, the coordinates of the multi-source geophysical data in the multi-dimensional data matrix obtained in step 1 are transformed to the same coordinate system; further processing is performed using Kriging interpolation or natural nearest neighbor interpolation to register their spatial coordinates; and then... The interpolated data is then standardized using z-score, where μ For column mean, σ The standard deviation is denoted as . Noise suppression is applied to the standardized data to obtain preprocessed geophysical data from the three sources, each corresponding to a matrix, for a total of three matrices.

4. The coal mine multi-source geophysical data fusion method based on principal component analysis as described in claim 1, characterized in that, In step 31, the correlation coefficient matrix R between the two column vectors is: in, In the formula, For each of two column vectors, one column vector and another column vector The correlation coefficient, =3; This represents the first element of a column vector. , These represent the means of the two column vectors, respectively. n The number of elements in a column vector.

5. The coal mine multi-source geophysical data fusion method based on principal component analysis as described in claim 1, characterized in that, In step 32, the eigenvalues ​​of the correlation coefficient matrix R are calculated. And arrange them in order of size, that is Then, calculate the corresponding eigenvalues ​​respectively. eigenvectors This requires ,Right now ,in, Representing vectors The j-th component.

6. The method for fusing multi-source geophysical data in coal mines based on principal component analysis as described in claim 1, characterized in that, In step 33, principal components The contribution rate The cumulative contribution rate is Select the first k principal components such that the cumulative contribution rate k is the smallest positive integer that satisfies this condition.

7. The coal mine multi-source geophysical data fusion method based on principal component analysis as described in claim 1, characterized in that, In step 34, the three principal components are as follows: , , .

8. The coal mine multi-source geophysical data fusion method based on principal component analysis as described in any one of claims 1 to 7, characterized in that, Step 3 also includes step 36, which involves geological interpretation of the imaging results from step 35: analyzing low-resistivity / high-resistivity anomalies in the imaging and, in conjunction with the geological background of the mine, determining the location, scale, and nature of adverse geological bodies.

9. A coal mine multi-source geophysical data fusion system based on principal component analysis, characterized in that, Specifically, it includes the following modules: Multi-source data acquisition module: Used to complete the following processes: synchronously deploy transient electromagnetic method survey lines, audio-frequency electro-optical observation points, and slot wave method survey networks; acquire attenuated voltage data using transient electromagnetic method, acquire apparent resistivity data at frequencies of 1Hz-2kHz using audio-frequency electro-optical method, acquire wave signals of 50-500Hz using slot wave method, and extract propagation velocity and amplitude attenuation coefficient from them; integrate the raw data obtained from transient electromagnetic method, audio-frequency electro-optical method, and slot wave method into a multi-source heterogeneous dataset, and perform denoising and imaging processing respectively, storing the processing results in the form of a multi-dimensional data matrix; The preprocessing module is used to perform spatial coordinate unification, data gridding, interpolation, resampling, data standardization, and normalization on the multidimensional data matrix obtained by the multi-source data acquisition module, resulting in three preprocessed sets of geophysical data, each corresponding to a matrix. The data fusion module performs the following operations: For each of the three preprocessed matrices output by the preprocessing module, it calculates the correlation coefficient matrix, and then calculates the component contribution rate and cumulative contribution rate based on the correlation coefficient matrix. Finally, it obtains the principal components based on the cumulative contribution rate and uses these principal components for imaging. Specifically, the process includes the following steps: S1. Expand each of the three preprocessed matrices obtained from the preprocessing module by columns to obtain a column vector for each type of geophysical data, resulting in a total of three column vectors; calculate the correlation coefficients between each pair of these three column vectors to obtain the correlation coefficient matrix of the three column vectors; S2, calculate the eigenvalues ​​and eigenvectors of the correlation coefficient matrix; S3, calculate the principal component contribution rate and cumulative contribution rate based on the eigenvalues ​​of the correlation coefficient matrix obtained in step 32, and select the principal component based on the cumulative contribution rate; S4, Multiply the feature vector obtained in step 32 by the three column vectors obtained in step 31 to obtain the three principal components respectively; S5 uses principal components for imaging to obtain the imaging results.

Images