Rock slice positioning detection method combining multiple in-situ test instruments

Abstract

The invention relates to the technical field of rock slice detection, and particularly discloses a rock slice positioning detection method combining multiple in-situ test instruments, which comprises the following steps: preparing a sample, slicing a rock sample to be detected, and transversely cutting the original sample into two parts to form a transverse section I and a transverse section II which completely correspond to each other in position and size, the parallelism error of the two sections is less than or equal to 50 microns, and the area difference is less than or equal to 1 Instrument calibration: designing a general sample table, and calibrating a plurality of in-situ test instruments by matching an SEM (scanning electron microscope), an AFM (atomic force microscope), a Raman spectrometer and micro-area X-ray fluorescence equipment; platinum and carbon nano-particle marks (the precision is smaller than 100 nm) deposited through electron beam induction are fixed to the edge of the thin sheet through an atomic-scale combination technology, compared with traditional micron-scale marks (the error is larger than 1 micron), the positioning precision is improved by 10 times or above, and nano-scale view fields of TEM and SEM instruments can be accurately matched.

Description

Technical Field

[0001] This invention relates to the field of rock thin section testing technology, and in particular to a method for locating and testing rock thin sections using multiple in-situ testing instruments in combination. Background Technology

[0002] The exploration and development of fossil fuels and mineral resources both rely on rock testing technology. As these fields continue to expand, the requirements for rock testing technology are also constantly increasing.

[0003] Traditional rock thin section examination often uses a single instrument or manual marking. For example, when using a polarizing microscope in conjunction with SEM, the field of view needs to be manually aligned, and the positioning error is usually in the micrometer range (1-10μm).

[0004] Furthermore, traditional methods rely on manual positioning, have large cross-instrument detection errors (>1μm), low efficiency (single sample analysis requires 24 hours), and cannot achieve dynamic process analysis.

[0005] Therefore, there is an urgent need for a more efficient and accurate method to guide and assist in the localization techniques of multiple in-situ tests, so as to promote the solution of rock science problems. Summary of the Invention

[0006] The purpose of this invention is to provide a method for locating and detecting thin sections of rock using multiple in-situ testing instruments in combination, so as to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for locating and detecting thin sections of rock using multiple in-situ testing instruments in combination, comprising the following steps:

[0008] S1: Sample preparation. The rock sample to be tested is sliced ​​and then transversely cut into two parts to form cross-sections I and II. The two sections are completely corresponding in position and size, and the parallelism error between the two sections is ≤50μm and the area difference is ≤1%.

[0009] S2: Instrument calibration, design of a universal sample stage, and compatibility with SEM, AFM, Raman spectrometers and micro-area X-ray fluorescence equipment to calibrate multiple in-situ testing instruments;

[0010] S3: Combined setup, based on the characteristics of rock thin sections and detection requirements, covers test location, depth and time factors to achieve joint detection;

[0011] S4: Data recording and analysis, recording the test data of each device and performing comprehensive analysis;

[0012] S5: Model Building: Based on the detection data recorded in step S4, construct a three-dimensional digital model of the rock thin section.

[0013] Preferably, in step S1, the surface of the sheet is subjected to chemical mechanical polishing, and epoxy resin is coated and cured at the edge of the sheet;

[0014] In step S1, ion beam deposition is used to deposit platinum or carbon nanoparticles on the surface of rock thin sections, enabling atomic-level bonding between the sample surface atoms and the deposited material.

[0015] Preferably, the feature is that: in step S2, a heating and cooling module is integrated to be compatible with temperature-sensitive analysis of SEM and Raman spectrometers, and a micro electromagnet is embedded at the bottom of the sample stage to control the submicron displacement of the sample stage among multiple instruments through an external magnetic field;

[0016] The temperature control calculation formula is as follows:

[0017] Temperature fluctuation range = (actual T - set T) ≤ ±5℃;

[0018] Wherein, T_actual_measured represents the temperature value actually measured by the temperature sensor device, and T_set refers to the target temperature value that has been set in advance.

[0019] Preferably, in step S2, an automatic recognition algorithm based on a convolutional neural network is developed to extract features from a standard grid image, including grid intersections, and automatically generate a coordinate system transformation matrix suitable for scanning electron microscopes and atomic force microscopes.

[0020] By using standard samples with known three-dimensional structures, including nanowire arrays, data were acquired in various instruments, and calibration accuracy was assessed through three-dimensional reconstruction error to ensure that the cross-instrument positioning error was less than 50 nm.

[0021] The positioning error formula is as follows:

[0022]

[0023] Where x1, y1, and z1 are the coordinates of the standard sample in the scanning electron microscope, and x2, y2, and z2 are the coordinates of the standard sample in the atomic force microscope.

[0024] Preferably, in step S3, the parameters for joint use of multiple in-situ testing devices are configured, sample calibration is performed, images are acquired in each device using standard grid samples, and a coordinate system transformation matrix is ​​constructed using a feature point matching algorithm.

[0025] Preferably, in step S3, a preliminary microscopic image of a rock thin section is input, and the distribution of the region of interest is predicted through a U-Net network to automatically generate the optimal detection sequence of performing Raman spectroscopy analysis first and then focused ion beam cutting.

[0026] Among them, Raman spectroscopy is non-destructive, while focused ion beam cutting is destructive;

[0027] A hardware triggering mechanism is established between devices, and the EEE1588 precision clock protocol is adopted.

[0028] Preferably, in step S4, a real-time data interface based on Python is developed to synchronously map the BSE image of SEM, the component distribution of Raman spectrum, and the elemental concentration data of X-ray fluorescence to a unified coordinate system, forming a dynamically superimposed four-dimensional data cube. The dynamically superimposed four-dimensional data cube includes spatial X, Y, and Z dimensions and a time t dimension.

[0029] Preferably, in step S4, the detection data of each device is collected through a data collection device, and outlier data points are automatically detected using the isolated forest algorithm and marked in the three-dimensional model.

[0030] Preferably, in step S5, for areas with complex mineral distribution, high-resolution transmission electron microscopy data is used, scanning electron microscopy data is used on the visualization area, and the data is fused using a pyramid layering algorithm to construct a three-dimensional model that includes both details and a global perspective.

[0031] The formula for the pyramid layering algorithm is as follows:

[0032]

[0033] Where M represents the fusion result obtained after processing, Pyramid represents the pyramid algorithm, TEM_data represents transmission electron microscopy, and SEM_data represents scanning electron microscopy. This represents a fusion operation.

[0034] Preferably, in step S5, optical flow analysis is introduced to analyze two-dimensional image sequences at different time points to reconstruct the dynamic process of mineral growth or alteration, forming a "four-dimensional digital twin model," i.e., a 3D space X

[0035] Y, z + time t dimension;

[0036] The vector calculation formula for the optical flow method is as follows:

[0037] I X u+I Y v+I t =0;

[0038] Among them, I X I Y I t Let u be the image gradient, and v be the pixel displacement vectors.

[0039] Compared with the prior art, the beneficial effects of the present invention are:

[0040] 1. This invention employs electron beam-induced deposition of platinum and carbon nanoparticles (accuracy <100nm) for labeling, which are then fixed to the edge of thin sections using atomic-level bonding technology. Compared to traditional micron-level labeling (error >1μm), this method improves positioning accuracy by more than 10 times and can precisely match the nanoscale field of view of TEM and SEM instruments. Combined with automatic feature extraction technology based on CNN algorithm, coordinate system calibration is completed within five minutes, with cross-instrument positioning error <50nm, significantly improving the accuracy of micro-area detection. This method is a creative combination of materials science, computer vision, and geological detection, achieving precise positioning and efficient detection of rock thin sections, and providing more comprehensive data support for the study of rock thin sections.

[0041] 2. This invention integrates a temperature control module (temperature fluctuation ≤ ±5℃) with an intelligent sample stage, and submicron-level displacement control (error < 50nm) with a micro-electromagnet. Combined with U-Net network prediction of regions of interest, it automatically generates the optimal path from non-destructive analysis to destructive analysis, reducing redundant scanning by up to 40%, improving overall detection efficiency by 3 times, and reducing sample loss rate from >30% in traditional methods to <5%. At the same time, the standardized process shortens the single-sample analysis time to 8 hours. This method has a high degree of automation and intelligence, which can significantly reduce the workload of researchers and improve research efficiency.

[0042] 3. This invention constructs a "four-dimensional data cube" (spatial XY-Z + time t) through a Python real-time data interface, integrating multi-source data such as SEM morphology, Raman composition, and X-ray elemental concentration, and tracking mineral changes through the time dimension. Combining optical flow and pyramid layering algorithms, it not only achieves cross-scale fusion of mineral distribution (resolution ≤2nm) and macroscopic structure, but also reconstructs mineral growth rate (and alteration path), forming a three-dimensional digital twin model of "microstructure-composition-time". This is the first time that dynamic quantitative analysis of rock microstructure evolution has been achieved, providing a scientific basis for the stability analysis and design of underground engineering.

[0043] 4. In the field of oil and gas reservoirs, this invention can accurately identify nanoscale pore structure characteristics, such as in shale gas reservoirs, providing key data for seepage simulation. In mineral exploration, it can reveal isotopic evidence of diagenetic stages and fluid evolution through cross-instrument combinations, such as the linkage between LA-ICP-MS and cathodoluminescence. In deep engineering, it can dynamically analyze rock crack propagation dynamics and predict mechanical stability using a four-dimensional model, providing forward-looking support for underground engineering design. This method breaks through the limitations of single instruments, constructs an integrated "location-analysis-modeling" process, and promotes the development of rock detection towards intelligence and dynamism. Attached Figure Description

[0044] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0045] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0046] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0047] Please see Figure 1 This invention provides a technical solution:

[0048] Example 1, please refer to Figure 1 As shown, a method for locating and detecting rock thin sections using multiple in-situ testing instruments includes the following steps:

[0049] Includes the following steps:

[0050] S1: Sample preparation. The rock sample to be tested is sliced ​​and then transversely cut into two parts to form cross-sections I and II. The two sections are completely corresponding in position and size, and the parallelism error between the two sections is ≤50μm and the area difference is ≤1%; to ensure the spatial consistency of the two sections and provide a matching basis for subsequent multi-instrument joint detection.

[0051] S2: Instrument calibration, design of a universal sample stage, and compatibility with SEM, AFM, Raman spectrometers and micro-area X-ray fluorescence equipment to calibrate multiple in-situ testing instruments;

[0052] S3: Combined setup, based on the characteristics of rock thin sections and detection requirements, covers test location, depth and time factors to achieve joint detection;

[0053] S4: Data recording and analysis, recording the test data of each device and performing comprehensive analysis;

[0054] S5: Model Building: Based on the detection data recorded in step S4, construct a three-dimensional digital model of the rock thin section.

[0055] Platinum and carbon nanoparticles, deposited using electron beam-induced deposition, are fixed to the edges of thin sections using atomic-level bonding technology. Compared to traditional micron-level markers, this method offers more than 10 times the positioning accuracy and can precisely match the nanoscale field of view of TEM and SEM instruments. Combined with automatic feature extraction technology based on CNN algorithms, coordinate system calibration is completed within five minutes, with cross-instrument positioning error <50nm, significantly improving the accuracy of micro-area detection. This method is a creative combination of materials science, computer vision, and geological testing, achieving precise positioning and efficient detection of rock thin sections, and providing more comprehensive data support for rock thin section research.

[0056] Example 2, please refer to Figure 1 As shown, in step S1, the surface of the thin film is chemically and mechanically polished, and epoxy resin is coated and cured at the edge of the thin film. The thickness of the epoxy resin coating is controlled between 5 and 10 micrometers to ensure that it can firmly adhere to the edge of the thin film after curing without affecting the subsequent electron beam induced deposition operation. The curing process adopts a constant temperature heating method, with the temperature controlled between 60 and 80 degrees Celsius for 2 to 4 hours to ensure that the epoxy resin is completely cured and forms a stable marking base.

[0057] Platinum or carbon nanoparticles are deposited on the surface of rock thin sections using ion beam deposition technology, enabling atomic-level bonding between the sample surface atoms and the deposited material. The platinum or carbon nanoparticle markers exhibit excellent stability and corrosion resistance, and can be clearly visualized under an electron microscope, providing reliable reference points for subsequent precise localization.

[0058] In step S2, a heating and cooling module is integrated to be compatible with temperature-sensitive analysis of SEM and Raman spectrometers. Furthermore, a micro electromagnet is embedded at the bottom of the sample stage to control the submicron displacement of the sample stage among multiple instruments via an external magnetic field.

[0059] The temperature control calculation formula is as follows:

[0060] Temperature fluctuation range = (actual T - set T) ≤ ±5℃;

[0061] Wherein, T_actual_measured represents the temperature value actually measured by the temperature sensor device, and T_set refers to the target temperature value that has been set in advance.

[0062] Temperature sensors are precisely installed near the sample stage to ensure that temperature changes on the sample stage can be captured in real time and accurately. Through a high-precision data acquisition system, the measured T data value is continuously recorded and compared with the T set value. When a temperature fluctuation is detected to exceed the preset range, the system automatically activates the adjustment mechanism to quickly adjust the sample stage temperature through the heating or cooling module, ensuring that the temperature remains stable throughout the entire test process.

[0063] In step S2, an automatic recognition algorithm based on a convolutional neural network is developed to extract features from a standard grid image, including grid intersections, and automatically generate a coordinate system transformation matrix suitable for scanning electron microscopes and atomic force microscopes.

[0064] The automatic recognition algorithm of convolutional neural networks can identify and mark key feature points in grid images, and then construct a high-precision coordinate transformation matrix. Through this matrix, images of rock thin sections taken under different instruments are accurately aligned to ensure the accuracy of positioning. The automatic recognition algorithm of convolutional neural networks is also highly efficient, capable of quickly processing large amounts of image data and improving the overall detection efficiency.

[0065] By using standard samples with known three-dimensional structures, including nanowire arrays, data were acquired in various instruments, and calibration accuracy was assessed through three-dimensional reconstruction error to ensure that the cross-instrument positioning error was less than 50 nm.

[0066] The positioning error formula is as follows:

[0067]

[0068] Where x1, y1, and z1 are the coordinates of the standard sample in the scanning electron microscope, and x2, y2, and z2 are the coordinates of the standard sample in the atomic force microscope.

[0069] By accurately measuring and recording the coordinate positions of standard samples in different instruments, a unified coordinate system is established. This system can accurately reflect the relative positions of rock thin sections under different instruments. The establishment of the coordinate system is a key step in achieving precise positioning across instruments.

[0070] In practice, multiple standard samples with known three-dimensional structures, such as nanowire arrays, are selected, and data is acquired using different in-situ testing instruments.

[0071] The coordinates of each standard sample in the first instrument's scanning electron microscope (X1, Y1, Z1) and in the second instrument's atomic force microscope (X2, Y2, Z2) will be precisely measured and recorded.

[0072] Using coordinate data, a coordinate transformation matrix is ​​calculated. This matrix can accurately transform the image of the rock thin section in the first instrument to the coordinate system of the second instrument, thereby achieving high-precision positioning.

[0073] In step S3, the parameters for joint use of multiple in-situ testing devices are configured, sample calibration is performed, images are acquired in each device using standard grid samples, and a coordinate system transformation matrix is ​​constructed using a feature point matching algorithm.

[0074] In step S3, a preliminary microscopic image of a rock thin section is input, and the distribution of the region of interest is predicted through a U-Net network. The optimal detection sequence of performing Raman spectroscopy analysis followed by focused ion beam cutting is automatically generated.

[0075] Among them, Raman spectroscopy is non-destructive, while focused ion beam cutting is destructive;

[0076] A hardware triggering mechanism is established between the devices, and the EEE1588 precision clock protocol is adopted to ensure that the secondary electron images of SEM and the elemental distribution data of micro-area X-ray fluorescence are strictly aligned in time.

[0077] By integrating a temperature control module (temperature fluctuation ≤ ±5℃) with a smart sample stage and submicron-level displacement control with a micro-electromagnet (error < 50nm), and combining it with U-Net network prediction of regions of interest, the system automatically generates the optimal path from non-destructive analysis to destructive analysis, reducing redundant scanning by up to 40%, improving overall detection efficiency by 3 times, and reducing sample loss rate from >30% in traditional methods to <5%. At the same time, the standardized process shortens the single-sample analysis time to 8 hours. This method is highly automated and intelligent, significantly reducing the workload of researchers and improving research efficiency. It constructs a "four-dimensional data cube" (spatial XY-Z + time t) through a real-time Python data interface, integrating multi-source data such as SEM morphology, Raman spectroscopy composition, and X-ray elemental concentration, and tracking mineral changes over time. Combining optical flow and pyramid layering algorithms, it not only achieves cross-scale fusion of mineral distribution (resolution ≤2nm) and macroscopic structure, but also reconstructs mineral growth rates (accuracy ≤0.1μm / h) and alteration paths, forming a three-dimensional digital twin model of "microstructure-composition-time." This is the first time that dynamic quantitative analysis of rock microstructure evolution has been achieved, providing a scientific basis for the stability analysis and design of underground engineering projects.

[0078] Example 3, please refer to Figure 1 As shown, in step S4, a real-time data interface based on Python is developed to synchronously map the BSE image of SEM, the component distribution of Raman spectrum and the elemental concentration data of X-ray fluorescence to a unified coordinate system, forming a dynamically superimposed four-dimensional data cube. The dynamically superimposed four-dimensional data cube includes spatial X, Y, Z dimensions and time t dimension.

[0079] In step S4, the detection data of each device is collected through the data collection device, and outlier data points are automatically detected using the isolated forest algorithm and marked in the three-dimensional model.

[0080] In step S5, for areas with complex mineral distribution, high-resolution transmission electron microscopy data is used, and scanning electron microscopy data is used on the visualization area. The data is then fused using a pyramid layering algorithm to construct a 3D model that includes both details and a global perspective.

[0081] The formula for the pyramid layering algorithm is as follows:

[0082]

[0083] Where M represents the fusion result obtained after processing, Pyramid represents the pyramid algorithm, TEM_data represents transmission electron microscopy, and SEM_data represents scanning electron microscopy. This represents a fusion operation.

[0084] During the fusion process, high-resolution rock thin-section images are acquired using transmission electron microscopy, which provides detailed structural information about the rock thin sections. Scanning electron microscopy is then used to acquire more macroscopic image data in the visualization area, which helps to construct a global view of the rock thin sections. The two different scales of image data are then fused using a pyramid layering algorithm to obtain a fused result that includes both details and a global perspective.

[0085] The pyramid layering algorithm first preprocesses the image data from transmission electron microscopy and scanning electron microscopy, including denoising and contrast enhancement operations to improve image quality. Based on the principle of the pyramid layering algorithm, the image data is decomposed layer by layer to form an image pyramid of multiple scales. At each scale, the image data from transmission electron microscopy and scanning electron microscopy are fused through a fusion operation to obtain the fusion result at that scale.

[0086] In this way, the advantages of both transmission electron microscopy and scanning electron microscopy can be fully utilized to construct a three-dimensional model that has both high-resolution details and a global perspective, providing more accurate and comprehensive information for the localization and detection of rock thin sections.

[0087] In step S5, optical flow analysis is introduced to analyze two-dimensional image sequences at different time points, reconstructing the dynamic process of mineral growth or alteration, forming a "four-dimensional digital twin model," namely, 3D spatial dimensions X, Y, z + time dimension t.

[0088] The vector calculation formula for the optical flow method is as follows:

[0089] I X u+I Y v+I t =0;

[0090] Among them, I X I YI t Let u be the image gradient, and v be the pixel displacement vectors.

[0091] Optical flow analysis, by analyzing a sequence of two-dimensional images of rock thin sections taken at different time points, can capture minute changes in mineral growth or alteration. By calculating the optical flow vector of each pixel—the displacement of the pixel between consecutive images—it is possible to track changes in mineral structure over time. The image gradient part of the optical flow vector calculation formula reflects the changes in grayscale values ​​in the image, while the pixel displacement vector represents the direction and distance of pixel movement from the previous frame to the current frame. Using optical flow information, the dynamic process of mineral growth or alteration can be reconstructed. Combining this with three-dimensional spatial information forms a "four-dimensional digital twin model" that includes the time dimension. The model not only provides three-dimensional structural information of the rock thin section but also demonstrates the dynamic process of mineral changes over time, providing a more comprehensive and in-depth perspective for the analysis and research of rock thin sections.

[0092] In the field of oil and gas reservoirs, it can accurately identify nanoscale pore structure features, such as in shale gas reservoirs, providing key data for seepage simulation. In mineral exploration, it can reveal isotopic evidence of diagenetic stages and fluid evolution through cross-instrument combinations, such as the linkage between LA-ICP-MS and cathodoluminescence. In deep engineering, it can dynamically analyze rock crack propagation dynamics and predict mechanical stability using four-dimensional models, providing forward-looking support for underground engineering design. This method breaks through the limitations of single instruments, constructs an integrated "location-analysis-modeling" process, and promotes the development of rock detection towards intelligence and dynamism.

[0093] This invention employs electron beam-induced deposition of platinum and carbon nanoparticles (accuracy <100nm), which are fixed to the edge of thin sections using atomic-level bonding technology. Compared to traditional micron-level markers (error >1μm), this method improves positioning accuracy by more than 10 times and can precisely match the nanoscale field of view of TEM and SEM instruments. Combined with automatic feature extraction technology based on CNN algorithm, coordinate system calibration is completed within five minutes, with cross-instrument positioning error <50nm, significantly improving the accuracy of micro-area detection. This method is a creative combination of materials science, computer vision, and geological detection, achieving precise positioning and efficient detection of rock thin sections, providing more comprehensive data support for rock thin section research. By integrating a temperature control module (temperature fluctuation ≤±5℃) into the intelligent sample stage and submicron-level displacement control of micro-electromagnets (error <50nm), combined with U-Net network prediction of regions of interest, the optimal path of "non-destructive analysis → destructive analysis" is automatically generated, reducing redundant scanning by up to 40%, improving overall detection efficiency by 3 times, and reducing sample loss rate from >30% in traditional methods to <5%. At the same time, the standardized process shortens the single-sample analysis time to 8 hours. This method features a high degree of automation and intelligence, significantly reducing the workload of researchers and improving research efficiency. It constructs a "four-dimensional data cube" (spatial XY-Z + time t) through a real-time Python data interface, integrating multi-source data such as SEM morphology, Raman spectroscopy composition, and X-ray elemental concentration, and tracks mineral changes over time. Combining optical flow and pyramid layering algorithms, it not only achieves cross-scale fusion of mineral distribution (resolution ≤2nm) and macroscopic structure, but also reconstructs mineral growth rates (accuracy ≤0.1μm / h) and alteration paths, forming a three-dimensional digital twin model of "microstructure-composition-time," achieving for the first time a microstructure-composition-time model of rocks. Dynamic quantitative analysis of structural evolution provides a scientific basis for the stability analysis and design of underground engineering. In the field of oil and gas reservoirs, it can accurately identify nanoscale pore structure characteristics, such as in shale gas reservoirs, providing key data for seepage simulation. In mineral exploration, it reveals isotopic evidence of diagenetic stages and fluid evolution through cross-instrument combinations, such as the linkage between LA-ICP-MS and cathodoluminescence. In deep engineering, it dynamically analyzes rock crack propagation dynamics and combines four-dimensional models to predict mechanical stability, providing forward-looking support for underground engineering design. This method breaks through the limitations of single instruments, constructs an integrated "location-analysis-modeling" process, and promotes the development of rock detection towards intelligence and dynamism.

[0094] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; 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 some or all of the technical features can be replaced; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for locating and detecting thin sections of rock using multiple in-situ testing instruments in combination, characterized in that: Includes the following steps: S1: Sample preparation. The rock sample to be tested is sliced ​​and then transversely cut into two parts to form cross-sections I and II. The two sections are completely corresponding in position and size, and the parallelism error between the two sections is ≤50μm and the area difference is ≤1%. S2: Instrument calibration, design of a universal sample stage, and compatibility with SEM, AFM, Raman spectrometers and micro-area X-ray fluorescence equipment to calibrate multiple in-situ testing instruments; S3: Combined setup, based on the characteristics of rock thin sections and detection requirements, covers test location, depth and time factors to achieve joint detection; S4: Data recording and analysis, recording the test data of each device and performing comprehensive analysis; S5: Model Building: Based on the detection data recorded in step S4, construct a three-dimensional digital model of the rock thin section.

2. The method for locating and detecting thin sections of rock using multiple in-situ testing instruments in conjunction as described in claim 1, characterized in that: In step S1, the surface of the sheet is subjected to chemical mechanical polishing, and epoxy resin is coated and cured at the edge of the sheet. In step S1, ion beam deposition is used to deposit platinum or carbon nanoparticles on the surface of rock thin sections, enabling atomic-level bonding between the sample surface atoms and the deposited material.

3. The method for locating and detecting thin sections of rock using multiple in-situ testing instruments in conjunction as described in claim 1, characterized in that: In step S2, a heating and cooling module is integrated to be compatible with temperature-sensitive analysis of SEM and Raman spectrometers. Furthermore, a micro electromagnet is embedded at the bottom of the sample stage to control the submicron displacement of the sample stage among multiple instruments via an external magnetic field. The temperature control calculation formula is as follows: Temperature fluctuation range = (actual T - set T) ≤ ±5℃; Wherein, T_actual_measured represents the temperature value actually measured by the temperature sensor device, and T_set refers to the target temperature value that has been set in advance.

4. The method for locating and detecting thin sections of rock using multiple in-situ testing instruments in conjunction as described in claim 1, characterized in that: In step S2, an automatic recognition algorithm based on a convolutional neural network is developed to extract features from a standard grid image, including grid intersections, and automatically generate a coordinate system transformation matrix suitable for scanning electron microscopes and atomic force microscopes. By using standard samples with known three-dimensional structures, including nanowire arrays, data were acquired in various instruments, and calibration accuracy was assessed through three-dimensional reconstruction error to ensure that the cross-instrument positioning error was less than 50 nm. The positioning error formula is as follows: Where x1, y1, and z1 are the coordinates of the standard sample in the scanning electron microscope, and x2, y2, and z2 are the coordinates of the standard sample in the atomic force microscope.

5. The method for locating and detecting thin sections of rock using multiple in-situ testing instruments in conjunction as described in claim 1, characterized in that: In step S3, the parameters for joint use of multiple in-situ testing devices are configured, sample calibration is performed, images are acquired in each device using standard grid samples, and a coordinate system transformation matrix is ​​constructed using a feature point matching algorithm.

6. The method for locating and detecting thin sections of rock using multiple in-situ testing instruments in conjunction as described in claim 1, characterized in that: In step S3, a preliminary microscopic image of a rock thin section is input, and the distribution of the region of interest is predicted through a U-Net network to automatically generate the optimal detection sequence of performing Raman spectroscopy analysis first and then focused ion beam cutting. Among them, Raman spectroscopy is non-destructive, while focused ion beam cutting is destructive; A hardware triggering mechanism is established between devices, and the EEE1588 precision clock protocol is adopted.

7. The method for locating and detecting thin sections of rock using multiple in-situ testing instruments in conjunction as described in claim 1, characterized in that: In step S4, a real-time data interface based on Python is developed to synchronously map the BSE image of SEM, the component distribution of Raman spectrum, and the elemental concentration data of X-ray fluorescence to a unified coordinate system, forming a dynamically superimposed four-dimensional data cube. The dynamically superimposed four-dimensional data cube includes spatial X, Y, and Z dimensions and a time t dimension.

8. The method for locating and detecting thin sections of rock using multiple in-situ testing instruments in conjunction as described in claim 1, characterized in that: In step S4, the detection data of each device is collected through the data collection device, and outlier data points are automatically detected using the isolated forest algorithm and marked in the three-dimensional model.

9. The method for locating and detecting thin sections of rock using multiple in-situ testing instruments in conjunction as described in claim 1, characterized in that: In step S5, for areas with complex mineral distribution, high-resolution transmission electron microscopy data is used, and scanning electron microscopy data is used on the visualization area. The data is then fused using a pyramid layering algorithm to construct a 3D model that includes both details and a global perspective. The formula for the pyramid layering algorithm is as follows: Where M represents the fusion result obtained after processing, Pyramid represents the pyramid algorithm, TEM_data represents transmission electron microscopy, and SEM_data represents scanning electron microscopy. This represents a fusion operation.

10. The method for locating and detecting thin sections of rock using multiple in-situ testing instruments in conjunction as described in claim 1, characterized in that: In step S5, optical flow analysis is introduced to analyze two-dimensional image sequences at different time points, reconstruct the dynamic process of mineral growth or alteration, and form a "four-dimensional digital twin model", namely, 3D space X, Y, z + time t dimension. The vector calculation formula for the optical flow method is as follows: I X u+I Y v+I t =0; Among them, I X I Y I t Let u be the image gradient, and v be the pixel displacement vectors.

Images