Coal mine in-service wellbore disease precision quantitative identification method based on multi-source heterogeneous driving

Abstract

This invention discloses a method for accurate quantitative identification of defects in in-service coal mine shafts based on multi-source heterogeneous driving, comprising: S1. Installing monitoring equipment on the annular structure of the coal mine vertical shaft to collect visual data, ultrasonic data, and stress data; S2. Preprocessing the collected data; S3. Calculating the information entropy and fusion weights of the three types of data after preprocessing based on the entropy weight method to complete feature layer fusion; S4. Establishing quantitative formulas for crack width, corrosion area, and spalling depth, and combining the fused feature layers to output the quantitative indicators and severity levels of defects through the final weights. This invention can achieve comprehensive perception of defects on the surface, inside, and under structural stress, with high quantitative accuracy and strong linkage to maintenance decisions. It is suitable for safety monitoring of in-service coal mine vertical shafts and can significantly reduce operation and maintenance costs and accident rates.

Description

Technical Field

[0001] This invention relates to the field of coal mine vertical shaft safety technology, specifically to a method for accurate quantitative identification of defects in in-service coal mine shafts based on multi-source heterogeneous driving. Background Technology

[0002] Coal mine shafts, serving as the core passageways for coal mining, are subject to long-term effects from shaft pressure, groundwater erosion, and mine dust corrosion, making them prone to defects such as cracks, shaft wall corrosion, and concrete spalling. Existing shaft defect identification technologies suffer from two major shortcomings:

[0003] (1) Blind spots in single data identification: Relying on visual images can only identify surface defects (such as crack direction) and cannot detect internal corrosion or deep cracks; Although ultrasonic detection can obtain internal structural information, it is difficult to quantify the area of ​​defects; Stress sensors can only report structural stress anomalies and cannot be associated with specific defect types, resulting in incomplete defect identification.

[0004] (2) Qualitative identification cannot support maintenance decisions: Existing methods mostly rely on qualitative judgments of "existence / non-existence of defects" and cannot output quantitative indicators such as crack width, corrosion area, and peeling depth. Maintenance plan formulation depends on human experience, which can easily lead to over-maintenance or under-maintenance, increasing safety risks and operation and maintenance costs.

[0005] Therefore, this application proposes a method for accurate quantitative identification of defects in in-service coal mine shafts based on multi-source heterogeneous driving, which can integrate multi-source monitoring data and achieve accurate quantification of defects to solve the above-mentioned technical problems. Summary of the Invention

[0006] The main objective of this invention is to provide a method for accurate quantitative identification of defects in in-service coal mine shafts based on multi-source heterogeneous driving, in order to solve the technical problems mentioned in the background art, such as the blind spots of single data identification and the inability of qualitative identification to support maintenance decisions.

[0007] The present invention solves the above-mentioned technical problems by adopting the following technical solutions:

[0008] A method for accurate quantitative identification of defects in in-service coal mine shafts based on multi-source heterogeneous driving, which includes the following steps executed by computer equipment:

[0009] S1. To address the needs of monitoring the annular structure of a coal mine vertical shaft, visual monitoring equipment (explosion-proof high-definition industrial camera and laser scanner), ultrasonic detection equipment (multi-frequency ultrasonic probe array), and stress monitoring equipment (fiber optic stress sensor) are deployed and installed on the annular structure of the coal mine vertical shaft to collect visual data, ultrasonic data, and stress data, ultimately achieving multi-dimensional data acquisition of the shaft.

[0010] S2. Perform noise reduction, correction, and smoothing preprocessing on the collected data respectively;

[0011] S3. Calculate the information entropy and fusion weight of the three types of data after preprocessing based on the entropy weight method to complete the feature layer fusion;

[0012] S4. Establish quantitative formulas for crack width, corrosion area, and spalling depth. Combine these with the fused feature layers and output the quantitative indicators and severity levels of the disease through final weighting.

[0013] Preferably, in step S1, the inner wall of the annular structure of the coal mine shaft is pre-installed with at least three vertical explosion-proof guide rails, which are evenly arranged in a ring at specified intervals along the inner wall of the shaft (conventionally, they are evenly arranged in a ring at 120° intervals along the inner wall of the shaft to achieve full coverage without blind spots, while ensuring accurate matching of visual data, laser three-dimensional coordinates and ultrasonic detection, with reference positions of 0°, 120° and 240° with the center of the shaft as the central angle).

[0014] The explosion-proof guide rail is equipped with an independently movable explosion-proof mobile carrier, which integrates explosion-proof mobile carrier one and explosion-proof mobile carrier two to form a dual independent bearing unit (vertical spacing ≥10m), which sequentially carries visual monitoring equipment and ultrasonic detection equipment. The explosion-proof mobile carrier is equipped with a depth encoder and a locking mechanism, with a positioning error ≤±5mm and vibration resistance ≥5g.

[0015] During the data acquisition process, a monitoring section is set up every 5m to 10m along the inner wall of the well shaft for stopping and collecting data. At this time, the monitoring sections of the visual monitoring, ultrasonic probe, and stress sensor are as completely overlapping as possible to ensure that the multi-source data can correspond to the same well wall area. Each monitoring section is a set of circular planes perpendicular to the well shaft axis.

[0016] The monitoring section focuses on the well neck (0-50m), well bottom, and sections where geological conditions change (such as rock strata interfaces, and the vicinity of the interface between loose layers and bedrock layers).

[0017] Preferably, the visual monitoring equipment uses an explosion-proof high-definition industrial camera and a laser scanner to monitor the direction of cracks and the boundary of corrosion areas on the inner surface of the well wall, wherein:

[0018] A combination of "multi-camera coverage + single scanner calibration" can be used, with the camera as the main component and the scanner as a supplement, to avoid functional redundancy.

[0019] Explosion-proof high-definition industrial cameras acquire images of the well wall surface, capturing the direction of cracks and the boundaries of corroded areas (relying on 5-megapixel resolution to achieve detailed identification).

[0020] The laser scanner outputs the three-dimensional coordinates (X / Y / Z) of each point on the well wall, calibrating the camera's "pixel accuracy" to avoid size calculation errors caused by image distortion;

[0021] The explosion-proof high-definition industrial camera has a resolution of ≥5 million pixels;

[0022] The laser ranging accuracy of the laser scanner is ±0.1mm;

[0023] The visual monitoring device is installed on the explosion-proof mobile carrier 1 on the vertical explosion-proof guide rail, and the axial movement monitoring is achieved through the explosion-proof guide rail;

[0024] Data is uploaded to the ground server in real time via an intrinsically safe wireless transmission module for mining applications;

[0025] The explosion-proof high-definition industrial camera is equipped with a rotatable angle adjustment bracket. When the visual monitoring device moves to a certain monitoring section, the angle adjustment bracket drives the camera / scanner to adjust the lens angle.

[0026] The explosion-proof high-definition industrial camera rotates via the angle adjustment bracket, allowing the lens to face the opposite well wall. The "field of view" covered by each explosion-proof high-definition industrial camera must include the edges of the fields of view of two adjacent cameras (to ensure that the fields of view of two adjacent cameras have an overlap of ≥10%, completely eliminating the monitoring blind spot in the circumferential direction). With the overlap of the fields of view of three explosion-proof high-definition industrial cameras, the entire circumference of the well is covered 360°. Depending on the compatibility between the "field of view" of the explosion-proof high-definition industrial camera and the diameter of the well, it can be increased to 4 cameras (90° interval) or 6 cameras (60° interval).

[0027] The laser scanner has a built-in 360° circular scanning function (no rotation required), with a scanning accuracy of ±0.1mm. It can obtain the three-dimensional coordinates of the entire monitoring section in a single scan, directly replacing the need for circumferential movement.

[0028] The laser scanner, with the aid of components such as a horizontal adjustment arm and an angle fine-tuning mechanism, can be adjusted to the target position "directly below the center of the monitoring section" or "coaxial with the explosion-proof high-definition industrial camera," and then locked in place. When "directly below the center," the laser emission direction is perpendicular to the well wall surface. When "coaxial with the explosion-proof high-definition industrial camera," it can be directly and synchronously positioned with the camera, which is more efficient and ensures precise matching between the "laser three-dimensional coordinates" and the "camera image pixels" (coaxial means that the image and coordinates of the same area come from the same optical axis), providing a more accurate data foundation for subsequent "crack width quantification calculation."

[0029] Preferably, in step S1, a multi-frequency ultrasonic probe array of an ultrasonic detection device is used during ultrasonic data acquisition to monitor the corrosion thickness and deep fracture depth inside the well wall. The multi-frequency ultrasonic probe array includes six multi-frequency ultrasonic probes with a detection frequency of 2.5–10 MHz. Different frequencies are used to adapt to different defects. Each probe can switch between three frequencies: 2.5 MHz, 5 MHz, and 10 MHz. The specific functions are as follows:

[0030] 1) 2.5MHz low frequency: Deep detection depth, suitable for locating "deep cracks inside the well wall" and "areas with steel reinforcement corrosion";

[0031] 2) 5MHz intermediate frequency: high detection accuracy (resolution ≤ 0.1mm), suitable for quantification of "internal corrosion thickness";

[0032] 3) 10MHz high frequency: Focuses on the shallow layer below the surface, adapted to the detection of "surface crack extension depth" (forming "depth-width" linked data with visually recognized surface cracks).

[0033] The detection radius of the multi-frequency ultrasonic probe is ≥0.5m, the penetration depth is ≤500mm, the detection range of the six multi-frequency ultrasonic probes completely covers the circumference of the inner wall of the well, and the overlap rate of the detection areas of adjacent multi-frequency ultrasonic probes is ≥10%.

[0034] The ultrasonic detection device includes multiple multi-frequency ultrasonic probes, which are installed on the explosion-proof mobile carrier of the vertical explosion-proof guide rail. Two multi-frequency ultrasonic probes are installed on each explosion-proof mobile carrier. The ultrasonic detection device for ultrasonic data acquisition and the visual monitoring device for visual data acquisition share the vertical explosion-proof guide rail to achieve axial movement monitoring.

[0035] Each multi-frequency ultrasonic probe has an integrated solid coupling pad on its detection end face. The coupling pad is a butyl rubber modified composite material with a thickness of 5 mm and an acoustic impedance of approximately 2.8 × 10⁻⁶ mm. 6 kg / (m 2 •s), with an anti-slip texture on the surface and chamfered edges;

[0036] The explosion-proof mobile carrier integrates a retractable actuator, including an adjustable ring bracket, an explosion-proof electric push rod, a parallel guide rail, and a pressure sensor. During ultrasonic detection, the explosion-proof mobile carrier moves along the vertical explosion-proof guide rail to the target monitoring section, the adjustable ring bracket rotates to a preset angle, and the probe position is 30° away from the reference position of the vertical explosion-proof guide rail, so that the 6 probes are pushed to circumferential positions of 30°, 90°, 150°, 210°, 270°, and 330° respectively, avoiding obstruction by the guide rail.

[0037] The explosion-proof electric push rod pushes the probe to make vertical contact with the well wall. After the data collection is completed, the probe is retracted and the data is uploaded to the ground server in real time through the intrinsically safe wireless transmission module for mining.

[0038] The ultrasonic detection equipment and visual monitoring equipment operate in the order of "visual detection first, ultrasonic detection later", covering all monitoring sections in sequence. The field of view overlap rate of the detection area of ​​the same monitoring section is ≥10%. The data is automatically aligned according to the section depth reference, and the alignment error is ≤±5mm, which does not affect the subsequent multi-source fusion.

[0039] Preferably, the stress monitoring device for stress data acquisition in step S1 is used to monitor sudden changes in wellbore stress (associated with spalling risk), and adopts a fiber optic stress sensor with a range of 0 to 200 MPa and an accuracy of ±0.5%FS.

[0040] The stress monitoring equipment is preferentially deployed in the well neck (which is greatly affected by surface load), the bottom of the well (which is greatly affected by mining), and sections where geological conditions change (such as near rock strata interfaces, loose layers and bedrock interfaces), and is on the same cross section as the visual and ultrasonic equipment to facilitate multi-source data alignment.

[0041] The fiber optic stress sensors are arranged in 3 to 6 groups along the inner wall of the wellbore in each monitoring section, adopting the principle of "equal interval + key reinforcement" - in the normal area, they are arranged at 120° intervals (corresponding to adjacent positions of 0°, 120° and 240° on the guide rail). If a slight crack has appeared in a certain area (predicted by the visual equipment), 3 additional sensors (60° interval) are added to cover a 30cm range on both sides of the crack.

[0042] The sensitive axis of the fiber optic stress sensor is matched with the stress monitoring target. When monitoring axial stress, the sensor is attached along the vertical (up and down) direction of the wellbore to capture the vertical compressive / tensile stress of the well wall (associated with spalling defects). When monitoring circumferential stress, the sensor is attached along the circumference of the wellbore to capture radial pressure changes (associated with crack propagation).

[0043] The fiber optic stress sensor is installed using a post-mount method, following a three-step process: surface pretreatment → bonding and fixing with explosion-proof epoxy adhesive for coal mines → protective encapsulation with explosion-proof stainless steel housing.

[0044] In conjunction with the design of the explosion-proof guide rail movement monitoring, the optical fiber is laid along the dedicated cable tray on the side of the explosion-proof guide rail inside the well shaft (the cable tray depth is ≥2cm and the width is ≥1.5cm), and is fixed with explosion-proof clamps every 1m (the clamp spacing is ≤30cm to avoid the optical fiber shaking); a "buffer bend" (bending radius ≥30cm to avoid tensile breakage) is set every 50m for the vertical optical fiber, and a 1m redundancy length is reserved (to accommodate slight deformation of the well shaft);

[0045] The underground optical fiber is collected through the explosion-proof junction box (intrinsically safe type) at the top of the shaft, and then transmitted to the ground monitoring room through the mine-specific optical fiber trunk line. The ground end is connected to the fiber optic grating demodulator, which converts the optical signal into a stress electrical signal. Finally, it is uploaded to the server through the mine-use intrinsically safe wireless transmission module and stored synchronously with the visual and ultrasonic data.

[0046] To achieve "stress-vision-ultrasound" data fusion, the sensor needs to form "three-source positioning" with the vision camera and ultrasound probe of the same cross section - the straight-line distance between each sensor and the adjacent vision camera and ultrasound probe is ≤0.5m, to ensure that the monitoring area completely overlaps and avoid data association blind spots.

[0047] Preferably, the preprocessing operation in step S2 includes:

[0048] S21. An adaptive median filtering algorithm is used to eliminate salt-and-pepper noise caused by mineral dust. The algorithm formula is as follows:

[0049] ;

[0050] Where f(x,y) is the pixel value of the original visual image; g(x,y) is the pixel value after denoising; med{·} is the median algorithm, used to eliminate isolated noise points while preserving crack edge features;

[0051] S22. The influence of wellbore concrete temperature (25-60℃) on ultrasonic propagation speed is corrected by the sound velocity compensation formula. At this time, the ultrasonic probe array is equipped with a self-calibration module, which integrates a 100mm thick concrete standard test block to support on-site calibration in the well. After calibration, the ultrasonic data detection error is ≤2%, which is compatible with the ultrasonic sound velocity compensation formula. The sound velocity compensation formula is as follows:

[0052] ;

[0053] in, Expressed as the velocity of sound in concrete at 25°C; The measured wellbore temperature (°C) is used. The corrected sound velocity is used to ensure the accuracy of ultrasonic depth detection;

[0054] S23. Finally, a central moving average filter is used to eliminate mechanical vibration interference. The filtering formula is as follows:

[0055] ;

[0056] in, This is the original stress value; This indicates the size of the sliding window (take an odd number between 3 and 7 to ensure that the sliding window is centered on the target data point and avoid data offset). This is the smoothed stress value, used to preserve stress abrupt change characteristics (corresponding to spalling defects); The time series index of the target smoothed data points (corresponding to the first) The original data points acquired during secondary stress sampling. =1,2,3...); Indexes of the original data points (range of values) that participate in the averaging within the sliding window. to ,like hour, , , ; hour, , , , , , , ).

[0057] Preferably, step S3 employs an entropy-weighted fusion model, allocating fusion weights based on the information entropy (credibility) of different data sources, specifically including:

[0058] S31. Extract the disease characteristic indicators based on visual features, ultrasonic features, and stress features respectively;

[0059] S32. Calculate the information entropy of each data source. The calculation formula is as follows:

[0060] ;

[0061] ;

[0062] in, These represent vision, ultrasound, and stress, respectively. To monitor the number of cross sections; For the first The first class feature The original feature value (i.e. the first) Class of monitoring indicators in the first (Original data collected at each monitoring section) For the first The first class of data The eigenvalue normalization result of the cross section (i.e., for) After normalization, the percentage of the data in all cross-sectional data of the same feature is obtained. The smaller the information entropy, the lower the data dispersion and the higher the credibility.

[0063] S33. Calculate the fusion weights The calculation formula is as follows:

[0064] ;

[0065] and ;

[0066] At this point, the credibility of data can be quantified by information entropy, and adaptive fusion of "high credibility data with high weight" can be achieved, which can solve the problem of multi-source data conflict in existing technologies (such as internal corrosion that is not visually recognized but detected by ultrasound, which is included in the recognition result through weight allocation).

[0067] Preferably, the disease characteristic indicators in step S31 specifically include:

[0068] Visual features: Crack pixel length Pixel area of ​​eroded region ;

[0069] Ultrasonic characteristics: internal corrosion thickness Depth of deep cracks ;

[0070] Stress characteristics: The difference between the measured value and the reference value is taken as the stress mutation value. σs.

[0071] Preferably, the crack width quantification formula in step S4 is used to calculate the actual crack width by combining the pixel precision of the visual image (P=0.05mm / pixel) with the three-dimensional coordinates of the laser scan. The specific formula is as follows:

[0072] ;

[0073] in, Weights for visual feature fusion; The length of the crack is in pixels; P For visual camera pixel precision; Represents visual width weights (data from step S3). Adjustment); Weights for ultrasound feature fusion; This refers to the ultrasonic echo time difference (the echo time difference between the upper and lower interfaces of the crack). The width of the crack detected by ultrasonic testing; Weighted by the ultrasonic width;

[0074] At this point, accuracy verification shows that the crack width error calculated by this formula is ≤0.03mm, which is better than single visual recognition (error ≤0.1mm).

[0075] Preferably, the corrosion area quantification formula in step S4 is used to fuse the surface area of ​​visual segmentation with the internal corrosion area detected by ultrasonic testing to calculate the total corrosion area. This solves the problem that a single visual approach cannot identify internal corrosion, achieving total area quantification of "surface + interior". The specific formula is as follows:

[0076] ;

[0077] in, This represents the number of pixels in the visually eroded area. The pixel area; The number of internal corrosion areas detected by ultrasonic testing (each detection point covers a circular area with a radius of r = 0.1 m). The area of ​​a single ultrasonic detection region.

[0078] Preferably, the spalling depth quantification formula in step S4 is used to correlate stress mutation values ​​with ultrasonic depth detection to calculate the wellbore spalling depth, enabling dual verification of depth through "structural response + direct detection," with an error ≤5mm. The specific formula is as follows:

[0079] ;

[0080] in, (Expressed as the elastic modulus of concrete). The strain corresponding to the sudden change in stress; The thickness of the wellbore wall is used to estimate the depth of structural deformation caused by spalling through stress back calculation. The actual depth of the exfoliation detected by ultrasonic testing; , All are weights. The weight of stress data in the peel depth quantization corresponds to the fusion weight of stress features in the feature layer fusion step. (That is, the credibility weight of the stress data source calculated by the entropy weight method). The weights of ultrasound data in the ablation depth quantization correspond to the fusion weights of ultrasound features in the feature layer fusion step. (That is, the credibility weight of the ultrasound data source calculated by the entropy weight method).

[0081] Preferably, in step S4, the quantitative indicators and severity levels are finally weighted and output using the quantitative indicators of crack width W, corrosion area S, and spalling depth H to establish a wellbore damage severity level matrix, so as to achieve a direct mapping between "quantitative indicators and maintenance priorities".

[0082] The severity levels of wellbore defects are categorized into four levels: minor (Level I), moderate (Level II), severe (Level III), and critical (Level IV). The quantitative indicators corresponding to each severity level of wellbore defects are as follows:

[0083] Slight (Level I): W<0.2mm, S<1.0m 2 H<10mm;

[0084] Moderate (Level II): 0.2mm ≤ W < 0.5mm, 1.0m 2 ≤S<3.0m 2 10mm≤H<30mm;

[0085] Severe (Level III): 0.5mm ≤ W < 1.0mm, 3.0m 2 ≤S<5.0m 2 30mm≤H<50mm;

[0086] Emergency (Level IV): W≥1.0mm, S≥5.0m 2 H≥50mm.

[0087] In another aspect, the present invention also discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the method described above.

[0088] In another aspect, the present invention also discloses a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the method described above.

[0089] As can be seen from the above technical solution, the present invention provides a method for accurate quantitative identification of defects in in-service coal mine shafts based on multi-source heterogeneous driving. Compared with the prior art, the present invention has the following advantages:

[0090] 1. This invention solves the problem of complex modification of traditional multi-equipment and multi-rail systems by combining three reusable guide rails with a dynamic telescopic device, thereby reducing the amount of well wall modification required.

[0091] 2. This invention, through the synergistic fusion of multi-source monitoring data (visual, ultrasonic, stress), can eliminate the blind spots of single data identification and achieve comprehensive perception of wellbore defects on the surface, inside and under structural stress.

[0092] 3. This invention can automatically adapt to different monitoring environments (such as increasing the ultrasonic weight when the concentration of mineral dust is high) by allocating the credibility weight of multi-source data, thereby improving the accuracy of multi-source data fusion and avoiding interference from poor-quality data on the identification results.

[0093] 4. This invention establishes a mapping model between multi-source data and quantitative indicators of defects (width, area, depth, severity level), and achieves real-time synchronization of multi-source data fusion results. It automatically classifies defects into severity levels I-IV and outputs quantitative results that can directly support maintenance decisions. By directly matching maintenance plans with quantitative indicators such as crack width, corrosion area, and spalling depth (e.g., a crack width of 0.3mm corresponds to a grouting pressure of 1.2MPa), maintenance decisions are transformed from "experience-driven" to "data-driven," reducing operation and maintenance costs by 30%, wellbore accident rate by 40%, and significantly improving the efficiency of safety management and control of in-service wellbores.

[0094] It should be understood that the descriptions in this section are not intended to identify key or essential features of embodiments of the invention, nor are they intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Of course, implementing any product of the invention does not necessarily require achieving all of the advantages described above simultaneously. Attached Figure Description

[0095] The accompanying drawings, which form part of this application, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings:

[0096] Figure 1 This is a schematic diagram of the overall method flow of the present invention;

[0097] Figure 2 This is a schematic diagram of the installation structure of the visual monitoring device of the present invention;

[0098] Figure 3 This is a schematic diagram of the installation structure of the ultrasonic detection device of the present invention;

[0099] Figure 4 This is a schematic diagram of the installation structure of the stress monitoring device of the present invention. Detailed Implementation

[0100] 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 a part of the embodiments of the present invention, and not all of them. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. 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.

[0101] For details in the embodiments, please refer to Figures 1 to 4 .

[0102] In the current technology for identifying defects in operating vertical shafts of coal mines, there are problems such as blind spots in identification based on single monitoring data, inability of qualitative judgment to support maintenance decisions, and lack of data credibility weight allocation. These problems lead to incomplete defect identification, reliance on manual experience in maintenance plan formulation, and interference of poor-quality data with identification results.

[0103] This invention proposes a method for precise quantitative identification of defects in in-service coal mine shafts based on multi-source heterogeneous driving. Through a four-layer technical system of "multi-source equipment deployment → data preprocessing → feature layer fusion → precise quantitative identification," and based on the fusion of visual, ultrasonic, and stress multi-source monitoring data, it achieves multi-dimensional and high-precision quantitative analysis of defects such as cracks, corrosion, and wall spalling in in-service coal mine vertical shafts. This provides crucial evidence for the decision-making and safety management of vertical shaft maintenance throughout its entire lifecycle, improving the comprehensiveness, accuracy, and engineering practicality of defect identification in in-service coal mine vertical shafts. Visual monitoring addresses "surface defect identification," ultrasonic detection addresses "internal structure perception," and stress monitoring addresses "structural stress correlation." This method fills the gap in a multi-dimensional and high-precision quantitative identification system for defects in in-service coal mine vertical shafts, and is of great significance for promoting the upgrading of coal mine shaft safety monitoring technology, enhancing the risk prevention and control capabilities of in-service shaft defects, and promoting the efficient and safe production of coal mines.

[0104] like Figure 1 As shown, the method specifically includes the following steps:

[0105] S1. To address the needs of monitoring the annular structure of a coal mine vertical shaft, visual monitoring equipment (explosion-proof high-definition industrial cameras and laser scanners), ultrasonic detection equipment (multi-frequency ultrasonic probe arrays), and stress monitoring equipment (fiber optic stress sensors) are deployed and installed on the annular structure of the coal mine vertical shaft to collect visual data, ultrasonic data, and stress data, ultimately achieving multi-dimensional data acquisition of the shaft.

[0106] For environments characterized by dampness, dust, and high vibration in underground coal mines, the camera lenses of visual monitoring equipment require additional protection during actual use, including a waterproof glass cover (5mm thick, ≥95% light transmittance, dust and impact resistant). Equipment cables (power and data cables) are run through stainless steel conduits designed for mining environments and connected to an intrinsically safe wireless transmission module (ensuring compliance with coal mine explosion-proof standards). This circular arrangement of cameras and scanners aims to achieve: 1) Full-circumference, blind-spot-free monitoring: By combining camera field-of-view overlap with full-section laser scanning, high-definition images and three-dimensional coordinate data are ensured across the 360° area of ​​the cylinder wall, preventing the overlooking of cracks, corrosion, and other defects due to angle omissions; 2) Precise data correlation: Within the same monitoring section, camera image pixels can be "anchored" to the three-dimensional coordinates of the laser scanner (e.g., a pixel in the image corresponds to the actual coordinates of the laser scan (X1, Y1, Z1)), providing a precise conversion between pixels and actual dimensions for subsequent crack width calculations.

[0107] The inner wall of the annular structure of the coal mine shaft is pre-installed with at least three vertical explosion-proof guide rails, which are evenly arranged in a ring at specified intervals along the inner wall of the shaft (conventionally, they are evenly arranged in a ring at 120° intervals along the inner wall of the shaft to achieve full coverage without blind spots, while ensuring accurate matching of visual data, laser three-dimensional coordinates and ultrasonic detection. The reference positions are 0°, 120° and 240° with the center of the shaft as the central angle). The explosion-proof guide rails are equipped with independently movable explosion-proof mobile carriers, which are used to integrate explosion-proof mobile carrier one and explosion-proof mobile carrier two to form a dual independent bearing unit (vertical spacing ≥10m), which sequentially carries the visual monitoring equipment and the ultrasonic detection equipment. The explosion-proof mobile carrier is equipped with a depth encoder and a locking mechanism, with a positioning error ≤±5mm and vibration resistance ≥5g.

[0108] During the data acquisition process, a monitoring section is set up every 5m to 10m along the inner wall of the well for stopping and collecting data. At this time, the monitoring sections of visual monitoring, ultrasonic probe, and stress sensor are as completely overlapped as possible to ensure that multi-source data can correspond to the same well wall area. Each monitoring section is a circular plane perpendicular to the well axis.

[0109] At this time, the monitoring section should focus on the well neck (0-50m), the bottom of the well, and the sections where geological conditions change (such as rock layer interfaces, the area near the interface between loose layers and bedrock layers).

[0110] In the specific implementation process:

[0111] (1) Visual monitoring equipment: Explosion-proof high-definition industrial camera + laser scanner are used to monitor the direction of cracks and the boundary of corrosion areas on the inner surface of the well wall, wherein:

[0112] A combination of "multi-camera coverage + single scanner calibration" can be used, with the camera as the main component and the scanner as a supplement, to avoid functional redundancy.

[0113] Explosion-proof high-definition industrial cameras acquire images of the well wall surface, capturing the direction of cracks and the boundaries of corroded areas (relying on 5-megapixel resolution to achieve detailed identification).

[0114] The laser scanner outputs the three-dimensional coordinates (X / Y / Z) of each point on the well wall, calibrating the camera's "pixel accuracy" to avoid size calculation errors caused by image distortion;

[0115] Explosion-proof high-definition industrial cameras have a resolution of ≥5 million pixels;

[0116] The laser ranging accuracy of the laser scanner is ±0.1mm;

[0117] The visual monitoring equipment is installed on an explosion-proof mobile carrier 1 with a vertical guide rail, and achieves axial movement monitoring through the explosion-proof guide rail. (Reference) Figure 2 ;

[0118] Data is uploaded to the ground server in real time via an intrinsically safe wireless transmission module for mining applications;

[0119] The explosion-proof high-definition industrial camera is equipped with a rotatable angle adjustment bracket. When the visual monitoring equipment moves to a certain monitoring section, the angle adjustment bracket drives the camera / scanner to adjust the lens angle.

[0120] The explosion-proof high-definition industrial camera rotates via an angle adjustment bracket to point the lens toward the opposite well wall. The "field of view" covered by each explosion-proof high-definition industrial camera must include the edges of the fields of view of two adjacent cameras (to ensure that the fields of view of two adjacent cameras have an overlap of ≥10%, completely eliminating blind spots in the circumferential direction). With the overlap of the fields of view of three explosion-proof high-definition industrial cameras, the entire circumference of the well is covered 360°. Depending on the compatibility between the "field of view" of the explosion-proof high-definition industrial camera and the diameter of the well, it can be increased to four (90° interval) or six (60° interval).

[0121] The laser scanner has a built-in 360° circular scanning function (no rotation required), with a scanning accuracy of ±0.1mm. A single scan can obtain the three-dimensional coordinates of the entire monitoring section, directly replacing the need for circumferential movement.

[0122] The laser scanner, with the help of components such as a horizontal adjustment arm and an angle fine-tuning mechanism, can be adjusted to the target position "directly below the center of the monitoring section" or "coaxial with the explosion-proof high-definition industrial camera" and then locked in place. When "directly below the center of the section", the laser emission direction is perpendicular to the well wall surface. When "coaxial with the explosion-proof high-definition industrial camera", it can be directly and synchronously positioned with the camera, which is more efficient and ensures that the "laser three-dimensional coordinates" and "camera image pixels" are accurately matched (coaxial means that the image and coordinates of the same area come from the same optical axis), providing a more accurate data foundation for subsequent "crack width quantification calculation".

[0123] (2) Ultrasonic detection equipment: A multi-frequency ultrasonic probe array is used during ultrasonic data acquisition to monitor the thickness of corrosion inside the well wall and the depth of deep cracks. The multi-frequency ultrasonic probe array includes 6 multi-frequency ultrasonic probes with a detection frequency of 2.5-10MHz. Different frequencies are used to adapt to different detection targets of different defects. Each probe can switch between three frequencies: 2.5MHz, 5MHz, and 10MHz. The specific functions are as follows:

[0124] 1) 2.5MHz low frequency: Deep detection depth, suitable for locating "deep cracks inside the well wall" and "areas with steel reinforcement corrosion";

[0125] 2) 5MHz intermediate frequency: high detection accuracy (resolution ≤ 0.1mm), suitable for quantification of "internal corrosion thickness";

[0126] 3) 10MHz high frequency: Focuses on the shallow layer below the surface, adapted to the detection of "surface crack extension depth" (forming "depth-width" linked data with visually recognized surface cracks).

[0127] The detection radius of the multi-frequency ultrasonic probe is ≥0.5m, the penetration depth is ≤500mm, the detection range of the 6 multi-frequency ultrasonic probes completely covers the circumference of the inner wall of the well, and the overlap rate of the detection areas of adjacent multi-frequency ultrasonic probes is ≥10%.

[0128] The ultrasonic detection equipment includes multiple multi-frequency ultrasonic probes, which are installed on an explosion-proof mobile carrier with a vertical guide rail. Each explosion-proof mobile carrier is equipped with two multi-frequency ultrasonic probes. The ultrasonic detection equipment for ultrasonic data acquisition and the visual monitoring equipment for visual data acquisition share the vertical explosion-proof guide rail to achieve axial movement monitoring.

[0129] Each multi-frequency ultrasonic probe integrates a solid coupling pad on its probe end face. The coupling pad is made of butyl rubber modified composite material, 5 mm thick, with an acoustic impedance of ≈2.8 × 10⁻⁶. 6 kg / (m 2 (s) The surface has an anti-slip texture, and the edges are chamfered;

[0130] like Figure 3As shown, the explosion-proof mobile carrier integrates a retractable actuator, including an adjustable ring bracket, an explosion-proof electric push rod, a parallel guide rail, and a pressure sensor. During ultrasonic detection, the explosion-proof mobile carrier moves along the vertical guide rail to the target monitoring section, and the adjustable ring bracket rotates to a preset angle. The probe position is 30° away from the vertical explosion-proof guide rail reference position, so that the 6 probes are pushed to the circumferential positions of 30°, 90°, 150°, 210°, 270°, and 330° respectively, avoiding obstruction by the guide rail.

[0131] The explosion-proof electric push rod pushes the probe to make vertical contact with the well wall. After the data collection is completed, the probe is retracted and the data is uploaded to the ground server in real time through the intrinsically safe wireless transmission module for mining.

[0132] The ultrasonic detection equipment and the visual monitoring equipment operate in the order of "visual detection first, ultrasonic detection later", covering all monitoring sections in sequence. The field of view overlap rate of the detection area of ​​the same monitoring section is ≥10%. The data is automatically aligned according to the section depth reference, and the alignment error is ≤±5mm, which does not affect the subsequent multi-source fusion.

[0133] (3) Stress monitoring equipment: used to monitor sudden changes in wellbore stress (associated with spalling risk), using fiber optic grating stress sensors. The range of the fiber optic grating stress sensors is 0 to 200 MPa, and the accuracy is ±0.5%FS.

[0134] Stress monitoring equipment should be preferentially placed in the well neck (which is greatly affected by surface loads), the bottom of the well (which is greatly affected by mining), and sections where geological conditions change (such as near rock strata interfaces, loose layers and bedrock interfaces), and should be placed on the same cross section as visual and ultrasonic equipment to facilitate multi-source data alignment.

[0135] like Figure 4 As shown, 3 to 6 groups of fiber optic stress sensors are arranged in a ring along the inner wall of the wellbore at each monitoring section, adopting the principle of "equal interval + key reinforcement" - in the normal area, they are arranged at 120° intervals (corresponding to adjacent positions of 0°, 120°, and 240° on the guide rail). If a slight crack has appeared in a certain area (predicted by the visual equipment), 3 additional sensors (60° interval) are added to cover a 30cm range on both sides of the crack.

[0136] The sensitive axis of the fiber optic stress sensor is matched with the stress monitoring target. When monitoring axial stress, the sensor is attached along the vertical (up and down) direction of the wellbore to capture the vertical compressive / tensile stress of the well wall (associated with spalling defects). When monitoring circumferential stress, the sensor is attached along the circumference of the wellbore to capture radial pressure changes (associated with crack propagation).

[0137] The fiber optic stress sensor is installed using a post-mounting method, following a three-step process: surface pretreatment → bonding and fixing with explosion-proof epoxy adhesive for coal mines → protective encapsulation with explosion-proof stainless steel housing.

[0138] In conjunction with the design of the explosion-proof guide rail movement monitoring, the optical fiber is laid along the dedicated cable tray on the side of the explosion-proof guide rail inside the well shaft (the cable tray depth is ≥2cm and the width is ≥1.5cm), and is fixed with explosion-proof clamps every 1m (the clamp spacing is ≤30cm to avoid the optical fiber shaking); a "buffer bend" (bending radius ≥30cm to avoid tensile breakage) is set every 50m for the vertical optical fiber, and a 1m redundancy length is reserved (to accommodate slight deformation of the well shaft);

[0139] The underground optical fiber is collected through an explosion-proof junction box (intrinsically safe type) at the top of the shaft, and then transmitted to the ground monitoring room via a dedicated mine optical fiber trunk line. The ground end is connected to a fiber optic grating demodulator, which converts the optical signal into a stress electrical signal. Finally, it is uploaded to the server through a mine-use intrinsically safe wireless transmission module and stored synchronously with visual and ultrasonic data.

[0140] In summary, the data acquisition system of this method combines three reusable guide rails with a dynamic telescopic device, which can solve the problem of complex modification of traditional multi-equipment and multi-guide rail systems, reduce the amount of wellbore modification, and replace conventional liquid coupling agent with solid coupling pads and temporary clamping operation, thereby eliminating the need for a supply / recovery system, simplifying the equipment structure, and adapting to downhole maintenance needs.

[0141] It should also be noted that, in order to achieve the fusion of stress-vision-ultrasound data, the sensor needs to form a "three-source positioning" with the vision camera and ultrasound probe of the same cross section - the straight-line distance between each sensor and the adjacent vision camera and ultrasound probe is ≤0.5m, to ensure that the monitoring area completely overlaps and avoid blind spots in data association.

[0142] The monitoring and quantification methods used here are compliant with the following standards: ultrasonic detection sound velocity correction refers to GB / T50344-2019 "Technical Standard for Building Structure Testing"; stress monitoring accuracy complies with GB / T13606-2007 "General Technical Conditions for Vibrating Wire Sensors of Geotechnical Testing Instruments and Geotechnical Engineering Instruments"; and the visual data processing flow conforms to MT / T1097-2008 "Technical Specification for Maintenance of Coal Mine Electromechanical Equipment", thus forming a standardized quantitative identification process.

[0143] S2. Perform noise reduction, correction, and smoothing preprocessing on the collected data. The specific operation process includes:

[0144] S21. An adaptive median filtering algorithm is used to eliminate salt-and-pepper noise caused by mineral dust. The algorithm formula is as follows:

[0145] ;

[0146] Where f(x,y) is the pixel value of the original visual image; g(x,y) is the pixel value after denoising; med{·} is the median algorithm, used to eliminate isolated noise points while preserving crack edge features;

[0147] S22. The influence of wellbore concrete temperature (25-60℃) on ultrasonic propagation speed is corrected by the sound velocity compensation formula. At this time, the ultrasonic probe array is equipped with a self-calibration module, which integrates a 100mm thick concrete standard test block to support on-site calibration in the well. After calibration, the ultrasonic data detection error is ≤2%, which is compatible with the ultrasonic sound velocity compensation formula. The sound velocity compensation formula is as follows:

[0148] ;

[0149] in, Expressed as the velocity of sound in concrete at 25°C; The measured wellbore temperature (°C) is used. The corrected sound velocity is used to ensure the accuracy of ultrasonic depth detection;

[0150] S23. Finally, a central moving average filter is used to eliminate mechanical vibration interference. The filtering formula is as follows:

[0151] ;

[0152] in, This is the original stress value; Indicates the size of the sliding window (take an odd number between 3 and 7 to ensure that the sliding window is centered on the target data point and avoid data offset); The smoothed stress value is used to preserve stress abrupt changes (corresponding to spalling disease). The time series index of the target smoothed data points (corresponding to the first) The original data points acquired during secondary stress sampling. =1,2,3...); Indexes of the original data points (range of values) that participate in the averaging within the sliding window. to ,like hour, , , ; hour, , , , , , , ).

[0153] S3. Calculate the information entropy and fusion weight of the three types of data after preprocessing based on the entropy weight method to complete the feature layer fusion.

[0154] At this point, an entropy weighting method-weighted fusion model is adopted, which allocates fusion weights based on the information entropy (credibility) of different data sources to avoid bias caused by a single data source dominating the fusion. Specifically, this includes:

[0155] S31. Extract the disease characteristic indicators from visual features, ultrasonic features, and stress features respectively, whereby the disease characteristic indicators specifically include:

[0156] Visual features: Crack pixel length Pixel area of ​​eroded region ;

[0157] Ultrasonic characteristics: internal corrosion thickness Depth of deep cracks ;

[0158] Stress characteristics: The difference between the measured value and the reference value is taken as the stress mutation value. σs;

[0159] S32. Calculate the information entropy of each data source. The calculation formula is as follows:

[0160] ;

[0161] ;

[0162] in, These represent vision, ultrasound, and stress, respectively. To monitor the number of cross sections; For the first The first class feature The original feature value (i.e. the first) Class of monitoring indicators in the first (Original data collected at each monitoring section) For the first The first class of data The eigenvalue normalization result of the cross section (i.e., for) After normalization, the percentage of the data in all cross-sectional data of the same feature is obtained. The smaller the information entropy, the lower the data dispersion and the higher the credibility.

[0163] S33. Calculate the fusion weights The calculation formula is as follows:

[0164] ;

[0165] and .

[0166] It should also be noted that, in specific embodiments, in order to make the weight allocation more aligned with the quantitative needs of specific diseases, the weight fusion process can be specifically adapted to the disease characteristic indicators of the current monitoring section based on the basic weights (the adjustment is based on the core contribution of each data source in different disease types):

[0167] If the visual data first identifies the crack feature indicators (such as the crack pixel length corresponding to the feature indicator) (≥50 pixels), then the basic weight of visual features. Make adaptive adjustments (such as after adjustment) Strengthen the contribution of visual data in the quantification of crack-type diseases;

[0168] If visual data first identifies the spalling feature indicators (such as the pixel area of ​​concrete spalling corresponding to the feature indicators) (≥100 pixels), then the basic weight of the stress feature. Basic weights of ultrasound features Make adaptive adjustments separately (e.g., after adjustment) , The synergistic contribution of stress-ultrasound data to the quantification of exfoliation-related diseases was enhanced.

[0169] Normalize the adjusted weights (to ensure) By doing so, the fusion weights associated with the disease characteristic indicators can be obtained.

[0170] In summary, by allocating credibility weights for multi-source data, it is possible to automatically adapt to different monitoring environments (such as increasing the weight of ultrasound when the concentration of mineral dust is high), improve the accuracy of multi-source data fusion, and avoid interference from poor-quality data on the identification results.

[0171] In a specific embodiment, this method can eliminate the blind spots of single data by synergistically fusing multi-source monitoring data (visual, ultrasonic, stress), and achieve all-round perception of wellbore defects on the surface, inside and under structural stress. At this time, by using an explosion-proof high-definition industrial camera + laser scanner to capture surface features (crack direction recognition rate ≥98%), multi-frequency ultrasonic probe array to detect internal structure (corrosion thickness error ≤2%), and fiber optic stress sensor to correlate stress state, it can significantly reduce the defect omission rate compared with traditional single methods and increase the internal corrosion recognition rate from 65% to 98%.

[0172] S4. Establish quantitative formulas for crack width, corrosion area, and spalling depth. Combined with the fused feature layer, output the quantitative indicators and severity level of the disease through final weighting, where:

[0173] (1) Crack width quantification formula, used to combine the pixel precision of the visual image (P=0.05mm / pixel) with the three-dimensional coordinates of the laser scan to calculate the actual crack width, the specific formula is as follows:

[0174] ;

[0175] in, Weights for visual feature fusion; The length of the crack is in pixels;P For visual camera pixel precision; Represents visual width weights (data from step S3). Adjustment); Weights for ultrasound feature fusion; This refers to the ultrasonic echo time difference (the echo time difference between the upper and lower interfaces of the crack). The width of the crack detected by ultrasonic testing; Weighted by the ultrasonic width;

[0176] At this point, accuracy verification shows that the crack width error calculated by this formula is ≤0.03mm, which is better than single visual recognition (error ≤0.1mm).

[0177] (2) The corrosion area quantification formula is used to integrate the surface area of ​​visual segmentation with the internal corrosion area detected by ultrasonic detection to calculate the total corrosion area. This can solve the problem that a single visual method cannot identify internal corrosion and realize the quantification of the total area of ​​"surface + interior". The specific formula is as follows:

[0178] ;

[0179] in, This represents the number of pixels in the visually eroded area. The pixel area; The number of internal corrosion areas detected by ultrasonic testing (each detection point covers a circular area with a radius of r = 0.1 m). The area of ​​a single ultrasonic detection region;

[0180] (3) The spalling depth quantification formula is used to correlate stress mutation values ​​with ultrasonic depth detection to calculate the spalling depth of the well wall. It can achieve dual verification of depth by "structural response + direct detection" with an error ≤ 5 mm. The specific formula is as follows:

[0181] ;

[0182] in, (Expressed as the elastic modulus of concrete). The strain corresponding to the sudden change in stress; The thickness of the wellbore wall is used to estimate the depth of structural deformation caused by spalling through stress back calculation. The actual depth of the exfoliation detected by ultrasonic testing; , All are weights (from step S3) Sure);

[0183] (4) By using the final weight output of the disease quantification index and the quantification index based on the three disease types of crack width W, corrosion area S and spalling depth H in the severity level process, a wellbore disease severity level matrix is ​​established to achieve the direct mapping of "quantification index - maintenance priority".

[0184] The severity levels of wellbore defects are divided into four categories: minor (Level I), moderate (Level II), severe (Level III), and critical (Level IV). The quantitative indicators corresponding to each severity level of wellbore defects are as follows:

[0185] Slight (Level I): W<0.2mm, S<1.0m 2 H<10mm;

[0186] Moderate (Level II): 0.2mm ≤ W < 0.5mm, 1.0m 2 ≤S<3.0m 2 10mm≤H<30mm;

[0187] Severe (Level III): 0.5mm ≤ W < 1.0mm, 3.0m 2 ≤S<5.0m 2 30mm≤H<50mm;

[0188] Emergency (Level IV): W≥1.0mm, S≥5.0m 2 H≥50mm.

[0189] In summary, by establishing a mapping model between multi-source data and quantitative indicators of defects (width, area, depth, and severity level), and achieving real-time synchronization of multi-source data fusion results, the system automatically classifies defects into severity levels I-IV. It can output quantitative results that can directly support maintenance decisions. By directly matching maintenance plans with quantitative indicators such as crack width, corrosion area, and spalling depth (e.g., a crack width of 0.3 mm corresponds to a grouting pressure of 1.2 MPa), maintenance decisions are transformed from "experience-driven" to "data-driven," reducing operation and maintenance costs by 30%, wellbore accident rate by 40%, and significantly improving the efficiency of safety management and control of in-service wellbores.

[0190] In a further embodiment, the main shaft of a coal mine has a diameter of 6m, a depth of 800m, and a wall thickness of [missing information]. h 0=500mm, elastic modulus of concrete E c=30GPa; Monitoring layout: 1 monitoring section is set up every 5m along the axis, for a total of 160 sections, with a focus on analyzing the 320m geological change section (visual prediction of minor cracks); Equipment parameters: visual camera pixel accuracy. P =0.05mm / pixel, ultrasonic probe detection radius r=0.1m, stress sensor range 0~200MPa, accuracy ±0.5%FS; environmental parameters: actual measured temperature downhole. t =35℃.

[0191] Multi-source equipment is deployed inside the wellbore, and three vertical explosion-proof guide rails (arranged at 0°, 120°, and 240°) are reused. Two explosion-proof mobile carriers carry visual / ultrasonic equipment respectively (with a vertical spacing of 12m). Six sets of stress sensors (spaced at 60° intervals) are arranged in a 320m section, and the monitoring sections of the three types of equipment completely overlap.

[0192] (1) Data preprocessing is performed after data acquisition from the three types of equipment:

[0193] ① Visual data denoising (adaptive median filtering), using the formula The pixel values ​​of the crack area in the original data (89 represents salt-and-pepper noise caused by mine dust), take the median value. Denoising calculation of the denoised pixel value Preserve the characteristics of the crack edges.

[0194] ② Ultrasonic data correction (sound velocity compensation), using the formula Calculated The corrected sound velocity is used for subsequent ultrasonic data calculations of crack width and spalling depth to ensure detection accuracy.

[0195] ③ Stress data smoothing (center moving average filtering), through (N=5, i=3), original data [12,13,28,14,13], after smoothing calculation Mutation value .

[0196] (2) Feature layer fusion

[0197] ① For the 320m monitoring section, extract disease characteristic indicators, visually. L v =800 pixels S v =40,000 pixels; ultrasound d u =12mm h u =25mm; Stress mutation value Δ σ s =12MPa.

[0198] ② Calculate information entropy, visual information entropy Ultrasonic information entropy Stress information entropy .

[0199] ③ Calculate the fusion weights (calculate the fusion weights), base weights , , After adjustment based on disease characteristics .

[0200] (3) Precise quantitative identification of diseases and quantitative identification of crack width (0.40mm (weighted average) × 1.05 (correction factor) = 0.42mm); Corrosion area quantification Quantification of peeling depth .

[0201] (4) According to the patent disease level matrix, the virus level of the 320m monitoring section is moderate (Level II). The output recommendation is: grouting reinforcement, grouting pressure 1.2MPa (adaptive parameter corresponding to crack width of 0.3-0.5mm).

[0202] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium, an optical medium, or a semiconductor medium, etc.

[0203] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0204] Furthermore, it should be noted that if any directional indication (such as up, down, left, right, front, back, etc.) is involved in the embodiments of the present invention, the directional indication is only used to explain the relative positional relationship and movement of each component in a specific posture. If the specific posture changes, the directional indication will also change accordingly.

[0205] Furthermore, if the embodiments of this invention involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the meaning of "and / or" throughout the text includes three parallel solutions; for example, "A and / or B" includes solution A, solution B, or a solution where both A and B are satisfied simultaneously. Furthermore, in the embodiments of this invention, "multiple" refers to two or more. Moreover, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

Claims

1. A method for accurate quantitative identification of in-service coal mine shaft defects based on multi-source heterogeneous driving, characterized in that, include: S1. Install monitoring equipment on the annular structure of the vertical shaft of the coal mine to collect visual data, ultrasonic data, and stress data; S2. Preprocess the collected data; S3. Calculate the information entropy and fusion weight of the three types of data after preprocessing based on the entropy weight method to complete the feature layer fusion; S4. Establish quantitative formulas for crack width, corrosion area, and spalling depth, and combine them with the fused feature layer to output the quantitative indicators and severity level of the disease through the final weight; In step S1, at least three vertical explosion-proof guide rails are pre-installed on the inner wall of the annular structure of the coal mine shaft, and are evenly arranged in a ring at specified intervals along the inner wall of the shaft. The vertical explosion-proof guide rail is equipped with an independently movable explosion-proof mobile carrier, which integrates two independent bearing units to sequentially carry the visual monitoring equipment and the ultrasonic detection equipment. During the data acquisition process, a monitoring section is set up every 5m to 10m along the inner wall of the well shaft for stopping and collecting data. Each monitoring section is a circular plane perpendicular to the axis of the well shaft. In step S1, a multi-frequency ultrasonic probe array is used in the ultrasonic data acquisition process. The ultrasonic detection device includes multiple multi-frequency ultrasonic probes, which are installed on the explosion-proof mobile carrier of the vertical explosion-proof guide rail. Two multi-frequency ultrasonic probes are installed on each explosion-proof mobile carrier. The ultrasonic data acquisition device and the visual data acquisition device share the vertical explosion-proof guide rail to achieve axial movement monitoring. The explosion-proof mobile carrier integrates a retractable actuator, including an adjustable ring bracket, an explosion-proof electric push rod, and a parallel guide rail. During ultrasonic detection, the explosion-proof mobile carrier moves along the vertical explosion-proof guide rail to the target monitoring section, the adjustable ring bracket rotates to a preset angle, and the probe position is 30° away from the reference position of the vertical explosion-proof guide rail, so that the 6 probes are pushed to the circumferential positions of 30°, 90°, 150°, 210°, 270°, and 330° respectively, avoiding obstruction by the guide rail. The explosion-proof electric push rod pushes the probe to make vertical contact with the well wall. After the data acquisition is completed, the probe is retracted, and the data is uploaded to the ground server in real time through the intrinsically safe wireless transmission module for mining.

2. The method for accurate quantitative identification of in-service coal mine shaft defects based on multi-source heterogeneous driving as described in claim 1, characterized in that, The preprocessing procedure in step S2 includes: S21. An adaptive median filtering algorithm is used to eliminate salt-and-pepper noise caused by mineral dust. The algorithm formula is as follows: ； Where f(x,y) is the pixel value of the original visual image; g(x,y) is the pixel value after denoising; med{·} is the median algorithm, which is used to eliminate isolated noise points while preserving crack edge features; S22. The effect of wellbore concrete temperature on ultrasonic propagation speed is corrected by the sound velocity compensation formula, which is: ； in, Expressed as the velocity of sound in concrete at 25°C; This refers to the measured wellbore temperature. The corrected sound velocity is used to ensure the accuracy of ultrasonic depth detection; S23. Finally, a central moving average filter is used to eliminate mechanical vibration interference. The filtering formula is as follows: ； in, This is the original stress value; Indicates the size of the sliding window; This is the smoothed stress value, used to preserve the characteristics of stress abrupt changes; A time-series index for the target smoothed data points; This is the index of the original data points that participate in the averaging within the sliding window.

3. The method for accurate quantitative identification of in-service coal mine shaft defects based on multi-source heterogeneous driving as described in claim 1, characterized in that, Step S3 employs an entropy-weighted fusion model, allocating fusion weights based on the information entropy of different data sources. Specifically, this includes: S31. Extract the disease characteristic indicators based on visual features, ultrasonic features, and stress features respectively; S32. Calculate the information entropy of each data source. The calculation formula is as follows: ； ； in, These represent vision, ultrasound, and stress, respectively. To monitor the number of cross sections; For the first The first class feature One original feature value; For the first The first class of data The normalization result of the eigenvalues ​​of the cross section, and the smaller the information entropy, the lower the data dispersion and the higher the reliability; S33. Calculate the fusion weights The calculation formula is as follows: ； and .

4. The method for accurate quantitative identification of in-service coal mine shaft defects based on multi-source heterogeneous driving as described in claim 3, characterized in that, The disease characteristic indicators in step S31 specifically include: Visual features: Crack pixel length Pixel area of ​​eroded region ; Ultrasonic characteristics: internal corrosion thickness Depth of deep cracks ; Stress characteristics: The difference between the measured value and the reference value is taken as the stress mutation value. σs.

5. The method for accurate quantitative identification of in-service coal mine shaft defects based on multi-source heterogeneous driving as described in claim 4, characterized in that, The crack width quantification formula in step S4 is used to calculate the actual crack width by combining the pixel precision of the visual image with the three-dimensional coordinates of the laser scan. The specific formula is as follows: ； in, Weights for visual feature fusion; The length of the crack is in pixels; P For visual camera pixel precision; Indicates visual width weight; Weights for ultrasound feature fusion; This refers to the ultrasonic echo time difference; The width of the crack detected by ultrasonic testing; This represents the ultrasonic width weight.

6. The method for accurate quantitative identification of in-service coal mine shaft defects based on multi-source heterogeneous driving as described in claim 4, characterized in that, The corrosion area quantification formula in step S4 is used to fuse the surface area of ​​visual segmentation with the internal corrosion area detected by ultrasonic testing to calculate the total corrosion area. The specific formula is as follows: ； in, This represents the number of pixels in the visually eroded area. The pixel area; The number of internal corrosion zones detected by ultrasonic testing; The area of ​​a single ultrasonic detection region.

7. The method for accurate quantitative identification of in-service coal mine shaft defects based on multi-source heterogeneous driving as described in claim 4, characterized in that, The spalling depth quantification formula in step S4 is used to correlate stress mutation values ​​with ultrasonic depth detection to calculate the wellbore spalling depth. The specific formula is as follows: ； in, Expressed as the elastic modulus of concrete; The strain corresponding to the sudden change in stress; For wellbore wall thickness; The actual depth of the exfoliation detected by ultrasonic testing; The weight of stress data in the quantification of spalling depth; The weight of ultrasound data in the quantification of exfoliation depth.

8. The method for accurate quantitative identification of in-service coal mine shaft defects based on multi-source heterogeneous driving as described in claim 1, characterized in that, In step S4, the quantitative indicators and severity levels are finally weighted and output using the quantitative indicators of crack width W, corrosion area S, and spalling depth H to establish a wellbore disease severity level matrix. The quantitative indicators corresponding to the severity level of wellbore defects are as follows: Slight: W<0.2mm, S<1.0m 2 H<10mm; Moderate: 0.2mm ≤ W < 0.5mm, 1.0m 2 ≤S<3.0m 2 10mm≤H<30mm; Severe: 0.5mm ≤ W < 1.0mm, 3.0m 2 ≤S<5.0m 2 30mm≤H<50mm; Urgent: W≥1.0mm, S≥5.0m 2 H≥50mm.

Images