Mine supporting product quality management and control method and system

Abstract

The present application belongs to the technical field of image processing, and particularly relates to a mine support product quality management and control method and system, which comprises the following steps: generating a target image sequence; performing model matching and image mapping on the target image after edge detection to obtain a target edge; extracting a target image feature point to generate a registration target image and a registration target image sequence; positioning a target area, calculating a target main shaft direction angle; calculating a target centroid coordinate according to a target area pixel point abscissa, ordinate and total number of pixel points; calculating a target asymmetric distortion degree according to the distance between the target centroid coordinate and the target main shaft straight line and the total number of target pixel points; in the registration target image sequence, calculating an anchor rod quality risk index, drawing an anchor rod quality risk baseline, and analyzing according to a preset threshold to trigger the system to issue an alarm and prompt replacement of the anchor rod. The present application realizes dynamic management and control of the anchor rod throughout its life cycle.

Description

Technical Field

[0001] This invention relates to the field of image processing technology. More specifically, this invention relates to a method and system for quality control of mine support products. Background Technology

[0002] In mining operations, support products are crucial for ensuring the safety of mine roadways. Among them, anchor bolts, as a key support structure, directly impact mine safety. With increasing mining depth and more complex geological conditions, anchor bolts are subjected to various forces during use, making them prone to deformation and damage. Failure to promptly monitor changes in anchor bolt quality can lead to loss of support function, potentially causing serious accidents such as roadway collapses, threatening the lives of construction workers and disrupting normal mine operations.

[0003] To detect the quality of anchor bolts, traditional methods rely on manual inspection, which is time-consuming and has a high rate of missed detections. Other technologies rely on single-point sensor monitoring, such as strain gauges, which are easily damaged by mine pressure, have high deployment costs, and the sensors can only monitor local stress and cannot correlate with multiple sources of risk such as surrounding rock displacement and environmental humidity. This results in no early warning before anchor bolt failure and passive accident prevention.

[0004] In related technologies, for example, Chinese patent document CN116703903B, entitled "A Machine Vision-Based Method for Anchor Bolt Repair Quality Inspection," discloses a method that utilizes machine vision technology to acquire images of the anchor bolt repair area, perform preprocessing, edge extraction, and feature analysis to achieve repair quality inspection. This method solves the problem of quality acceptance after anchor bolt repair and provides a digital means for evaluating repair effectiveness.

[0005] However, this method mainly focuses on the quality inspection of anchor bolts after repair, which is a phased inspection and cannot achieve dynamic quality control of anchor bolts from the initial installation to long-term use. During the long-term service of anchor bolts, their quality will gradually deteriorate, and this method cannot capture these gradual changes in a timely manner, resulting in a lag in the control of anchor bolt quality and making it difficult to meet the needs of long-term monitoring of anchor bolt quality for safe production in mines. Summary of the Invention

[0006] To address the aforementioned technical problem of the lack of dynamic management and control throughout the entire life cycle of anchor bolts, this invention provides solutions in the following aspects.

[0007] In a first aspect, the present invention provides a method for quality control of mine support products, comprising:

[0008] Anchor bolt pad images are continuously acquired and preprocessed at preset intervals. The resulting target images are arranged chronologically to generate a target image sequence. Model matching and image mapping are performed on the target images after edge detection to obtain the target edges. Feature points are extracted from the target images, and feature weights are calculated based on the feature points, the ratio of the maximum distance to the target edge, and the ratio of the minimum distance to all feature points. High-weight feature points are selected to register the target images, generating registered target images and a registered target image sequence. The target region is located according to the target edges, based on the line connecting the pair of target points with the largest distance in the target convex hull formed by the target region, the x-coordinates and y-coordinates of all pixels in the target region, and the distance between the target points and the target edges. The target region's pixel count ratio is used to calculate the target principal axis direction angle and centroid coordinates. Based on the perpendicular distance between the target centroid coordinates and the target point pair forming the target principal axis line, and the total number of pixels in the target region, the target asymmetric distortion is calculated. In the registered target image sequence, the anchor bolt quality risk index is calculated based on the maximum values ​​of the centroid coordinate distance, the change in the target principal axis direction angle, and the change in the target asymmetric distortion in all adjacent registered target images within a preset time period. An anchor bolt quality risk baseline is drawn based on the anchor bolt quality risk index within a preset time period, and analysis is performed according to a preset threshold. This triggers the system to issue an alarm and prompt for anchor bolt replacement.

[0009] Traditional methods suffer from long intervals of manual inspection, making it easy to miss problems. Sensors are prone to damage, costly, and can only monitor localized areas. Existing technologies can only perform phased inspections after repairs, unable to provide long-term tracking. This invention acquires images continuously and organizes them into a sequence, ensuring accurate alignment of images from different times. It then analyzes the angles, center of gravity, distortion, and other states of components to calculate risks and plot trend lines, achieving dynamic control from installation to long-term use. This allows for timely detection of subtle changes and sudden problems in components, preventing accidents caused by missed inspections or delayed judgments. It overcomes the shortcomings of traditional methods and existing technologies, making support quality control more timely and comprehensive.

[0010] Preferably, generating the target image sequence includes:

[0011] In the initial stage after the anchor bolt installation is completed and the quality is intact, an industrial camera is used to photograph the anchor bolt pad, and the obtained images are smoothed and denoised to obtain a grayscale image of the anchor bolt pad. According to a preset time period and a preset frequency, the anchor bolt pad is continuously photographed at the same location with the same camera and parameters to obtain an image sequence of the anchor bolt pad. All images in the anchor bolt pad image sequence are converted to grayscale to generate a target image sequence composed of multiple target images arranged according to the acquisition time.

[0012] Preferably, the feature weights satisfy the following expression:

[0013] ;

[0014] In the formula, This represents the feature weight of the i-th feature point; This represents the shortest distance from the i-th feature point to the target edge; This represents the maximum distance from all feature points in the target image to the target edge; e is a natural constant.

[0015] This invention calculates weights based on the distance from feature points to edges, with points farther away having higher weights. This filters out less disturbed and more stable points, allowing these points to play a major role in image alignment, reducing the influence of factors such as edge blurring and shadows, making the alignment results more reliable, and laying a more accurate foundation for subsequent analysis.

[0016] Preferably, generating the registration target image and the registration target image sequence includes:

[0017] Calculate the feature weights of all extracted feature points in the target image, sort these feature points from high to low according to their feature weights, select the top x feature points, use the RANSAC algorithm to remove erroneous matching pairs, and then calculate the geometric transformation matrix to complete the registration and generate the registered target image; repeat the above process on the target image sequence to obtain the registered target image sequence.

[0018] Preferably, the target principal axis direction angle satisfies the following expression:

[0019] ;

[0020] In the formula, The principal axis direction angle of the target region is represented; a and b are the pair of vertices that are farthest apart on the convex hull formed by the target region. , This represents the ordinate of the target point relative to the midpoints a and b; , This represents the x-coordinate of the target point relative to the midpoints a and b. The x-coordinates of a and b are not equal. If they are equal, the principal axis direction angle of the target is 90°. Represents the absolute value function; This represents the arctangent function.

[0021] This invention, by calculating the spindle direction angle, can clearly reflect the tilt or torsion of a component. When the component is subjected to changes in force, its tilt or torsion will be directly reflected in this angle. This angle allows for a direct assessment of whether the component has deviated from its normal state, making condition monitoring simpler and clearer.

[0022] Preferably, calculating the target centroid coordinates includes:

[0023] Count the number of pixels in the same target area and record it as the total number of target pixels; count the sum of the x-coordinates of all pixels in the target area and record it as the total x-coordinates of target pixels; count the sum of the y-coordinates of all pixels in the target area and record it as the total y-coordinates of target pixels; record the ratios between the total number of target pixels and the sum of the x-coordinates of target pixels and the sum of the y-coordinates of target pixels as the centroid coordinates of the target.

[0024] Preferably, the target asymmetric distortion degree satisfies the following expression:

[0025] ;

[0026] In the formula, Indicates the asymmetric distortion degree of the target; The x-coordinate and y-coordinate of the target centroid represent the target centroid; N represents the total number of target pixels, used to correlate the centroid offset distance with the area of ​​the anchor plate; A, B, and C represent the linear equation coefficients of the target principal axis line. Represents the absolute value function; This represents the normalization function.

[0027] This invention determines asymmetric distortion by calculating the distance from the center of mass to the principal axis, enabling the detection of uneven local stress on components. When a component experiences abnormal local stress, its center of mass deviates from the principal axis. This distance quantifies this deviation, allowing for more detailed quality assessment and preventing the overlooking of potential local problems.

[0028] Preferably, the calculation of the anchor bolt quality risk index includes:

[0029] The target centroid translation, target principal axis change, and target asymmetric speed are calculated based on the maximum value of the Euclidean distance between the target centroids, the maximum value of the principal axis direction angle change, and the maximum value of the asymmetric distortion change of all adjacent registered target images in the registered target image sequence.

[0030] The anchor bolt quality risk index satisfies the following expression:

[0031] ;

[0032] In the formula, , , This represents the degree of centroid translation, principal axis change, and asymmetric velocity change of the target in the i-th registered target image. This represents the maximum value function.

[0033] This invention uses the maximum value of three indicators as the risk index, enabling timely detection of any sudden problem in any dimension. Whether it's a component shift, angular change, or local distortion, as long as one of these changes is drastic, it can be quickly identified, preventing a single-dimensional risk from being masked by other normal indicators and improving the timeliness of early warnings.

[0034] Preferably, drawing a baseline for anchor bolt quality risk includes:

[0035] All anchor bolt quality risk indices calculated within a preset time period are organized in chronological order. The time information corresponding to each index is associated with the anchor bolt quality risk index to form a dataset containing the correspondence between time and risk index, which is stored in the system's memory. A two-dimensional coordinate system is established with time as the horizontal axis and anchor bolt quality risk index as the vertical axis. The risk index corresponding to each time point is marked as a data point in the two-dimensional coordinate system. Adjacent data points are connected sequentially by line segments to form a continuous curve, which is recorded as the anchor bolt quality risk baseline.

[0036] Secondly, the present invention provides a quality control system for mine support products, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned method for quality control of mine support products is implemented.

[0037] By adopting the above technical solution, a computer program is generated from the above-mentioned method for quality control of mine support products and stored in a memory so that it can be loaded and executed by a processor. Terminal equipment is then made based on the memory and processor for convenient use.

[0038] The beneficial effects of this invention are as follows: It shifts support quality control from a passive response to proactive prevention. Without relying on a large amount of manpower or easily damaged equipment, it can automatically track component status over a long period through image analysis, intuitively displaying risk changes and issuing timely warnings. This not only reduces the possibility of safety accidents and ensures the safety of construction personnel, but also makes mine management more efficient, reduces the risk of production interruptions due to support problems, provides strong support for stable mine production, and promotes the development of support quality control towards a more intelligent and reliable direction. Attached Figure Description

[0039] Figure 1 This is a flowchart illustrating a method for quality control of mine support products according to the present invention;

[0040] Figure 2 This is a schematic representation of the grayscale image of the anchor plate. Detailed Implementation

[0041] This invention discloses a method for quality control of mine support products, referring to... Figure 1This includes steps S1-S4:

[0042] S1: Collect anchor plate images continuously at a preset frequency within a preset time period and perform preprocessing. Arrange the obtained target images according to time to generate a target image sequence.

[0043] It should be noted that in mine support, the stability of anchor bolts is crucial for safety. As a key connecting component, the position, angle, and shape changes of the anchor bolt pad can indirectly reflect the stress and mass of the anchor bolt. This invention first acquires target images and sequences. Initially, images of the pad are captured and processed as a baseline. Subsequently, images are continuously captured and processed at preset time intervals to generate sequences, providing data for analysis. Because images captured at different times may have deviations, direct comparison is difficult to determine deformation; therefore, registration is necessary. Registration is completed using highly stable feature points through edge detection and other operations, generating registered images and sequences to ensure that images can be compared under the same standard, meeting the needs of dynamic monitoring. After registration, the pad's posture and deformation are analyzed. The principal axis direction angle is determined using a convex hull algorithm, the centroid coordinates are calculated, and the asymmetric distortion degree of deviation is analyzed to describe the overall posture. Simultaneously, three indicators, including the target centroid translation degree, are calculated, and the maximum value is taken as the quality risk index to capture instantaneous extreme deformation. Finally, the temporal changes of this index are analyzed, stored over time, and plotted as a baseline. An alarm is triggered when the index exceeds a preset threshold twice consecutively, prompting staff to replace the anchor bolt, forming a dynamic closed-loop management system for effective quality control.

[0044] It should be noted that the anchor plate is a key connecting component between the anchor bolt and the surrounding rock, directly bearing the tensile force transmitted by the anchor bolt and distributing the force to the surrounding rock. When the anchor bolt deforms, loosens, or is damaged under stress, the position, angle, and shape of the plate will change accordingly. By photographing the plate, the stress state and quality changes of the anchor bolt can be indirectly reflected. Compared to directly photographing the anchor bolt body, the plate has a larger area and a relatively fixed position, making it easier to capture clear images, facilitating long-term comparative analysis, and providing a reliable basis for judging the stability of the anchor bolt. The preset frequency is determined according to the geological conditions of the mine, such as once every 30 minutes in high-risk areas and once every 2 hours in low-risk areas, to ensure that gradual changes in the anchor bolt can be captured.

[0045] Specifically, in the initial stage after the anchor bolt installation is completed and the quality is intact, an industrial camera is used to photograph the anchor bolt pad, and the obtained images are smoothed and denoised to obtain a grayscale image of the anchor bolt pad. According to a preset time period and a preset frequency, the anchor bolt pad is continuously photographed at the same location with the same camera and parameters to obtain an image sequence of the anchor bolt pad. All images in the anchor bolt pad image sequence are converted to grayscale to generate a target image sequence composed of multiple target images arranged according to the acquisition time.

[0046] This completes the acquisition of the target image sequence.

[0047] S2: Perform model matching and image mapping on the target image after edge detection to obtain the target edge; extract feature points of the target image, and register the target image according to the feature points, the ratio of the maximum distance and the minimum distance between all feature points and the target edge, and the distance between the feature points and the target edge to generate a registered target image and a registered target image sequence.

[0048] It should be noted that when images of the anchor plate are taken at different times, slight variations in camera position or different environmental conditions may result in some discrepancies in the images, such as inconsistencies in the position and angle of the anchor plate. Directly comparing images cannot accurately determine whether the anchor has deformed; therefore, image registration and stabilization are necessary to ensure that images from different times can be compared under the same standard.

[0049] Specifically, Canny edge detection is performed on the target image. A YOLO model trained with a large number of matching anchor bolt pad images is used to match the anchor bolt pad edge contours. A closure operation is then used to repair the anchor bolt pad edge contours, resulting in closed anchor bolt pad edge contours, denoted as the anchor bolt edge contours. These anchor bolt edge contours are then mapped back to the pixel coordinate system of the target image using coordinate indexing, and the anchor bolt edge contours in the target image are denoted as the target edges. Feature points in the target image are extracted using the SIFT algorithm, and the shortest distance from each feature point to the target edge is calculated, denoted as the target distance. Based on the target distance, the spatial confidence weight of the feature points is calculated, denoted as the feature weights, including:

[0050] It should be noted that during the registration process of the target image, feature points far from the target edge are less affected by edge blurring, shadow interference, and deformation error, and have higher stability. Therefore, they should be given higher weights and given priority in registration. The formula determines the feature weight of the feature points by the distance between the feature points and the target edge.

[0051] The feature weights satisfy the following expression:

[0052] ;

[0053] In the formula, This represents the feature weight of the i-th feature point, which is used to measure the stability weight of this feature point in image registration; This represents the shortest distance from the i-th feature point to the target edge. The larger the distance, the farther the feature point is from the target edge. This represents the maximum distance from all feature points in the target image to the target edge, used to normalize the distance and make the formula fit a single target image; e is a natural constant. middle, The larger the value, the farther the feature point is from the target edge. The larger, The smaller, The smaller, The closer to 1, then Less than 1 and closer to 1, that is The closer a value is to 1, the further away the feature point is from the target edge, and the more suitable it is as a registration reference.

[0054] It should be noted that by repeating the above process on the target image sequence and utilizing the temporal relationship of the multi-frame registration results, the dynamic monitoring requirements of the anchor bolts can be met, the long-term deformation trend of the anchor bolts can be captured, and a data foundation can be provided for subsequent long-term monitoring.

[0055] Preferably, the feature weights of all extracted feature points in the target image are calculated, and these feature points are sorted from high to low according to their feature weights. The top x feature points are selected, and erroneous matching pairs are removed using the RANSAC algorithm. Then, the geometric transformation matrix is ​​calculated to complete the registration, generating a registered target image. The above process is repeated on the target image sequence to obtain a registered target image sequence.

[0056] S3: Locate the target area, calculate the target principal axis direction angle based on the line position connecting the target point pairs; calculate the target centroid coordinates based on the ratio of the x-coordinates and y-coordinates of all pixels in the target area to the total number of pixels in the target area; calculate the target asymmetric distortion degree based on the distance between the target centroid coordinates and the target principal axis line and the total number of target pixels; in the registered target image sequence, calculate the anchor bolt quality risk index based on the changes in the centroid coordinates, principal axis angle, and target asymmetric distortion degree of the target area in all adjacent registered target images.

[0057] It's important to note that in mine support scenarios, the principal axis angle of the anchor pad is a crucial indicator for determining the anchor's stress state. The complex mining environment, with its variations in rock stress and long-term loads, causes anchor deformation, which in turn leads to a synchronous shift in the pad. The change in the pad's principal axis angle visually reveals the anchor's tilting or torsional trend. Using the convex hull algorithm to determine target point pairs essentially aims to capture the most representative extension direction of the pad. In actual mine support systems, anchor forces have a primary transmission path, and the most representative extension direction of the pad is closely related to the anchor's stress axis. By identifying this direction, it's possible to accurately detect whether the anchor is deviating or whether there are quality issues affecting support safety, such as uneven stress distribution.

[0058] Specifically, a coordinate system is constructed in the target image with the horizontal direction to the right as the x-axis and the vertical direction upward as the y-axis, denoted as the target coordinate system. The anchor plate area formed by all pixels on and within the target edge is denoted as the target region. The pixels in the target region form a convex hull set, denoted as the target convex hull. Points on the convex hull are paired, and all vertex pairs of the convex hull are traversed to find the pair of vertices with the farthest distance, denoted as the target point pair. The pixel coordinates of the target point pair are obtained. When the x-coordinates of the target point pair are not equal, the principal axis direction angle of the target region is represented by the pixel coordinates, denoted as the target principal axis direction angle, including:

[0059] The target principal axis direction angle satisfies the following expression:

[0060] ;

[0061] In the formula, The principal axis direction angle of the target region is represented; a and b are the pair of vertices that are farthest apart on the convex hull formed by the target region. , This represents the ordinate of the target point relative to the midpoints a and b; , This represents the x-coordinate of the target point relative to the midpoints a and b. The x-coordinates of a and b are not equal. If they are equal, the principal axis direction angle of the target is 90°. Represents the absolute value function; This represents the arctangent function.

[0062] In the formula, It represents the vertical projection length of the line segment formed by the target point; It represents the horizontal projection length of the line segment formed by the target point; This represents constructing a right triangle with its longest diagonal as the hypotenuse, where the lengths of the two legs are respectively... and Main axis orientation angle It is the acute angle formed by the hypotenuse and the horizontal leg in the right triangle.

[0063] At this point, the principal axis direction angle of the target area has been obtained.

[0064] It should be noted that after determining the principal axis direction angle of the pad, further calculation of its center of mass coordinates is of significant supplementary importance. The mining environment is complex, and the principal axis direction alone is sometimes insufficient to fully reflect the stress state. The center of mass is the point where the mass of an object is concentrated, while the principal axis direction is the extension direction of the geometric shape. Theoretically, for a symmetrical object with uniform mass, the center of mass should lie on the principal axis. Therefore, by calculating the center of mass coordinates and analyzing their positional relationship with the determined principal axis, the non-uniformity of the pad's mass distribution can be quantified, revealing a more refined asymmetric deformation caused by localized stress concentration, thus enabling more comprehensive quality control.

[0065] Specifically, the number of pixels in the same target area is counted and recorded as the total number of target pixels; the sum of the x-coordinates of pixels in all target areas is counted and recorded as the total x-coordinates of target pixels; the sum of the y-coordinates of pixels in all target areas is counted and recorded as the total y-coordinates of target pixels; and the ratios between the total number of target pixels and the sum of the x-coordinates and the sum of the y-coordinates of target pixels are recorded as the centroid coordinates of the target.

[0066] At this point, the target centroid coordinates of the target region have been obtained.

[0067] It should be noted that, in the above steps, this invention obtained the principal axis direction angle used to describe the overall attitude of the anchor plate, and the target centroid coordinates used to describe the center of mass of the anchor plate. The degree of deviation between the principal axis direction angle and the target centroid coordinates is a key indicator for measuring whether the plate has undergone asymmetric deformation. Therefore, this step aims to perform precise quantitative calculation of this deviation to obtain an asymmetric distortion degree that can reflect the uneven local stress in the target area.

[0068] Preferably, a straight line is fitted based on the coordinates of the target point pair, and this straight line is denoted as the target principal axis line; the perpendicular distance from the target centroid coordinates to the target principal axis line is calculated, the total number of pixels in the target region is calculated and denoted as the target region area, and the asymmetric distortion degree of the target region is represented by the perpendicular distance from the target centroid coordinates to the target principal axis line and the target region area, denoted as the target asymmetric distortion degree, including:

[0069] ;

[0070] In the formula, Indicates the asymmetric distortion degree of the target; The x-coordinate and y-coordinate of the target centroid represent the target centroid; N represents the total number of target pixels, used to correlate the centroid offset distance with the area of ​​the anchor plate; A, B, and C represent the linear equation coefficients of the target principal axis line. Represents the absolute value function; This represents the normalization function.

[0071] In the formula, This represents the distance from the target's centroid to the line connecting the target's principal axes. When the value is 0, the centroid lies on the target principal axis, indicating that the target area is completely symmetrical and has no distortion. In the case of a small anchor plate, even if the centroid offset is small, the denominator will be small, and the final D value may be large. For a small plate, the same offset may mean more severe deformation or greater risk. Conversely, when the anchor plate area is large, the same centroid offset results in a larger denominator, and the final D value will be relatively smaller. The larger the value, the greater the asymmetric distortion of the target.

[0072] It should be noted that traditional state assessment methods based on comparing current and baseline images cannot detect reciprocating deformations that occur within the monitoring period but eventually return to their original state. Such reciprocating deformations, such as those caused by blasting impacts or periodic mining pressures, are key factors leading to metal fatigue and loosening of supports. Therefore, this step constructs an assessment model for maximum historical offset to capture the instantaneous extreme deformations that may occur within the monitoring period and pose the greatest threat to structural safety. This model aims to identify and record the historical peak values ​​of various deformation indices relative to the initial state. Even if subsequent deformation recovers, this peak value may indicate that the product has experienced stress exceeding its elastic limit, potentially resulting in plastic deformation or permanent damage.

[0073] Specifically, the centroid coordinates of all target regions in the registered target image sequence are calculated to form a target centroid coordinate sequence; based on the translation of the target centroid coordinates in the target centroid coordinate sequence, the degree of translation of the target centroid coordinates of the target region within a preset time period is calculated and denoted as the target centroid translation degree, including:

[0074] The degree of translation of the target centroid satisfies the following expression:

[0075] ;

[0076] In the formula, This indicates the degree of centroid translation in the i-th registered target image; , Indicates the x and y coordinates of the centroid of the target in the i-th registered target image; , Indicates the x and y coordinates of the centroid of the target in the (i-1)th registered target image; Represents the absolute value function; Represents the maximum value function; This represents the normalization function.

[0077] In the formula, The Euclidean distance between the centroid of the target in the i-th registered target image and the centroid of the target in the (i-1)-th registered target image represents the distance the centroid of the target moves relative to the previous time point at the i-th time point. This represents the maximum translation distance among all adjacent registered target images in the registered target image sequence; The larger the value, the greater the instantaneous impact force on the anchor bolt, and the higher the risk of loosening or shear failure.

[0078] It should be noted that the change in the principal axis direction angle of the anchor plate is a direct manifestation of the anchor bolt's tilting, torsion, or attitude instability caused by uneven external forces. To quantify this change, this step calculates the maximum rate of change of the principal axis direction angle between adjacent registered target images throughout the entire monitoring period, which is taken as the degree of change of the target principal axis.

[0079] Preferably, the principal axis orientation angles of all target regions in the registered target image sequence are calculated to form a principal axis orientation angle sequence; based on the magnitude of the change in the principal axis orientation angles in the principal axis orientation angle sequence, the degree of change of the principal axis orientation angles of the target region within a preset time period is calculated and denoted as the degree of change of the principal axis, including:

[0080] The degree of change of the target principal axis satisfies the following expression:

[0081] ;

[0082] In the formula, This indicates the degree of change in the principal axis of the target in the i-th registered target image; , The principal axis orientation angles of the i-th and (i-1)-th registered target images are represented. Represents the absolute value function; Represents the maximum value function; This represents the normalization function.

[0083] In the formula, This indicates the change in the principal axis direction angle between adjacent registered target images; This represents the maximum value of the change in the principal axis orientation angle among all adjacent registered target images in the registered target image sequence; This indicates the fastest deformation that occurs in the anchor plate within a preset time period. The larger the value, the more likely the anchor may have been subjected to tilting, torsion, or uneven impact, and the greater the degree of change in its target principal axis.

[0084] It should be noted that the rapid change in the asymmetric distortion of the anchor plate reflects a drastic adjustment in its internal stress distribution, which may be a precursor to material yielding or microcrack propagation. To capture this most dangerous internal state change, this step calculates the maximum rate of change of asymmetric distortion between adjacent registered target images within a preset time period, i.e., the asymmetric rate of change.

[0085] Preferably, the target asymmetric distortion degree composed of all target regions in the registered target image sequence is calculated to form a target asymmetric distortion degree sequence; based on the rate of change of the target asymmetric distortion degree in the target asymmetric distortion degree sequence, the target asymmetric distortion rate of the target region within a preset time period is calculated and denoted as the target asymmetric speed change, including:

[0086] The target asymmetric speed change satisfies the following expression:

[0087] ;

[0088] In the formula, This indicates that the target in the i-th registered target image undergoes asymmetric velocity variation; , This represents the degree of asymmetric distortion of the target region in the i-th and (i-1)-th registered target images; Represents the absolute value function; Represents the maximum value function; This represents the normalization function.

[0089] In the formula, This represents the change in asymmetric distortion of the target region between adjacent registered target images; This represents the maximum value of the change in asymmetric distortion among all adjacent registered target images in the registered target image sequence; This indicates the most drastic change in internal stress within the anchor plate within a preset time period. The larger the value, the more likely the plate material is undergoing rapid plastic deformation, and the greater the risk to its structural integrity.

[0090] It should be noted that the anchor bolt quality risk index is defined using a maximum value function, which is based on the characteristic that a single risk dimension failure in mining support scenarios can lead to safety hazards. Since the three types of deformation in anchor bolts—translation, principal axis deflection, and asymmetric distortion—correspond to different failure modes (translation reflects loosening, principal axis change reflects tilting, and asymmetric distortion reflects localized stress concentration), any drastic change in any type of deformation can independently constitute a support risk.

[0091] Preferably, the anchor bolt quality risk index satisfies the following expression:

[0092] ;

[0093] In the formula, , , This represents the degree of centroid translation, principal axis change, and asymmetric velocity change of the target in the i-th registered target image. Represents the maximum value function; This means that the anchor bolt quality risk index of the system is directly defined as the highest level among all quantities, so that the anchor bolt quality risk index meets the requirement of having the highest sensitivity to any single-dimensional sudden risk. Even if the translational and rotational impact of the anchor bolt pad is small, as long as there is a violent asymmetric speed change inside, This will immediately reflect the high risk, thus avoiding the problem of the risk being averaged out by other normal indicators, and ensuring the timeliness and effectiveness of the early warning.

[0094] S4: Draw the anchor bolt quality risk baseline based on the anchor bolt quality risk index within a preset time period, analyze it according to the preset threshold, trigger the system to issue an alarm and prompt the anchor bolt to be replaced.

[0095] It should be noted that this step, through analysis of the time-series changes in the anchor bolt quality risk index, captures persistent risks that may be masked by a single fluctuation, avoids misjudgments caused by accidental factors, and ensures timely response to real safety hazards, thereby achieving dynamic closed-loop management of the anchor bolt status.

[0096] Specifically, all anchor bolt quality risk indices calculated within a preset time period are organized in chronological order. The time information corresponding to each index is associated with the anchor bolt quality risk index to form a dataset containing the time-risk index correspondence, which is then stored in the system's memory. A two-dimensional coordinate system is established with time as the horizontal axis and the anchor bolt quality risk index as the vertical axis. The risk index corresponding to each time point is marked as a data point in the two-dimensional coordinate system. Adjacent data points are connected sequentially by line segments to form a continuous curve, which is denoted as the anchor bolt quality risk baseline.

[0097] Preferably, a first threshold is preset. If the anchor bolt quality risk baseline exceeds the first threshold twice consecutively at a certain moment, the system marks this position and sends an alarm command to the staff, who then replace the anchor bolt. It should be noted that the first threshold is set by the implementers according to the actual implementation situation; for example, the first threshold can be set to 0.75.

[0098] This completes the quality control of mine support products.

[0099] This invention also discloses a quality control system for mine support products, including a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement a quality control method for mine support products according to the present invention.

[0100] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0101] While this specification has shown and described numerous embodiments of the invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will occur to those skilled in the art without departing from the spirit and essence of the invention. It should be understood that various alternatives to the embodiments of the invention described herein may be employed in the practice of this invention.

Claims

1. A method for quality control of mine support products, characterized in that, include: The anchor plate images are continuously acquired at a preset time and preprocessed. The resulting target images are then arranged by time to generate a target image sequence. The target image after edge detection is subjected to model matching and image mapping to obtain the target edge; Extract feature points from the target image, and calculate feature weights based on the feature points, the ratio of the maximum distance to the target edge to the maximum distance to all feature points, and the ratio of the minimum distance to the target edge. High-weight feature points are selected for registration, generating the registered target image and the sequence of registered target images; The target region is located according to the target edge. Based on the position of the line connecting the target point pairs with the largest distance in the target convex hull formed by the target region, the ratio of the x-coordinate and y-coordinate of all pixels in the target region to the total number of pixels in the target region, the target principal axis direction angle and the target centroid coordinates are calculated. Based on the vertical distance between the target centroid coordinates and the target principal axis line formed by connecting the target point pairs, and the total number of pixels in the target region, the target asymmetric distortion degree is calculated. In the registered target image sequence, the anchor bolt quality risk index is calculated based on the maximum value of the centroid coordinate distance of the target region, the change in the target principal axis direction angle, and the change in the target asymmetric distortion degree in all adjacent registered target images within a preset time period. An anchor quality risk baseline is drawn based on the anchor quality risk index within a preset time period. The analysis is performed based on a preset threshold, triggering the system to issue an alarm and prompt for anchor replacement. The feature weights satisfy the following expression: In the formula, This represents the feature weight of the i-th feature point; This represents the shortest distance from the i-th feature point to the target edge; This represents the maximum distance from all feature points in the target image to the target edge; e is the natural constant. The target principal axis direction angle satisfies the following expression: In the formula, The principal axis direction angle of the target region is represented; a and b are the pair of vertices that are farthest apart on the convex hull formed by the target region. , This represents the ordinate of the target point relative to the midpoints a and b; , This represents the x-coordinate of the target point relative to the midpoints a and b. The x-coordinates of a and b are not equal. If they are equal, the principal axis direction angle of the target is 90°. Represents the absolute value function; Represents the arctangent function; Calculating the target centroid coordinates includes: counting the number of pixels in the same target region, denoted as the total number of target pixels; counting the sum of the x-coordinates of pixels in all target regions, denoted as the total x-coordinates of target pixels; counting the sum of the y-coordinates of pixels in all target regions, denoted as the total y-coordinates of target pixels; and denoting the ratios between the total number of target pixels and the sum of the x-coordinates of target pixels, and the sum of the y-coordinates of target pixels, respectively, as the target centroid coordinates. The target asymmetric distortion degree satisfies the following expression: In the formula, Indicates the asymmetric distortion degree of the target; The x-coordinate and y-coordinate of the target centroid represent the target centroid; N represents the total number of target pixels, used to correlate the centroid offset distance with the area of ​​the anchor plate; A, B, and C represent the linear equation coefficients of the target principal axis line. Represents the normalization function; The anchor bolt quality risk index is calculated by: calculating the target centroid translation degree, target principal axis change degree, and target asymmetric speed change based on the maximum value of the Euclidean distance between the target centroids of all adjacent registered target images in the registered target image sequence, the maximum value of the principal axis direction angle change, and the maximum value of the asymmetric distortion degree change. The anchor bolt quality risk index satisfies the following expression: In the formula, , , This represents the degree of centroid translation, principal axis change, and asymmetric velocity change of the target in the i-th registered target image. Represents the maximum value function; A first threshold is preset. If the anchor bolt quality risk baseline exceeds the first threshold twice in a row at a certain moment, the position at that moment is marked, and the system sends an alarm command to the staff, who then replace the anchor bolt.

2. The method for quality control of mine support products according to claim 1, characterized in that, The generation of the target image sequence includes: In the initial stage after the anchor bolt installation is completed and the quality is intact, an industrial camera is used to photograph the anchor bolt pad, and the obtained images are smoothed and denoised to obtain a grayscale image of the anchor bolt pad. According to a preset time period and a preset frequency, the anchor bolt pad is continuously photographed at the same location with the same camera and parameters to obtain an image sequence of the anchor bolt pad. All images in the anchor bolt pad image sequence are converted to grayscale to generate a target image sequence composed of multiple target images arranged according to the acquisition time.

3. The method for quality control of mine support products according to claim 1, characterized in that, The generation of the registration target image and the registration target image sequence includes: Calculate the feature weights of all extracted feature points in the target image, sort these feature points from high to low according to their feature weights, select the top x feature points, use the RANSAC algorithm to remove erroneous matching pairs, and then calculate the geometric transformation matrix to complete the registration and generate the registered target image; repeat the above process on the target image sequence to obtain the registered target image sequence.

4. The method for quality control of mine support products according to claim 1, characterized in that, The drawing of the anchor bolt quality risk baseline includes: All anchor bolt quality risk indices calculated within a preset time period are organized in chronological order. The time information corresponding to each index is associated with the anchor bolt quality risk index to form a dataset containing the correspondence between time and risk index, which is stored in the system's memory. A two-dimensional coordinate system is established with time as the horizontal axis and anchor bolt quality risk index as the vertical axis. The risk index corresponding to each time point is marked as a data point in the two-dimensional coordinate system. Adjacent data points are connected sequentially by line segments to form a continuous curve, which is recorded as the anchor bolt quality risk baseline.

5. A quality control system for mine support products, characterized in that, include: A processor and a memory, wherein the memory stores computer program instructions that, when executed by the processor, implement a method for quality control of mine support products according to any one of claims 1-4.

Images