A bolt pre-tightening force off-line detection method and system

Abstract

The application provides a bolt pre-tightening force offline detection method and system, and the specific steps of the method include the following: when a bolt to be detected meets installation conditions, spraying a fluorescent marking line on a joint surface of a nut and a connecting piece and collecting a reference marking line image, and projecting a fluorescent marking point on a bolt head end surface and collecting a reference light point image; calculating reference displacement features and reference deformation features according to the images respectively; during detection, synchronously collecting a current marking line image and a current light point image of the bolt to be detected, analyzing changes of current displacement features and current deformation features, and comprehensively judging a loosening risk of the bolt to be detected. The application can complete extraction and analysis of macro displacement and micro deformation double features at one time, accurately locate the bolt with a loosening risk, and realize daily patrol of a large number of bolts in an offline state.

Description

Technical Field

[0001] This invention relates to the field of bolt preload detection technology, specifically to an offline bolt preload detection method and system. Background Technology

[0002] The preload of bolts is directly related to the reliability and safety of the entire connection structure. During use, the preload of bolts will decrease due to factors such as vibration, load fluctuation and material deformation. Therefore, it is crucial to detect the preload of bolts efficiently and reliably. Common methods for detecting bolt preload are generally online monitoring with high-precision sensors or manual inspection, which generally suffer from problems such as high cost, low efficiency and long cycle.

[0003] In the prior art, CN118196058A discloses an image-based bolt loosening detection method. It fits the circumcircle of the bolt head based on the image of the bolt head, searches for the two intersection points of the inner and outer regions of the circumcircle with the marker line, and uses the central angle corresponding to the arc between the two intersection points to determine whether the bolt is loose. This scheme requires a lot of calculations in the implementation process, is very cumbersome, and relies on a single feature for recognition, which cannot eliminate interference factors. Therefore, the prior art lacks a solution that can perform efficient detection in an offline state. Summary of the Invention

[0004] The purpose of this invention is to provide an offline method and system for detecting bolt preload, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] An offline method for detecting bolt preload includes the following steps:

[0007] Step 1: When the installation conditions are met, take the mating surface between the nut and the connector on the bolt to be tested as the marking area, spray a fluorescent marking line across the mating surface in the marking area, and collect the reference marking line image of the bolt to be tested in the marking area. When the installation conditions are met, take the head end face of the bolt to be tested as the projection area, and project multiple fluorescent marking points in the projection area based on preset parameters, and collect the reference light point image of the bolt to be tested in the projection area.

[0008] Step 2: Extract the position information of the luminescent pixels from the reference mark line image, calculate the centroid of the fluorescent mark line, use the abscissa of the centroid as the reference displacement feature of the bolt to be tested, and extract the position information of each light spot from the reference light spot image, calculate the moment of inertia of all light spots relative to their centroid, and use the moment of inertia as the reference deformation feature of the bolt to be tested.

[0009] Step 3: During detection, the current marking line image of the bolt to be detected is acquired in the marking area, the position information of the luminescent pixels is extracted from the current marking line image, the centroid of the fluorescent marking line is calculated, and the abscissa of the centroid is used as the current displacement feature of the bolt to be detected. In addition, the current light spot image of the bolt to be detected is acquired in the projection area, the position information of each light spot is extracted from the current light spot image, and the moment of inertia of all light spots relative to their centroid is calculated. The moment of inertia is used as the current deformation feature of the bolt to be detected.

[0010] Step 4: Determine the displacement state of the bolt to be tested based on the relative displacement deviation between the current displacement feature and the reference displacement feature; determine the deformation state of the bolt to be tested based on the relative deformation deviation between the current deformation feature and the reference deformation feature; and comprehensively judge the risk of loosening of the bolt based on the displacement state and deformation state of the bolt to be tested.

[0011] Furthermore, the fluorescent marking line is parallel to the axis of the bolt to be tested in space, and the spatial projection line on the plane where the bolt head end face is located coincides with the midline of the bolt head end face. Under specific light source excitation conditions, a reference marking line image of the bolt to be tested is acquired in the marking area. A plane rectangular coordinate system is established with the bolt head end face as the projection area, the midpoint of the head end face as the origin, and the spatial projection line of the bolt's fluorescent marking line on this plane as the vertical axis. Fluorescent marking points are projected in the projection area. Under specific light source excitation conditions, a reference light spot image of the bolt to be tested is acquired in the projection area.

[0012] Furthermore, the logic for extracting the position information of the luminescent pixels and calculating the centroid of the fluorescent marker line to obtain the displacement features of the bolt to be detected is as follows:

[0013] A fixed-size image is acquired for the marked area, with the vertical axis of the image parallel to the extension direction of the fluorescent marker line, and the fluorescent marker line positioned in the center of the image. The image is converted to grayscale and noise reduction is performed. All pixels in the grayscale image are binarized according to a set segmentation threshold. If the grayscale value is less than the segmentation threshold, it is determined to be a background pixel; otherwise, it is determined to be a foreground pixel. The coordinates of all foreground pixels are collected to form a set of luminous pixels. The abscissa of the geometric center point of this set is calculated by summing the abscissas of all pixels in the luminous pixel set and dividing the result by the total number of luminous pixels to obtain the abscissa of the geometric center point of the set. The value of this abscissa is used as the displacement feature.

[0014] Furthermore, the logic for extracting the position information of each light spot from the light spot image is as follows:

[0015] A fixed-size image is acquired for the projection area, with the imaging direction consistent with the coordinate system of the bolt head end face, and the bolt head end face is placed in the center of the image; the image is converted into a grayscale image and noise reduction is performed; all pixels in the grayscale image are binarized according to the set segmentation threshold; if the grayscale value is less than the segmentation threshold, it is determined to be a background pixel; otherwise, it is determined to be a foreground pixel, and a binary image is generated.

[0016] The specific steps for scanning the binary image and dividing it into pixel sets for each light point are as follows:

[0017] Define a label image with the same size as the binary image and one-to-one pixel correspondence, with all pixel values ​​initialized to 0; define a neighborhood set containing the pixel values ​​of the four positions above, below, above right, and to the left of the current pixel, and mark any position as empty if it does not exist; define a counter with an initial value of 1, and assign the current value of the counter to the target pixel each time a new label is assigned, and then increment the counter value by 1.

[0018] Starting from the top left corner of the binary image, each pixel is read sequentially from left to right and top to bottom. If the current pixel is background, it is skipped, and the next pixel is read. If the current pixel is foreground, its neighborhood set is read. If its neighborhood set is all background or empty, a new label is assigned to the corresponding pixel in the label image. If its neighborhood set contains foreground elements, the neighborhood set of the corresponding pixel is read in the label image, and the smallest non-zero label in the read neighborhood set is assigned to the pixel in the label image. All non-zero distinct labels in the neighborhood set of the pixel in the label image are recorded as equivalent.

[0019] After scanning, all labels with equivalent relationships are grouped into a label set, and the label with the smallest value in each label set is selected as the root label of the label set. Scan the label image, and for each non-zero pixel, replace its pixel value with the root label of its label set. The coordinates of pixels with the same root label form a light spot pixel set, which is the position information of each light spot in the light spot image.

[0020] Furthermore, based on the position information of each light spot, the moment of inertia of all light spots relative to their centroids is calculated, and the logic for obtaining the deformation characteristics of the bolt to be detected is as follows:

[0021] For each set of light point pixels, calculate the coordinates of the center point of the light point based on the pixels it contains. Sum the x-coordinates of all the pixels in the set and divide the result by the total number of pixels in the set to obtain the x-coordinate of the center point of the light point pixel set. Sum the y-coordinates of all the pixels in the set and divide the result by the total number of pixels in the set to obtain the y-coordinate of the center point of the light point pixel set.

[0022] Similarly, the coordinates of the center point of each light spot in the light spot image are obtained. From this, the centroid of all light spots is calculated. The sum of the abscissas of the center points of all light spots is divided by the total number of light spots to obtain the abscissa of the centroid of the light spot image. The sum of the ordinates of the center points of all light spots is divided by the total number of light spots to obtain the ordinate of the centroid of the light spot image.

[0023] Based on the above results, calculate the moment of inertia of all light spots in the spot image relative to their centroids. For the center point of each light spot, square the difference between its x-coordinate and the x-coordinate of the centroid, and the difference between its y-coordinate and the y-coordinate of the centroid. Add the two squared values ​​to obtain the squared distance from the center point to the centroid. Sum the squared distances from all center points to the centroid to obtain the moment of inertia of the spot image. Use this moment of inertia as the deformation feature.

[0024] Furthermore, the relative displacement deviation is calculated as follows: calculate the difference between the current displacement feature and the reference displacement feature, divide the absolute value of the difference by the length of the fluorescent marker line in the reference marker line image, and then multiply by 100%. The result is the relative displacement deviation between the current displacement feature and the reference displacement feature. If the relative displacement deviation is less than or equal to the first displacement threshold, it is determined to be tight. If the relative displacement deviation is greater than the first displacement threshold and less than or equal to the second displacement threshold, it is determined to be slightly moving. If the relative displacement deviation is greater than the second displacement threshold, it is determined to be loose.

[0025] Furthermore, the relative deformation deviation is calculated as follows: calculate the difference between the current deformation feature and the reference deformation feature, divide the absolute value of the difference by the reference deformation feature, and then multiply by 100%. The result is the relative deformation deviation between the current deformation feature and the reference deformation feature. If the relative deformation deviation is less than or equal to the first deformation threshold, it is judged as normal. If the relative deformation deviation is greater than the first deformation threshold and less than or equal to the second deformation threshold, it is judged as a warning. If the relative deformation deviation is greater than the second deformation threshold, it is judged as abnormal.

[0026] Furthermore, based on the displacement and deformation states of the bolt to be tested, the risk of loosening of the bolt is comprehensively assessed, specifically as follows:

[0027] A loosening risk index and a risk coupling coefficient are defined. The loosening risk index is calculated by multiplying the risk coupling coefficient by the relative displacement deviation and the relative deformation deviation, and then summing the product with the relative displacement deviation and the relative deformation deviation. The result is the loosening risk index. The risk level of the bolt to be tested is determined based on the loosening risk index: if the loosening risk index is less than or equal to the first risk threshold, it is determined to be low risk; if the loosening risk index is greater than the first risk threshold and less than or equal to the second risk threshold, it is determined to be medium risk; if the loosening risk index is greater than the second risk threshold, it is determined to be high risk, and the system will automatically alarm.

[0028] The present invention also provides an offline bolt preload detection system, which is used to perform the above-described offline bolt preload detection method, including:

[0029] The marking creation module is used to, when the installation conditions are met, use the mating surface between the nut and the connector on the bolt to be tested as the marking area, spray a fluorescent marking line across the mating surface in the marking area, and collect a reference marking line image of the bolt to be tested in the marking area. Also, when the installation conditions are met, use the head end face of the bolt to be tested as the projection area, and project multiple fluorescent marking points in the projection area based on preset parameters, and collect a reference light spot image of the bolt to be tested in the projection area.

[0030] The reference feature extraction module is used to extract the position information of the luminescent pixels from the reference mark line image, calculate the centroid of the fluorescent mark line, use the abscissa of the centroid as the reference displacement feature of the bolt to be tested, and extract the position information of each light spot from the reference light spot image, calculate the moment of inertia of all light spots relative to their centroid, and use the moment of inertia as the reference deformation feature of the bolt to be tested.

[0031] The detection feature extraction module is used to acquire the current marking line image of the bolt to be detected in the marking area during detection, extract the position information of the luminescent pixels from the current marking line image, calculate the centroid of the fluorescent marking line, and use the abscissa of the centroid as the current displacement feature of the bolt to be detected. It also acquires the current light spot image of the bolt to be detected in the projection area, extracts the position information of each light spot from the current light spot image, calculates the moment of inertia of all light spots relative to their centroid, and uses the moment of inertia as the current deformation feature of the bolt to be detected.

[0032] The risk assessment module is used to determine the displacement state of the bolt to be tested based on the relative displacement deviation between the current displacement feature and the reference displacement feature, to determine the deformation state of the bolt to be tested based on the relative deformation deviation between the current deformation feature and the reference deformation feature, and to comprehensively assess the loosening risk of the bolt based on the displacement state and deformation state of the bolt to be tested.

[0033] Compared with the prior art, the beneficial effects of the present invention are:

[0034] It can draw fluorescent marking lines and fluorescent marking points in one go, and simultaneously collect and analyze images of both features offline, enabling daily inspection of batch bolts. The detection efficiency is significantly improved compared to online monitoring with high-precision sensors or manual inspection.

[0035] By integrating the dual characteristics of microscopic deformation and macroscopic displacement for comprehensive decision-making, the system can accurately locate "key suspect" targets with a risk of loosening, thereby making the subsequent manual review work more targeted and significantly reducing the number of bolts that need to be measured one by one. Moreover, the mechanism of mutual verification of dual characteristics can overcome the defects of single feature detection that are easily affected by environmental interference, enabling the system to perform accurate detection even in complex environments. Attached Figure Description

[0036] Figure 1 This is a schematic diagram of the overall method flow of the present invention;

[0037] Figure 2 This is a three-dimensional schematic diagram of the fluorescent label provided in an embodiment of the present invention;

[0038] Figure 3 This is a spot image of the bolt head end face provided in an embodiment of the present invention;

[0039] Figure 4 This is a schematic diagram of displacement feature calculation provided in an embodiment of the present invention;

[0040] Figure 5 This is a schematic diagram of deformation feature calculation provided in an embodiment of the present invention;

[0041] Figure 6 The displacement relative deviation-loosening risk index fitting curve provided in the embodiments of the present invention;

[0042] Figure 7 This is a fitting curve of relative deformation deviation and loosening risk index provided in an embodiment of the present invention;

[0043] Figure 8 This is a structural block diagram of the overall system of the present invention. Detailed Implementation

[0044] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0045] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0046] Example:

[0047] Please see Figures 1 to 7 The present invention provides a technical solution:

[0048] An offline method for detecting bolt preload includes the following steps:

[0049] Step 1: When the installation conditions are met, take the mating surface between the nut and the connector on the bolt to be tested as the marking area, spray a fluorescent marking line across the mating surface in the marking area, and collect the reference marking line image of the bolt to be tested in the marking area. When the installation conditions are met, take the head end face of the bolt to be tested as the projection area, and project multiple fluorescent marking points in the projection area based on preset parameters, and collect the reference light spot image of the bolt to be tested in the projection area.

[0050] Furthermore, the fluorescent marking line is parallel to the axis of the bolt to be tested in space, and the spatial projection line on the plane where the bolt head end face is located coincides with the midline of the bolt head end face. Under specific light source excitation conditions, a reference marking line image of the bolt to be tested is acquired in the marking area. A plane rectangular coordinate system is established with the bolt head end face as the projection area, the midpoint of the head end face as the origin, and the spatial projection line of the bolt's fluorescent marking line on this plane as the vertical axis. The number of fluorescent marking points projected is a random integer from 3 to 15. All marking points are randomly distributed in the projection area, and the Euclidean distance between any two marking points is not less than 5% of the side length of the projection area. Under specific light source excitation conditions, a reference light spot image of the bolt to be tested is acquired in the projection area.

[0051] The goal of this step is to establish fluorescent marking lines and dots on the bolt to be inspected. This step must be performed after the bolt is initially installed and a specified preload is applied. The marking area is defined as the mating surface between the nut and the connector on the bolt. This area is the most vulnerable physical interface in a bolted connection structure to loosening. Preload decay directly leads to rotation at this interface. Therefore, spraying marking lines onto the mating surface between the nut and the connector converts subtle, difficult-to-observe mechanical changes into quantifiable visual displacement characteristics.

[0052] When spraying the marking lines, it is stipulated that the marking lines are parallel to the bolt axis in three-dimensional space. This ensures that the extension direction of the marking lines is consistent with the main force direction of the bolt, and ensures that when loosening occurs, the displacement vector of the marking lines in the imaging plane can most directly reflect the relative displacement in the axial direction, thereby establishing a linear mapping relationship between the displacement measurement value and the physical loosening amount.

[0053] Furthermore, this embodiment employs a dual-feature fusion decision-making method to determine the bolt preload. Besides the displacement features established based on the fluorescent marker lines, it also needs to consider the subsequent operation of establishing deformation features based on the fluorescent marker points. However, a long-standing technical challenge lies in the fact that the bolt head end face is typically a rotationally symmetric figure. This geometric characteristic results in the lack of an absolute, inherent angular reference. However, the imaging orientation during subsequent detection must be absolutely consistent with the projection orientation when establishing the reference. Otherwise, any minute angular deviation will cause the acquired light spot image to rotate as a whole. The system will incorrectly identify this rotational change as tensile or shear deformation of the bolt itself, thus calculating a distorted moment of inertia and leading to a catastrophic misjudgment.

[0054] To address this issue, this embodiment introduces a pre-sprayed fluorescent marking line as a global reference. The core technical principle is that by selecting the spraying position of the fluorescent marking line, its spatial projection line on the plane where the bolt head end face is located is forced to coincide with the geometric center line of the head end face, thereby artificially defining an absolute and unique angular reference direction, namely the projection line direction of the marking line.

[0055] Refer to the relative positions of the fluorescent marker lines and fluorescent marker dots. Figure 2 As shown, this ensures that, regardless of the baseline establishment phase or any subsequent detection, the coordinate system orientation used when projecting the marker points and the image orientation used when acquiring the light spot image and calculating the moment of inertia are both consistent with this unique and unchanging baseline. Therefore, even if there are installation angle errors between different detections, the system can automatically align and correct the images based on this unified baseline, thereby fundamentally eliminating the error in moment of inertia calculation caused by viewing angle rotation and ensuring the high reliability and accuracy of the deformation state determination results.

[0056] Furthermore, the choice to project fluorescent markers onto the bolt head face is based on a comprehensive consideration of profound mechanical principles and practical engineering needs. The bolt head plays a crucial role in the connection system by transmitting clamping force to the connected components. When the preload changes, the axial stress generated in the bolt shank is directly and effectively transmitted to the bolt head. In addition, the head face is typically a machined, flat surface, providing a stable and repeatable two-dimensional reference plane for optical projection and imaging, avoiding image distortion and measurement errors caused by curved or irregular surfaces. Therefore, by monitoring the change in the moment of inertia of the marker pattern on the head face, the preload state can be effectively assessed.

[0057] As mentioned earlier, using the spatial projection line of the fluorescent marking line on the plane of the bolt head end face as the longitudinal baseline is equivalent to artificially defining an absolute and unique angular reference direction, ensuring that the projected marking points and the acquired images are based on the same direction and angle. Based on this, a Cartesian coordinate system is established on the bolt head end face, with the midpoint of the head end face as the origin and the spatial projection line of the bolt's fluorescent marking line on this plane as the longitudinal axis. Fluorescent marking points are then projected within this area.

[0058] The number of fluorescent markers projected is a random integer from 3 to 15. When the number of projections is 2, the centroid of the marker is equivalent to the midpoint of the line connecting the two points. Any tiny error at any point will cause the centroid to shift, making the calculated moment of inertia extremely unstable and lacking in anti-interference capability. Only when the number of markers is 3 or more does calculating the moment of inertia through the centroid of the markers have practical significance. However, since the effective area of ​​the bolt head end face is limited, too many markers will lead to an excessively small average distance between the points, making them prone to sticking together in the image, increasing the difficulty and error of light spot segmentation and recognition. Moreover, when the number of markers exceeds a certain limit, adding more points contributes negligibly to improving the deformation detection sensitivity, but significantly increases the computational burden of image processing. Therefore, setting a moderate upper limit can ensure that even when individual markers fail due to contamination or wear, there are still a sufficient number of effective points for accurate calculation. The range of 3-15 satisfies the minimum requirement for stable solution of the mathematical model while avoiding resource waste and recognition interference, making it the preferred range for achieving reliable and efficient detection.

[0059] Furthermore, the Euclidean distance between any two marker points must be controlled to be no less than 5% of the side length of the projection area. This design aims to: ensure the distinguishability of the marker points and prevent markers that are too close together from sticking together during imaging, thus being judged as a single light spot during image recognition and causing errors in the moment of inertia calculation; additionally, controlling the minimum spacing optimizes the spatial distribution, ensuring that the marker points are uniformly distributed on the plane of the bolt head end face, thereby calculating the moment of inertia that best represents the bolt deformation state. The formula for calculating the Euclidean distance between two marker points is: ;

[0060] in, These are the coordinates of the i-th marker point in a Cartesian coordinate system. The coordinates of the j-th marker point in the same coordinate system.

[0061] The 5% of the side length of the projection area is an optimal empirical value based on a comprehensive consideration of image processing principles, engineering practices, and algorithm robustness.

[0062] After determining the spraying position and angle based on the above standards, the reflective properties of the fluorescent pigment are utilized to illuminate the marking and projection areas with ultraviolet light of specific wavelengths as excitation sources. This light source possesses high monochromaticity and directivity, and its core output band can precisely match the absorption peak of the fluorescent material used, exciting the fluorescent material to emit reflected light. Furthermore, a bandpass optical filter is used for imaging. The central transmission band of this filter is precisely set to coincide with the peak of the fluorescence emission spectrum, while its cutoff range is designed to almost completely block the excitation light band and other ambient stray light. This design constitutes a highly efficient spectral filtering system, primarily receiving the characteristic light signal emitted by the fluorescent marker. Ambient light from sunlight, factory lighting, etc., cannot be received because its spectrum is strongly suppressed by the filter. This results in only the fluorescent marker area exhibiting high brightness in the acquired image under complex natural or industrial lighting environments, while the background remains almost pure black, thus achieving an extremely high signal-to-noise ratio.

[0063] Step 2: Extract the position information of the luminescent pixels from the reference mark line image, calculate the centroid of the fluorescent mark line, use the abscissa of the centroid as the reference displacement feature of the bolt to be tested, extract the position information of each light spot from the reference light spot image, calculate the moment of inertia of all light spots relative to their centroid, and use the moment of inertia as the reference deformation feature of the bolt to be tested.

[0064] Furthermore, the logic for extracting the position information of the luminescent pixels from the marker line image and calculating the centroid of the fluorescent marker line to obtain the displacement features of the bolt to be detected is as follows:

[0065] When acquiring the image of the marker line, ensure that the vertical baseline of the image is aligned with the axis of the bolt, and that the marker line is located in the central area of ​​the image. Acquire a fixed-size image of the marker area.

[0066] The resulting image is converted to grayscale using the following formula: ;

[0067] in, The coordinates are The grayscale value of the pixel is given by R, G, and B, which are the component values ​​of the pixel in the red, green, and blue channels, respectively.

[0068] Gaussian filtering is used to denoise the obtained grayscale image. A sliding window convolution is performed on the image using a Gaussian-based convolution kernel. The output value of each pixel is the weighted average of its neighboring pixels, with the weights determined by the Gaussian function. The center pixel has the largest weight, and the weights decrease as pixels become more distant. The expression for the two-dimensional Gaussian function is: ;

[0069] in, These are the coordinates of a point within the convolution kernel relative to the center point. It is the standard deviation of the Gaussian distribution.

[0070] Based on the set kernel size and standard deviation, the weight values ​​at each position within the kernel are calculated: the Gaussian function value of each coordinate point is calculated, all weight values ​​are summed to obtain the total weight, each weight value is divided by the total weight, and normalization is performed to finally obtain a convolution kernel matrix K with an element-wise sum of 1. For each pixel in the grayscale image, the filtered value is: ;

[0071] in, From arrive The summation range is defined, and the size of the neighborhood window is determined. It is the pixel grayscale value of the input image in the neighborhood; These are the weight values ​​of the convolution kernel at the corresponding positions.

[0072] When the convolution kernel slides to the image boundary, a mirror fill method is used: that is, it is assumed that the pixel values ​​outside the image boundary are mirror reflections of the pixels inside the boundary, avoiding the introduction of invalid values ​​or black borders at the image edges, and ensuring that boundary pixels are also effectively filtered. Through Gaussian filtering, high-frequency noise in the image is effectively suppressed, while the main edge features of the image are well preserved, providing high-quality input data for subsequent image processing.

[0073] For the denoised grayscale image, binarization is performed based on a set segmentation threshold to separate the foreground and background. First, the optimal segmentation threshold is determined using Otsu's method based on the inter-class variance. The formula for inter-class variance is: ;

[0074] in, The candidate threshold; Below the threshold Pixel ratio; For values ​​above the threshold Pixel ratio; Below the threshold The average gray level of the pixels; For values ​​above the threshold The average grayscale of the pixels.

[0075] Traverse all possible thresholds The integers from 0 to 255 are used to calculate the corresponding inter-class variances, and the threshold that maximizes the inter-class variance is selected. As the optimal threshold T.

[0076] The optimal threshold T is used to binarize all pixels in the grayscale image. The determination formula is as follows: ;

[0077] in, For a binary image at coordinates Pixel value at the location; This represents the foreground pixel; This represents the background pixels.

[0078] Collect the coordinates of all foreground pixels to form a set of luminous pixels, and calculate the x-coordinate of the geometric center point of this set. The summation of the x-coordinates of all pixels in the set of luminous pixels, divided by the total number of luminous pixels, yields the x-coordinate of the geometric center point of the set; the calculation formula is: ;

[0079] Where M is the number of luminous pixels. Let x be the x-coordinate of the m-th pixel; The value is used as the displacement feature.

[0080] Furthermore, the logic for extracting the position information of each light spot from the light spot image is as follows:

[0081] See Figure 3 As shown in (a), when acquiring the spot image, the longitudinal baseline of the image is aligned with the spatial projection line of the fluorescent marking line of the bolt on the bolt head end face, and the geometric midpoint of the head end face is located at the center of the image. A fixed-size image is acquired for the projection area. See also Figure 3As shown in (b), the obtained image is converted to a grayscale image using the same method and then subjected to noise reduction. Based on the segmentation threshold T described above, the obtained grayscale image is converted to a binary image, and the binary image is segmented by light points. The specific method is as follows:

[0082] Define a label image with the same size as the binary image and one-to-one pixel correspondence, with all pixel values ​​initialized to 0; define a neighborhood set containing the pixel values ​​of the four positions above, below, above right, and to the left of the current pixel, and mark any position as empty if it does not exist; define a counter with an initial value of 1, and assign the current value of the counter to the target pixel each time a new label is assigned, and then increment the counter value by 1.

[0083] Starting from the top left corner of the binary image, each pixel is read sequentially from left to right and top to bottom. If the current pixel is 0, it is skipped, and the next pixel is read. If the current pixel is 255, the neighborhood set of that pixel is read. If all values ​​in its neighborhood set are 0 or the set is empty, a new label is assigned to the corresponding pixel in the label image. If there is a 255 in its neighborhood set, the neighborhood set of the corresponding pixel is read in the label image, the smallest non-zero label in the read neighborhood set is assigned to the pixel in the label image, and the neighborhood set of the pixel in the label image is recorded. All non-zero distinct labels in the neighborhood set of the pixel in the label image are equivalent.

[0084] After scanning, all labels with equivalent relationships are grouped into a label set, and the label with the smallest value in each label set is selected as the root label of the label set. Scan the label image, and for each non-zero pixel, replace its pixel value with the root label of its label set. The coordinates of pixels with the same root label form a light spot pixel set, which is the position information of each light spot in the light spot image.

[0085] Furthermore, based on the position information of each light point, the moment of inertia of all light points relative to their center of mass is calculated. The specific steps are as follows:

[0086] For each pixel set containing a light point, calculate the coordinates of its center point based on the pixels it contains. Sum the x-coordinates of all pixels in the set and divide the result by the total number of pixels in the set to obtain the x-coordinate of the center point of the pixel set. Sum the y-coordinates of all pixels in the set and divide the result by the total number of pixels in the set to obtain the y-coordinate of the center point of the pixel set. The calculation formula is as follows: ;

[0087] in, For the set of pixels of the kth light point, This represents the total number of pixels contained in the set. Let the coordinates be the center point coordinates of the k-th light spot;

[0088] Based on the center point coordinates of each light spot in the light spot image, calculate the centroid of all light spots. The sum of the x-coordinates of the center points of all light spots, divided by the total number of light spots, yields the x-coordinate of the centroid of the image for that light spot; the sum of the y-coordinates of the center points of all light spots, divided by the total number of light spots, yields the y-coordinate of the centroid of the image for that light spot; the calculation formula is: ;

[0089] Where N is the number of projected light spots;

[0090] Based on the above results, calculate the moment of inertia of all light spots in the spot image relative to their centroids. For the center point of each light spot, square the difference between its x-coordinate and the x-coordinate of the centroid, and the difference between its y-coordinate and the y-coordinate of the centroid. Add the two squared values ​​to obtain the squared distance from the center point to the centroid. Sum the squared distances from all center points to the centroid to obtain the moment of inertia of the spot image. The calculation formula is:

[0091] ;

[0092] The moment of inertia is used as the deformation characteristic.

[0093] Step 3: During detection, the current marking line image of the bolt to be detected is acquired in the marked area. The position information of the luminescent pixels is extracted from the current marking line image. The centroid of the fluorescent marking line is calculated. The abscissa of the centroid is used as the current displacement feature of the bolt to be detected. The current light spot image of the bolt to be detected is acquired in the projection area. The position information of each light spot is extracted from the current light spot image. The moment of inertia of all light spots relative to their centroid is calculated. The moment of inertia is used as the current deformation feature of the bolt to be detected.

[0094] Furthermore, under ideal conditions where no displacement occurs, the fluorescent marker line of a bolt is fixed in position in the image, and its center of gravity's horizontal coordinate can be considered a reference value. When the bolt undergoes axial slippage or lateral shearing under external loads (such as wind vibration or structural settlement), the entire marker line will move as a whole in the image. The most direct and stable manifestation of this movement is the change in its overall geometric center (i.e., center of gravity) in the horizontal direction (horizontal coordinate).

[0095] Compared to selecting only a single point on the marked line, calculating the centroid of the entire line has significant advantages. It can effectively average out individual pixel position anomalies caused by localized unevenness in the fluorescent coating, image noise, or minor blemishes, thus obtaining a more stable and representative position indicator. Therefore, the change in the centroid's horizontal coordinate relative to the reference value can reliably reflect the overall macroscopic displacement deviation of the bolt.

[0096] Furthermore, since the fluorescent dot array is projected onto a specific area of ​​the bolt surface, when the bolt undergoes elastic or plastic deformation (such as bending or twisting), its surface is no longer an ideal plane, causing the relative positions between the projected light dots to change. In physics, the moment of inertia describes the degree of dispersion of an object's mass distribution relative to its center of mass; in this scenario, it is cleverly used to describe the dispersion of the light dot array in space.

[0097] If the bolt surface is flat and undeformed, the light spot cluster will maintain a regular and compact initial distribution, and the calculated moment of inertia will be a small reference value. When the bolt bends or twists, its surface becomes uneven, causing some light spots to be "lifted" or "pressed down," deviating from their original plane. This significantly increases the spatial dispersion of the light spot cluster relative to its centroid; that is, the sum of the squares of the distances between the light spots (i.e., the moment of inertia) will increase significantly. The moment of inertia is a scalar quantity that does not depend on the specific direction of deformation but is sensitive to deformation in all directions. Whether the deformation is unidirectional bending or complex twisting, as long as the surface geometry changes, the value of the moment of inertia will change accordingly. Therefore, it is a very effective comprehensive indicator for quantifying the overall degree of surface deformation. The change in the moment of inertia relative to the reference value directly reflects the microscopic or macroscopic deformation deviation of the bolt.

[0098] Step 4: Determine the displacement state of the bolt to be tested based on the relative displacement deviation between the current displacement feature and the reference displacement feature; determine the deformation state of the bolt to be tested based on the relative deformation deviation between the current deformation feature and the reference deformation feature; and comprehensively judge the risk of loosening of the bolt based on the displacement state and deformation state of the bolt to be tested.

[0099] Furthermore, the displacement state of the bolt to be tested is determined based on the relative displacement deviation between the current displacement feature and the reference displacement feature. The relative displacement deviation is calculated as follows: calculate the difference between the current displacement feature and the reference displacement feature, divide the absolute value of the difference by the length of the fluorescent marker line in the reference marker line image, and then multiply by 100%. The result is the relative displacement deviation between the current displacement feature and the reference displacement feature. If the relative displacement deviation is less than or equal to the first displacement threshold, it is determined to be tight. If the relative displacement deviation is greater than the first displacement threshold and less than or equal to the second displacement threshold, it is determined to be slightly moving. If the relative displacement deviation is greater than the second displacement threshold, it is determined to be loose.

[0100] Calculate the relative displacement deviation The formula is: ;

[0101] in, For the current displacement characteristics, The reference displacement feature is L, which is the length of the fluorescent marker line in the reference marker line image.

[0102] like If it is, then it is determined to be secure; if If it is, then it is determined to be a micro-motion; if If the displacement threshold is 4%, it is determined to be loose. The determination of the first displacement threshold (4%) and the second displacement threshold (10%) is based on statistical analysis of a large number of offline images of bolts under known preload conditions. Specifically:

[0103] The first displacement threshold (4%) is set as the upper limit of the statistical displacement of the marker line caused by non-loosening factors such as temperature fluctuations and slight vibrations under normal service conditions. Setting the threshold at this level can effectively avoid false judgments and ensure that the system remains silent when there is no risk of loosening.

[0104] The second displacement threshold (10%) corresponds to the critical state where the bolt undergoes macroscopic slippage or the nut rotates, and the preload is significantly lost. This threshold ensures high reliability of displacement determination; once this limit is exceeded, it indicates that the connection is in a clear failure state and immediate intervention is required.

[0105] See Figure 4 As shown in (a), when the bolt to be tested is in a tightened state, nine luminous pixels are extracted from its reference mark line image, with coordinates as follows:

[0106] (5.03, 2.10), (4.97, 2.90), (5.01, 3.80), (5.00, 4.50), (5.05, 5.20)

[0107] (4.98, 6.10), (5.02, 6.70), (5.06, 7.50), (5.03, 8.30);

[0108] Summing over all x-coordinates: ;

[0109] Divide the result by the number of luminous pixels:

[0110] ;

[0111] The reference displacement characteristic value of the bolt to be tested is 5.02;

[0112] See Figure 4 As shown in (b), after the bolt has been used for a period of time, in a subsequent inspection, 9 luminous pixels were still extracted from the current mark line image of the bolt, but they had been misaligned, with the coordinates as follows:

[0113] (4.98, 2.11), (4.99, 2.92), (5.01, 3.84), (5.02, 4.50), (6.34, 5.25)

[0114] (6.27, 6.10), (6.28, 6.79), (6.34, 7.50), (6.19, 8.28);

[0115] Summing over all x-coordinates: ;

[0116] Divide the result by the number of luminous pixels:

[0117] ;

[0118] The current displacement characteristic value of the bolt to be tested is 5.60;

[0119] Calculate the difference between the current displacement characteristic and the reference displacement characteristic: Divide the absolute value of the result by the length of the reference fluorescent marker line ( Figure 4 (a) The length is 6.20) multiplied by 100%: ;

[0120] The relative displacement deviation is then calculated as follows: ,because It can be determined that the bolt has rotated slightly, but the relative displacement deviation is close to the second displacement threshold, which poses a certain risk of loosening and requires attention.

[0121] Furthermore, the deformation state of the bolt to be tested is determined based on the relative deviation between the current deformation feature and the reference deformation feature. The relative deviation is calculated as follows: calculate the difference between the current deformation feature and the reference deformation feature, divide the absolute value of the difference by the reference deformation feature, and then multiply by 100%. The result is the relative deviation between the current deformation feature and the reference deformation feature. If the relative deviation is less than or equal to the first deformation threshold, it is considered normal. If the relative deviation is greater than the first deformation threshold but less than or equal to the second deformation threshold, it is considered a warning. If the relative deviation is greater than the second deformation threshold, it is considered abnormal.

[0122] Calculate the relative deviation of deformation The formula is: ;

[0123] in, As the baseline deformation characteristic, This represents the current deformation characteristics;

[0124] like If it is normal, then it is considered normal; if If it is, then it is judged as a warning; if If the deformation threshold is not met, it is considered abnormal. The determination of the first deformation threshold (3%) and the second deformation threshold (8%) is also based on statistical analysis of a large number of offline images of bolts under known preload conditions. Specifically:

[0125] The setting of the first deformation threshold (3%) is mainly based on the measurement repeatability accuracy of this detection system and the elastic deformation range of the bolt material. Extensive repeatability experiments have confirmed that, under constant preload, the inherent fluctuation (noise) of the measured moment of inertia of this system is far below this value. Therefore, setting the threshold to 3% can effectively distinguish the system's measurement noise from the actual material deformation signal, ensuring that no false alarms occur when the bolt is within its safe elastic range.

[0126] The second deformation threshold (8%) corresponds to the macroscopic characterization of significant plastic deformation of the bolt material or critical failure of the preload. Experimental data shows that when the relative deviation of the moment of inertia exceeds 8%, the bolted connection has usually entered an irreversible damage accumulation stage, its preload has significantly decreased, and there is a high risk of failure. The purpose of setting this threshold is to promptly capture and warn of this potential, progressive failure risk, providing a critical time window for preventative maintenance.

[0127] See Figure 5 As shown in (a), when the bolt to be tested is in a tightened state, the center points of 7 light spots are extracted from its reference light spot image, and their coordinates are as follows:

[0128] (4.00, 5.35), (4.51, 4.50), (4.67, 6.28), (5.00, 5.06)

[0129] (5.50, 4.66), (5.83, 6.40), (6.22, 5.51);

[0130] Summing over all x and y coordinates separately: ;

[0131] ;

[0132] Divide the result by the total number of light spots to obtain the x and y coordinates of the centroid of the reference light spot image:

[0133] ;

[0134] Based on the above results, calculate the moment of inertia of all light spots in the reference light spot image relative to their centroids:

[0135] ;

[0136] ;

[0137] The reference deformation characteristic value of the bolt is 6.96;

[0138] See Figure 5 As shown in (b), after the bolt has been used for a period of time, during a subsequent inspection, the center points of 7 light spots were extracted from the current light spot image of the bolt, with coordinates as follows:

[0139] (3.95, 5.33), (4.47, 4.47), (4.73, 6.25), (5.02, 5.05)

[0140] (5.55, 4.63), (5.83, 6.45), (6.25, 5.54);

[0141] Summing over all x and y coordinates separately:

[0142] ;

[0143] ;

[0144] Divide the result by the total number of light spots to obtain the x and y coordinates of the centroid of the current light spot image:

[0145] ;

[0146] Based on the above results, calculate the moment of inertia of all light spots in the current light spot image relative to their centroids:

[0147] ;

[0148] ;

[0149] The current characteristic value of the bolt's deformation is 7.34.

[0150] Calculate the difference between the current deformation feature and the reference deformation feature: Divide the absolute value of the result by the reference deformation characteristic and then multiply by 100%: ;

[0151] The relative deviation of deformation is then calculated as follows: ,because It can be determined that the bolt has undergone moderate deformation. However, the deformation characteristics are microscopic. In other words, when a clearly measurable deformation occurs, the bolt displacement may have already shifted significantly. Therefore, it needs to be determined as a warning state. The preload state should be comprehensively judged by combining the measured displacement data.

[0152] Furthermore, based on the displacement and deformation states of the bolt to be tested, the loosening risk of the bolt is comprehensively judged. A loosening risk index and a risk coupling coefficient are defined. The loosening risk index is calculated as follows: multiply the risk coupling coefficient by the relative displacement deviation and the relative deformation deviation, and sum the resulting product with the relative displacement deviation and the relative deformation deviation. The result is the loosening risk index. The risk level of the bolt to be tested is determined based on the loosening risk index: if the loosening risk index is less than or equal to the first risk threshold, it is judged as low risk; if the loosening risk index is greater than the first risk threshold and less than or equal to the second risk threshold, it is judged as medium risk; if the loosening risk index is greater than the second risk threshold, it is judged as high risk, and the system automatically alarms.

[0153] The specific formula for calculating the loosening risk index R is as follows: ;

[0154] in The risk coupling coefficient is a constant greater than 0.

[0155] The risk level of the bolt to be inspected is determined based on the calculated loosening risk index R: If... If it is determined to be low risk; It is classified as medium risk; if If the risk is deemed high, the system will automatically issue an alarm; among them The first risk threshold, This is the second risk threshold.

[0156] This step is crucial for transforming the system from "single-feature alarm" to "intelligent risk warning." It's important to recognize that bolt loosening is a gradual process where macroscopic displacement and microscopic deformation are not independent events but are profoundly coupled: deformation weakens the bolt's tightening ability, thus promoting displacement; conversely, displacement causes stress redistribution, exacerbating local deformation. To quantify this synergistic effect, this embodiment constructs a nonlinear risk fusion model to calculate the loosening risk index R. This is a linear superposition term, representing the basic risk contributed by displacement and deformation when they develop independently. For nonlinear coupling terms, it quantifies the additional risk gain caused by the interaction of two anomalous states when they occur simultaneously; the coupling coefficient is... This determines the strength of the secondary gain. Some data on the loosening risk index are shown in Table 1.

[0157] Table 1: Partial Loosening Risk Index Data

[0158]

[0159] See Figure 6 and Figure 7As shown in Table 1, it is clear from the data that the loosening risk index R is related to the relative deviation of the displacement. and relative deviation of deformation They all show a positive correlation, specifically as follows:

[0160] relative displacement deviation The impact on the loosening risk index R is significant: when The risk of easing increased significantly, rising from 0.2% to 7.7%, with the risk index R increasing from 0.73 to 14.48. This increase was particularly pronounced in [specific context - likely related to specific events]. After exceeding 3%, the growth rate of the R value accelerated, indicating that the increase in displacement has a significant accelerating effect on the improvement of the loosening risk index.

[0161] relative deformation deviation It also has a significant impact on the loosening risk index R: when As the percentage increased from 0.53% to 6.72%, the R-value also increased. Once it exceeds 3%, the increase in R value becomes more pronounced, and the increase in surface deformation will also significantly increase the loosening risk index.

[0162] The coupling effect between relative displacement deviation and relative deformation deviation: In most cases, when and When R increases synchronously, the increase in R value is more dramatic. For example, when It increased from 3.5% to 7.7%, while The percentage increased from 3.13% to 6.72%, while the R-value jumped from 6.66 to 14.48, indicating that the combined effect of the two factors significantly amplifies the risk.

[0163] Therefore, increases in both displacement and deformation lead to a rise in the loosening risk index, and there is a synergistic reinforcing effect between the two. In practical engineering or risk assessment, it is necessary to monitor and control changes in displacement and deformation simultaneously, especially when their values ​​are high, to prevent the loosening risk index from rapidly climbing to a dangerous level.

[0164] Furthermore, it should be noted that, through theoretical analysis combined with boundary calculations for typical working conditions, it can be determined that... The value range is from 0.03 to 0.08. When When the coefficient of performance is below 0.03, the contribution of the coupling term to the model will be negligible, leading to situations where high-risk operating conditions occur without triggering an alarm; while when... Above 0.08, the coupling effect is excessively amplified, causing any change in the bolt to trigger a system alarm. Similarly, after repeated verification, a first risk threshold can be determined. The value range is 7.0 to 9.0, the second risk threshold. The value range is from 16.0 to 20.0.

[0165] In this embodiment, the determination of all threshold ranges is to seek an optimal engineering balance between the two contradictory goals of "avoiding false alarms" and "preventing false alarms".

[0166] Please see Figure 8 The present invention also provides an offline bolt preload detection system, which is used to perform the above-described offline bolt preload detection method, including:

[0167] The marking creation module is used to, when the installation conditions are met, use the mating surface between the nut and the connector on the bolt to be tested as the marking area, spray a fluorescent marking line across the mating surface in the marking area, and collect a reference marking line image of the bolt to be tested in the marking area. Also, when the installation conditions are met, use the head end face of the bolt to be tested as the projection area, and project multiple fluorescent marking points in the projection area based on preset parameters, and collect a reference light spot image of the bolt to be tested in the projection area.

[0168] The reference feature extraction module is used to extract the position information of the luminescent pixels from the reference mark line image, calculate the centroid of the fluorescent mark line, use the abscissa of the centroid as the reference displacement feature of the bolt to be tested, and extract the position information of each light spot from the reference light spot image, calculate the moment of inertia of all light spots relative to their centroid, and use the moment of inertia as the reference deformation feature of the bolt to be tested.

[0169] The detection feature extraction module is used to acquire the current marking line image of the bolt to be detected in the marking area during detection, extract the position information of the luminescent pixels from the current marking line image, calculate the centroid of the fluorescent marking line, and use the abscissa of the centroid as the current displacement feature of the bolt to be detected. It also acquires the current light spot image of the bolt to be detected in the projection area, extracts the position information of each light spot from the current light spot image, calculates the moment of inertia of all light spots relative to their centroid, and uses the moment of inertia as the current deformation feature of the bolt to be detected.

[0170] The risk assessment module is used to determine the displacement state of the bolt to be tested based on the relative displacement deviation between the current displacement feature and the reference displacement feature, to determine the deformation state of the bolt to be tested based on the relative deformation deviation between the current deformation feature and the reference deformation feature, and to comprehensively assess the loosening risk of the bolt based on the displacement state and deformation state of the bolt to be tested.

[0171] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0172] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0173] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0174] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for detecting the pre-tightening force of a bolt, characterized in that, The specific steps include: Step 1: When the installation condition is met, a fluorescent marking line is sprayed on the joint surface of the nut to be detected and the connecting piece in the marking area, a reference marking line image of the bolt to be detected is collected in the marking area, and when the installation condition is met, a plurality of fluorescent marking points are projected in the projection area based on the preset parameters, and a reference light point image of the bolt to be detected is collected in the projection area; Step 2: The position information of the light-emitting pixel points is extracted from the reference marking line image, the center of gravity of the fluorescent marking line is calculated, the horizontal coordinate of the center of gravity is taken as the reference displacement feature of the bolt to be detected, and the position information of each light point is extracted from the reference light point image, the inertia moment of all light points relative to the center of mass is calculated, and the inertia moment is taken as the reference deformation feature of the bolt to be detected; Step 3: When detecting, the current marking line image of the bolt to be detected is collected in the marking area, the position information of the light-emitting pixel points is extracted from the current marking line image, the center of gravity of the fluorescent marking line is calculated, the horizontal coordinate of the center of gravity is taken as the current displacement feature of the bolt to be detected, and the current light point image of the bolt to be detected is collected in the projection area, the position information of each light point is extracted from the current light point image, the inertia moment of all light points relative to the center of mass is calculated, and the inertia moment is taken as the current deformation feature of the bolt to be detected; Step 4: The displacement state of the bolt to be detected is determined according to the displacement relative deviation of the current displacement feature and the reference displacement feature, the deformation state of the bolt to be detected is determined according to the deformation relative deviation of the current deformation feature and the reference deformation feature, and the loosening risk of the bolt to be detected is comprehensively determined according to the displacement state and the deformation state of the bolt to be detected.

2. The method of claim 1, wherein: The fluorescent marking line is parallel to the axis of the bolt to be detected in space, and the spatial projection line of the bolt head end surface in the plane coincides with the center line of the bolt head end surface. Under the condition of a specific light source excitation, the reference marking line image of the bolt to be detected is collected in the marking area; the bolt head end surface is taken as the projection area and a plane rectangular coordinate system is established, the midpoint of the head end surface is taken as the origin, the spatial projection line of the fluorescent marking line of the bolt in the plane is taken as the vertical axis direction, and the fluorescent marking points are projected in the projection area. Under the condition of a specific light source excitation, the reference light point image of the bolt to be detected is collected in the projection area.

3. The method of claim 2, wherein: The position information of the light-emitting pixel points is extracted and the center of gravity of the fluorescent marking line is calculated, and the logic of the displacement feature of the bolt to be detected is obtained as follows: Collect a fixed-size image of the marking area, the imaging longitudinal axis direction is parallel to the extension direction of the fluorescent marking line, and the fluorescent marking line is in the central region of the picture; convert the image into a gray-scale image and perform noise reduction processing, and perform binary determination on all pixel points in the gray-scale image according to the set segmentation threshold value, if the gray value is less than the segmentation threshold value, it is determined as a background pixel point, otherwise it is determined as a foreground pixel point; collect all foreground pixel point coordinates to form a light-emitting pixel point set, and calculate the horizontal coordinate of the geometric center point of the set, the specific method is: sum all the horizontal coordinates of the pixel points contained in the light-emitting pixel point set, divide the result by the total number of light-emitting pixel points, and the horizontal coordinate of the geometric center point of the set is obtained; the value of the horizontal coordinate is taken as the displacement feature.

4. The method of claim 3, wherein: The logic for extracting the position information of each light point from the light point image is: Collect a fixed-size image of the projection area, the imaging direction is consistent with the coordinate system of the bolt head end face, and the bolt head end face is in the central region of the picture; convert the image into a gray-scale image and perform noise reduction processing, and perform binary determination on all pixel points in the gray-scale image according to the set segmentation threshold value, if the gray value is less than the segmentation threshold value, it is determined as a background pixel point, otherwise it is determined as a foreground pixel point, and a binary image is generated; Scan the binary image to divide each light point pixel set, the specific steps are: Define a label image with the same size as the binary image and one-to-one correspondence of pixel points, all pixel values are initialized to 0; define a neighborhood set containing the pixel values of the left upper, upper, right upper and left positions of the current pixel point, if any of them does not exist, it is recorded as empty; define a counter with an initial value of 1, assign the current value of the counter to the target pixel point after each new label is assigned, and then the counter value is incremented by 1; Start from the top left corner of the binary image, read each pixel point in order from left to right and from top to bottom, if the current pixel point is background, skip to the next pixel point, if the current pixel point is foreground, read the neighborhood set of the pixel point, if all the neighborhood sets are background or the set is empty, assign a new label to the corresponding pixel point in the label image; If there is foreground in the neighborhood set, read the neighborhood set of the corresponding pixel point in the label image, assign the smallest non-zero label in the read neighborhood set to the label image pixel point, and record all non-zero equivalent labels in the neighborhood set of the label image pixel point. After scanning, all labels with equivalent relationship are classified into a label set, and the smallest label value in each label set is selected as the root label of the label set; scan the label image, replace the pixel value of each non-zero pixel point with the root label of the label set it belongs to, and the pixel point coordinates with the same root label form a light point pixel set, which is the position information of each light point in the light point image.

5. The method of claim 4, wherein: The logic for calculating the inertia moment of all light points relative to their centroid according to the position information of each light point is: For each light point pixel set, the center point coordinates of the light point are calculated according to the pixels contained in the set, the sum of all horizontal coordinates of the pixels contained in the set is divided by the total number of pixels in the set to obtain the horizontal coordinate of the center point of the light point pixel set, and the sum of all vertical coordinates of the pixels contained in the set is divided by the total number of pixels in the set to obtain the vertical coordinate of the center point of the light point pixel set; Similarly, the center point coordinates of each light point in the light point image are obtained, and the centroid of all light points is calculated, the sum of the horizontal coordinates of the center points of all light points is divided by the total number of light points to obtain the horizontal coordinate of the centroid of the light point image, and the sum of the vertical coordinates of the center points of all light points is divided by the total number of light points to obtain the vertical coordinate of the centroid of the light point image; According to the above results, the inertia moment of all light points in the light point image relative to the centroid thereof is calculated, for each center point of the light point, the square of the difference between the horizontal coordinate thereof and the horizontal coordinate of the centroid, and the square of the difference between the vertical coordinate thereof and the vertical coordinate of the centroid are calculated, and the sum of the two square values is obtained to obtain the square of the distance from the center point to the centroid; The sum of the squares of the distances from all center points to the centroid is obtained to obtain the inertia moment of the light point image; the inertia moment is used as the deformation feature.

6. The method of claim 5, wherein: The displacement relative deviation is calculated by calculating the difference between the current displacement feature and the reference displacement feature, dividing the absolute value of the difference by the length of the fluorescent marker line in the reference marker line image, and multiplying 100% to obtain the displacement relative deviation between the current displacement feature and the reference displacement feature; if the displacement relative deviation is less than or equal to the first displacement threshold, it is determined to be tightened; if the displacement relative deviation is greater than the first displacement threshold and less than or equal to the second displacement threshold, it is determined to be micro-motion; If the displacement relative deviation is greater than the second displacement threshold, it is determined to be loose.

7. The method according to claim 6, characterized in that: The deformation relative deviation is calculated by calculating the difference between the current deformation feature and the reference deformation feature, dividing the absolute value of the difference by the reference deformation feature, and multiplying 100% to obtain the deformation relative deviation between the current deformation feature and the reference deformation feature; If the deformation relative deviation is less than or equal to the first deformation threshold, it is determined to be normal; If the deformation relative deviation is greater than the first deformation threshold and less than or equal to the second deformation threshold, it is determined to be a warning; If the deformation relative deviation is greater than the second deformation threshold, it is determined to be abnormal.

8. The method of claim 7, wherein: According to the displacement state and deformation state of the bolt to be detected, the loosening risk of the bolt is comprehensively judged, specifically: A loosening risk index and a risk coupling coefficient are defined, the calculation method of the loosening risk index is that the risk coupling coefficient is multiplied by the displacement relative deviation and the deformation relative deviation, and the sum of the products and the displacement relative deviation and the deformation relative deviation is obtained, and the result is the loosening risk index; the risk level of the bolt to be detected is determined according to the loosening risk index: if the loosening risk index is less than or equal to the first risk threshold, it is determined to be low risk; If the loosening risk index is greater than the first risk threshold and less than or equal to the second risk threshold, it is determined to be medium risk; If the loosening risk index is greater than the second risk threshold, it is determined to be high risk, and the system automatically alarms.

9. A bolt pre-load off-line detection system characterized by: The bolt pre-tightening force offline detection system is used for executing the bolt pre-tightening force offline detection method of any one of claims 1-8, and comprises: a mark creating module, configured to, when the installation condition is met, take a joint surface of a nut on the bolt to be detected and a connecting piece as a mark area, spray a fluorescent mark line across the joint surface in the mark area, and collect a reference mark line image of the bolt to be detected for the mark area, and when the installation condition is met, take an end surface of a head of the bolt to be detected as a projection area, project a plurality of fluorescent mark points in the projection area based on preset parameters, and collect a reference light point image of the bolt to be detected for the projection area; a reference feature extraction module, configured to extract position information of light-emitting pixel points from the reference mark line image, calculate a barycenter of the fluorescent mark line, take an abscissa of the barycenter as a reference displacement feature of the bolt to be detected, and extract position information of each light point from the reference light point image, calculate an inertia moment of all light points relative to a centroid thereof, and take the inertia moment as a reference deformation feature of the bolt to be detected; a detection feature extraction module, configured to, when detection is performed, collect a current mark line image of the bolt to be detected for the mark area, extract position information of light-emitting pixel points from the current mark line image, calculate a barycenter of the fluorescent mark line, take an abscissa of the barycenter as a current displacement feature of the bolt to be detected, and collect a current light point image of the bolt to be detected for the projection area, extract position information of each light point from the current light point image, calculate an inertia moment of all light points relative to a centroid thereof, and take the inertia moment as a current deformation feature of the bolt to be detected; a risk judgment module, configured to judge a displacement state of the bolt to be detected according to a displacement relative deviation of the current displacement feature and the reference displacement feature, judge a deformation state of the bolt to be detected according to a deformation relative deviation of the current deformation feature and the reference deformation feature, and comprehensively judge a loosening risk of the bolt according to the displacement state and the deformation state of the bolt to be detected.

Images