A porcelain insulator glaze quality laser detection and evaluation system

Abstract

The present application relates to the technical field of intelligent detection, in particular to a porcelain insulator glaze quality laser detection and evaluation system, which comprises a glaze photoelectric acquisition module, an umbrella skirt morphology reconstruction module, a defect feature screening module, a glaze light loss quantification module and an image detection and evaluation module.In the present application, the background component is stripped by combining with the medium reflectivity correction factor to extract the scanning row barycenter, effectively solving the positioning deviation problem caused by the complex curved surface light path distortion, improving the three-dimensional topology reconstruction accuracy, locking the defect center and calculating the radiation energy attenuation, deeply quantifying the light loss sweeping attenuation sequence of the defect edge, realizing the accurate mapping of the microscopic medium energy dissipation behavior, strengthening the anti-interference extraction ability of the hidden bubble in the strong background, reducing the feature misjudgment risk caused by environmental noise, optimizing the calculation process of the micro-pore scattering constant and the energy absorption ratio, enhancing the robustness of the non-uniformity evaluation result, and ensuring the objectivity and reliability of quality investigation.

Description

Technical Field

[0001] This invention relates to the field of intelligent detection technology, and in particular to a laser detection and evaluation system for the glaze quality of porcelain insulators. Background Technology

[0002] The field of intelligent detection technology is a comprehensive application of sensing, visual imaging, and optical measurement to automatically quantify and acquire the state and physical form of a target object. It systematically covers the deployment of various sensors in the target environment to capture physical signals such as reflected light waves, sound waves, scattering, or electromagnetic changes on the object's surface. The acquired continuous physical parameters are converted into discrete digital matrices, and the geometric dimensions, surface continuity, or medium distribution of the target object are quantitatively measured through expected matrix algebra operations and fixed numerical comparison conditions to establish a non-contact physical inspection process. The traditional laser inspection and evaluation system for porcelain insulator glaze quality refers to a collection of equipment used to inspect for surface anomalies such as microcracks, bubbles, pinholes, and glaze defects in the glassy glaze of insulators in power transmission lines. The traditional system uses a linear laser emitter to project a monochromatic laser beam of a specific wavelength onto the uniformly rotating arc-shaped glaze of the porcelain insulator. An industrial area array camera, positioned at a specific geometric reflection angle, captures the laser triangulation rangefinder image modulated by the glaze morphology. The system extracts the centerline pixels of the laser beams row by row in the pixel coordinate system and calculates the spatial three-dimensional point cloud coordinates of the glaze surface using the camera's intrinsic and extrinsic parameter calibration matrix. Then, it extracts pixels with abrupt changes in grayscale values ​​and three-dimensional coordinates with discontinuous spatial depth values ​​from the camera-acquired images. These are compared pixel by pixel with the reference insulator's three-dimensional coordinate array and the standard grayscale matrix, serving as the specific operational steps and execution basis for identifying changes in the microscopic physical morphology of the glaze.

[0003] Traditional detection mechanisms rely on monochromatic laser beam projection and fixed-angle visual imaging to perform physical measurements, directly extracting the center pixel of the light stripe and constructing a three-dimensional spatial point cloud with intrinsic and extrinsic parameter matrices. They only perform a single model difference comparison based on grayscale abrupt changes and depth tomographic coordinates. This shallow comparison mode leads to the neglect of optical path distortion on complex curved glaze surfaces, resulting in a large number of false defect interference terms being added to the acquisition array. At the same time, the mechanical pixel-by-pixel subtraction mechanism lacks in-depth analysis of the dielectric loss characteristics, causing fine bubbles and hidden features to be masked by strong noise, thus limiting the reliability of the assessment. Summary of the Invention

[0004] To address the technical problems existing in the prior art, this invention provides a laser detection and evaluation system for the glaze quality of porcelain insulators. The technical solution is as follows: On the one hand, a laser detection and evaluation system for the glaze quality of porcelain insulators is provided, the system comprising: The glaze photoelectric acquisition module captures the rising edge of the pulse level signal of the photoelectric rotary encoder, matches the laser emitter of the preset trigger reference trigger line structure, and works with the industrial area array camera to capture the reflected laser light strip on the umbrella skirt arc surface. The effective component after extracting the reflectivity correction factor of the insulating layer medium and removing the background energy interference is extracted. The centroid of the scanning line is extracted to generate the glaze reflected light intensity energy response matrix. The umbrella skirt morphology reconstruction module calculates the extreme values ​​of pixel brightness sequence gradient based on the glaze surface reflected light intensity energy response matrix, constructs instantaneous radial quantity, extracts instantaneous normal vector by fusing the curvature radius of the umbrella skirt arc surface and the refractive index of the glass phase matrix, updates the dynamic compensation matrix by matching the cosine value of the reference normal vector deflection, and outputs the physical topology array of the umbrella skirt glaze layer spatial morphology. The defect feature screening module is based on the physical topology array of the spatial morphology of the umbrella skirt glaze layer. It compares the gradient difference in the depth direction of adjacent spatial points, locks the geometric center of the defect outside the preset threshold of the depth fault, extracts the pixel grayscale of the radiation path, and calculates the energy attenuation by combining the reflection intensity and transmittance of the glaze grain boundary to obtain the light intensity attenuation feature set of the radiation path of the defect edge. The glaze optical loss quantification module calculates the optical loss sweep attenuation rate sequence that varies with direction based on the light intensity attenuation feature set of the radiation path at the defect edge, integrates the scattering coefficient of the glass substrate and the scattering constant of the micropores of the glaze surface, extracts the isotropic difference value that satisfies the deformation tolerance, calculates the medium absorption ratio, and outputs the non-uniformity detection result of the glaze optical loss.

[0005] As a further aspect of the present invention, the glaze surface reflective light intensity energy response matrix includes normalized pixel gray levels, synthetic energy response distribution weights, and pulse level amplitudes after background suppression; the umbrella-skirt glaze layer spatial morphology physical topology array includes point cloud three-dimensional vertex coordinates, local surface normal feature vectors, and spatial topological adjacency weight matrix; the defect edge radiation path light intensity attenuation feature set includes radiation direction energy attenuation rate, defect boundary light intensity contrast, and unit path dielectric loss constant; and the non-uniformity detection result of glaze light loss includes anisotropic attenuation deviation coefficient, dielectric layer energy absorption rate, and isotropic correction value caused by diffuse reflection.

[0006] As a further aspect of the present invention, the glaze surface photoelectric acquisition module includes: The rotor circumferential phase synchronization capture submodule counts the rising edge of the output pulse level of the insulator rotating shaft, calculates the time difference with the preset trigger reference, sends the start command to the linear structure laser emitter, and obtains the synchronous control trigger pulse quantity through circumferential phase angle modulus calculation. The arc surface light strip exposure response submodule receives the synchronous control trigger pulse quantity, drives the camera to capture the reflected laser light strip on the arc surface of the umbrella skirt, performs nonlinear mapping by combining the reflectivity correction factor of the insulating layer medium, extracts the two-dimensional coordinates corresponding to the gray-scale extreme points in the width direction of the light strip, and obtains the original pixel set of the arc surface of the umbrella skirt. The light intensity matrix centroid mapping submodule loads the original pixel set of the umbrella skirt arc surface, performs scan line centroid extraction, corrects the slope offset of the light intensity response curve, arranges the grayscale data of multiple frames of exposed pixels into a two-dimensional spatial array, and generates the glaze surface reflective light intensity energy response matrix.

[0007] As a further aspect of the present invention, the umbrella skirt morphology reconstruction module includes: The brightness gradient extreme radius extraction submodule performs adjacent point operations on the brightness sequence based on the glaze surface reflected light intensity energy response matrix, locks the pixel horizontal coordinates corresponding to the positive and negative extreme values ​​of the gradient, extracts and calculates the sub-pixel center, and merges the geometric center to form the glaze surface contour feature chain to obtain the instantaneous extreme radial quantity set. The refractive index dynamic compensation submodule introduces the instantaneous polar radial quantity set, integrates the curvature radius of the umbrella skirt arc surface design with the refractive index of the glass phase matrix, extracts discrete compensation variables by comparing the cosine value of the normal vector deflection, updates the basic calibration matrix, and obtains the dynamic compensation mapping matrix. The spatial coordinate topology mapping submodule calls the dynamic compensation mapping matrix to perform matrix linear transformation on the instantaneous radial quantity set, maps the pixel coordinates to three-dimensional Euclidean space, performs point cloud density normalization and establishes a spatial topology set including normal vector information, and outputs the physical topology array of the umbrella skirt glaze layer spatial morphology.

[0008] As a further aspect of the present invention, the discrete compensation variable refers to the optical path offset caused by the refractive index of the glass phase matrix, calculated based on the slope of the tangent at the sampling point of the instantaneous radial quantity set, combined with the curvature radius of the umbrella skirt arc surface design, and calculated according to Snell's law.

[0009] As a further aspect of the present invention, the defect feature screening module includes: The depth fault threshold locking submodule, based on the physical topology array of the spatial morphology of the umbrella skirt glaze layer, compares the gradient difference in the depth direction of adjacent points to determine candidate spatial coordinate points that exceed the threshold. By evaluating the connected domain characteristics of the coordinate distribution in the abnormal area, it records the spatial location index and obtains the suspected depth fault coordinate points. The geometric centroid physical positioning submodule retrieves the coordinates of the suspected depth fault, calculates the arithmetic mean of the coordinate cluster to determine the geometric centroid, and performs a depth compensation operation in combination with the glaze layer thickness constant to obtain the coordinates of the defect geometric center. The radiation path energy detection submodule maps the coordinates of the geometric center of the defect, extracts the gray values ​​along the eight-directional radiation path in the glaze reflection light intensity energy response matrix, and combines the glaze grain boundary reflection intensity and transmittance parameters to calculate the light intensity energy attenuation at the beginning and end of the radiation path, thus obtaining the feature set of light intensity attenuation of the radiation path at the defect edge.

[0010] As a further aspect of the present invention, the enamel light loss quantification module includes: The attenuation difference calculation submodule calculates the unidirectional light intensity attenuation rate and filters out extreme values ​​based on the light intensity attenuation feature set of the radiation path at the defect edge, determines the isotropic level of light transmission inside the glaze, and associates it with the unidirectional absorption path length to obtain the directional attenuation rate sequence. The scattering per-directional difference molecular module, based on the directional attenuation rate sequence, introduces the scattering coefficient of the glass phase matrix and the diffuse reflection constant of the glaze micropores, calculates the isotropic offset and performs weight correction in combination with the lattice arrangement direction, and generates the attenuation anisotropy coefficient. The energy absorption calculation submodule, based on the attenuation anisotropy coefficient, accumulates the unidirectional light intensity attenuation rate that meets the conditions, obtains the total attenuation accumulation value, compares it with the preset absorption constant, calculates the energy absorption ratio per unit volume of the transparent medium, matches the preset threshold interval of bubble characteristics, and performs nonlinear fitting in combination with the refractive index fluctuation component of the local area to obtain the non-uniformity detection result of enamel optical loss.

[0011] As a further aspect of the present invention, the unidirectional light intensity attenuation rate and extreme value screening refers to extracting the incident energy and outgoing energy of the sampling path of the light intensity attenuation feature set of the radiation path at the defect edge, calculating the logarithmic loss value of light intensity per unit length of the sampling path, and locking the maximum and minimum values ​​of energy attenuation by comparing the loss values ​​under the different spatial dimensions.

[0012] As a further aspect of the present invention, the system also includes an image detection and evaluation module: Based on the non-uniformity detection results of the glaze optical loss, the image detection and evaluation module extracts the bubble feature interval and performs spatial filtering on the glaze grain boundary morphology. By mapping the umbrella skirt glaze layer topology array, it calibrates the discrete points of microbubbles and generates a spatial distribution trajectory map of microbubble defects in the insulator. The spatial distribution trajectory map of microbubble defects in the insulator includes the coordinates of the centroid of the three-dimensional defect cluster, the geometric connectivity features of the microbubbles, and the surface projection distribution points based on the image detection results.

[0013] As a further aspect of the present invention, the image detection and evaluation module includes: The feature threshold filtering submodule, based on the non-uniformity detection result of the enamel light loss, starts image detection, compares the bubble feature interval with the preset threshold and performs spatial domain filtering, performs coordinate inverse transformation and calculates the geometric area-to-depth ratio to obtain the coordinate point of the center of the range. The defect topology redrawing submodule performs image detection and preset defect space template comparison based on the center coordinate point of the conforming range, fuses and classifies the discrete point set of defect space, calibrates the distribution trajectory of internal microbubbles and analyzes the connectivity features of point cloud, and generates a defect topology structure set. The distribution map integration submodule extracts the image detection index, summarizes the polar diameter parameters and image detection index of the defect topology set, combines the three-dimensional surface mesh parameters in the physical topology array of the umbrella skirt glaze layer spatial morphology, performs a mapping transformation from spatial coordinates to surface unfolding, aggregates the microbubble spatial trajectories, and generates a spatial distribution trajectory map of insulator microbubble defects.

[0014] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: By combining the medium reflectivity correction factor to strip the background component to extract the centroid of the scan line, and fusing the radius of curvature and the matrix refractive index to update the dynamic compensation matrix, the positioning deviation problem caused by the optical path distortion of complex curved surfaces is effectively solved, the accuracy of three-dimensional topology reconstruction is improved, the defect center is locked and the radiation energy attenuation is calculated, the optical loss sweep attenuation sequence of the defect edge is deeply quantified, the energy dissipation behavior of the micro-medium is accurately mapped, the anti-interference extraction capability of hidden bubbles in strong background is strengthened, the risk of feature misjudgment caused by environmental noise is reduced, the calculation process of micropore scattering constant and energy absorption ratio is optimized, the robustness of non-uniformity evaluation results is enhanced, and the objectivity and reliability of quality inspection are ensured. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of a laser detection and evaluation system for the glaze quality of porcelain insulators provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the system framework of the present invention; Figure 3 This is a flowchart of the glaze surface photoelectric acquisition module in this invention; Figure 4 This is a flowchart of the umbrella skirt morphology reconstruction module in this invention; Figure 5 This is a flowchart of the defect feature screening module in this invention; Figure 6 This is a flowchart of the glaze optical loss quantification module in this invention; Figure 7 This is a flowchart of the image detection and evaluation module in this invention. Detailed Implementation

[0017] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0018] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0019] This invention provides a laser detection and evaluation system for the glaze quality of porcelain insulators, such as... Figure 1-2 The diagram shown illustrates a laser inspection and evaluation system for the glaze quality of porcelain insulators. This system includes: The glaze photoelectric acquisition module captures the rising edge of the pulse level signal of the photoelectric rotary encoder, matches the laser emitter of the preset trigger reference trigger line structure, and works with the industrial area array camera to capture the reflected laser light strip on the umbrella skirt arc surface. The effective component after extracting the reflectivity correction factor of the insulating layer medium and removing the background energy interference is extracted. The centroid of the scanning line is extracted to generate the glaze reflected light intensity energy response matrix. The umbrella skirt morphology reconstruction module is based on the glaze surface reflective light intensity energy response matrix. It calculates the extreme values ​​of the pixel brightness sequence gradient, constructs the instantaneous radial quantity, integrates the curvature radius of the umbrella skirt arc surface and the refractive index of the glass phase matrix to extract the instantaneous normal vector, matches the reference normal vector deflection cosine value to update the dynamic compensation matrix, and outputs the physical topology array of the umbrella skirt glaze layer spatial morphology. The defect feature screening module is based on the physical topology array of the spatial morphology of the umbrella skirt glaze layer. It compares the gradient difference in the depth direction of adjacent spatial points, locks the geometric center of the defect outside the preset threshold of the depth fault, extracts the pixel grayscale of the radiation path, and calculates the energy attenuation by combining the reflection intensity and transmittance of the glaze grain boundary to obtain the light intensity attenuation feature set of the radiation path of the defect edge. The glaze optical loss quantification module calculates the optical loss sweep attenuation rate sequence that varies with direction based on the light intensity attenuation feature set of the radiation path at the defect edge, integrates the scattering coefficient of the glass substrate and the scattering constant of the micropores of the glaze, extracts the isotropic difference value that satisfies the deformation tolerance, calculates the medium absorption ratio, and outputs the non-uniformity detection results of the glaze optical loss. The image detection and evaluation module extracts bubble feature regions and performs spatial filtering based on the non-uniformity detection results of glaze optical loss. It then calibrates the discrete points of microbubbles by mapping the topological array of the umbrella skirt glaze layer, generating a spatial distribution trajectory map of microbubble defects in the insulator.

[0020] The glaze surface reflectance intensity energy response matrix includes normalized pixel gray levels, synthetic energy response distribution weights, and pulse level amplitudes after background suppression. The umbrella skirt glaze layer spatial morphology physical topology array includes point cloud three-dimensional vertex coordinates, local surface normal feature vectors, and spatial topological adjacency weight matrix. The defect edge radiation path light intensity attenuation feature set includes radiation direction energy attenuation rate, defect boundary light intensity contrast, and unit path dielectric loss constant. The non-uniformity detection results of glaze light loss include anisotropic attenuation deviation coefficient, dielectric layer energy absorption rate, and isotropic correction value caused by diffuse reflection. The insulator microbubble defect spatial distribution trajectory map includes three-dimensional defect cluster centroid coordinates, microbubble geometric connectivity features, and surface projection distribution points based on image detection results.

[0021] Specifically, such as Figure 2 , 3 As shown, the glaze surface photoelectric acquisition module includes: The rotor circumferential phase synchronization capture submodule counts the rising edge of the output pulse level of the insulator rotating shaft, calculates the time difference with the preset trigger reference, sends the start command to the linear structure laser emitter, and obtains the synchronous control trigger pulse quantity through circumferential phase angle modulus calculation. The system receives alternating high and low level pulse signals from the photoelectric rotary encoder and extracts the timestamp of the abrupt transition from low to high in the continuous square wave pulse level. It retrieves the absolute time record value fixed during the initialization phase of the main control board clock chip as the preset trigger reference timestamp. The system then performs a subtraction operation between the pulse level rising edge timestamp and the preset trigger reference timestamp to generate the difference time node data. For example, extracting the rising edge timestamp of 10250 microseconds and subtracting the preset trigger reference timestamp of 10000 microseconds yields a difference time node of 250 microseconds. The rotor circumferential phase synchronization capture submodule sends the data packet containing the difference time node to the line structure laser emitter driver controller via the serial communication bus, issuing an enable command by pulling the control pin high. The system synchronously reads the cumulative total rotation angle parameter of the rotating shaft from the photoelectric rotary encoder interface, retrieves the rotation step angle parameter from the configuration register, and performs a modulo operation on the cumulative total rotation angle parameter and the rotation step angle parameter to obtain the circumferential phase angle remainder. Finally, it reads the step pulse equivalent coefficient from the equipment calibration file and performs a multiplication operation between the circumferential phase angle remainder and the step pulse equivalent coefficient to obtain the synchronous control trigger pulse quantity.

[0022] The arc surface light strip exposure response submodule receives the synchronous control trigger pulse quantity, drives the camera to capture the reflected laser light strip on the arc surface of the umbrella skirt, performs nonlinear mapping by combining the reflectivity correction factor of the insulating layer medium, extracts the two-dimensional coordinates corresponding to the gray-level extreme points in the width direction of the light strip, and obtains the original pixel set of the arc surface of the umbrella skirt. The system receives synchronous control trigger pulses, converts them into hardware interrupt trigger signals, and sends them to the external trigger pin of the global shutter industrial camera to control the image sensor to capture the laser light strip image of the umbrella skirt arc surface reflective structure. It retrieves the insulating layer reflectivity correction factor from memory and constructs a multilayer perceptron network (MLB) containing an input layer, hidden layer, and output layer to process image pixel data. Input layer neurons correspond to the horizontal pixel sequence of the image, and the output layer uses a linear mapping function to output the corrected pixel grayscale values. The original grayscale values ​​of each pixel are input into the MLB network, and a division mapping operation is performed in the hidden layer using the insulating layer reflectivity correction factor. For example, dividing the original grayscale value of 150 by the correction factor 0.8 yields the theoretical grayscale value of 187.5. The system compares the corrected grayscale values ​​of adjacent pixels along the image's row scanning direction, identifies the pixel position where the grayscale value reaches its peak, extracts its column index in the image matrix as the horizontal coordinate, and its row index as the vertical coordinate, obtaining a two-dimensional coordinate dataset containing coordinate parameters. This dataset is then aggregated to generate the original pixel set of the umbrella skirt arc surface.

[0023] The light intensity matrix centroid mapping submodule loads the original pixel set of the umbrella skirt arc surface, performs scan line centroid extraction, corrects the slope offset of the light intensity response curve, arranges the grayscale data of multiple frames of exposed pixels into a two-dimensional spatial array, and generates the glaze surface reflective light intensity energy response matrix. Load the original pixel set of the umbrella skirt's curved surface and read the corrected grayscale values ​​corresponding to the coordinate parameters. Traverse the nodes of the original pixel set line by line, extracting pixels that are consecutively arranged on the same scan line and exceed the grayscale background noise threshold. Multiply the horizontal coordinate parameters of each pixel with the corresponding corrected grayscale value and sum them to obtain the grayscale quality moment parameter. Directly sum the corrected grayscale values ​​of all pixels in the row to obtain the total quality parameter. Divide the grayscale quality moment parameter by the total quality parameter to obtain the horizontal coordinate position of the scan line's centroid. For example, dividing the grayscale quality moment parameter 312150 by the total quality parameter 487 yields a horizontal coordinate position of 640.965. Extract the difference in the horizontal coordinate positions of the centroids of adjacent scan lines as the actual response curve slope. Retrieve the theoretical slope parameter of the defect-free standard sample, subtract the theoretical slope parameter from the actual response curve slope to calculate the slope offset, and update the horizontal coordinate values ​​of each pixel by subtracting the slope offset from the calculated centroid horizontal coordinate position. A two-dimensional array containing row and column indices is created. The grayscale data of each frame of exposure obtained by continuous exposure is filled into the corresponding data cells according to the updated coordinates to generate the glaze surface reflective light intensity energy response matrix.

[0024] Specifically, such as Figure 2 , 4 As shown, the umbrella skirt morphology reconstruction module includes: The brightness gradient extreme radius extraction submodule performs adjacent point operations on the brightness sequence based on the glaze surface reflected light intensity energy response matrix, locks the pixel horizontal coordinates corresponding to the positive and negative extreme values ​​of the gradient, extracts and calculates the sub-pixel center, and merges the geometric center to form the glaze surface contour feature chain to obtain the instantaneous extreme radial quantity set. The glaze surface reflectance energy response matrix is ​​read, and the corrected grayscale value sequence under the corresponding row and column indices is retrieved. Subtraction is performed on the corrected grayscale values ​​in adjacent data cells in ascending column indices to obtain the gradient values ​​between pixels. The calculated continuous gradient value sequence is compared with the zero-point reference value. Data nodes whose gradient values ​​change from positive to negative are designated as positive extrema, and data nodes whose gradient values ​​change from negative to positive are designated as negative extrema. The corresponding column scan index of the extrema is recorded as the pixel's horizontal coordinate. The horizontal coordinates of the positive and negative extrema are summed and divided by a constant 2 to obtain the sub-pixel center feature coordinates. The coordinates of the insulator's mechanical rotation geometric center are extracted. The Euclidean distance between the sub-pixel center feature coordinates and the mechanical rotation geometric center coordinates is calculated. The obtained Euclidean distance is multiplied by the magnification coefficient to form the instantaneous radial distance set corresponding to the physical size. For example, adding the x-coordinate of the positive extreme point (645) and the x-coordinate of the negative extreme point (635) and dividing by 2 gives the center coordinate (640). Subtracting the rotation center coordinate (440) gives the distance (200). Multiplying by the magnification (0.226 mm per pixel) gives the instantaneous radial distance parameter (45.2 mm).

[0025] The refractive index dynamic compensation submodule introduces the instantaneous polar radial quantity set, integrates the curvature radius of the umbrella skirt arc surface design with the refractive index of the glass phase matrix, extracts discrete compensation variables by comparing the cosine value of the normal vector deflection, updates the basic calibration matrix, and obtains the dynamic compensation mapping matrix. Discrete compensation variables refer to the calculation of the normal vector deflection cosine value based on the tangent slope of the instantaneous radial quantity set at the sampling point, combined with the curvature radius of the umbrella skirt arc surface, and the optical path offset caused by the refractive index of the glass phase matrix is ​​calculated according to Snell's law. Read the instantaneous polar radius parameters, retrieve the umbrella skirt arc surface design curvature radius parameters from the drawing configuration library, and retrieve the glass phase matrix refractive index parameters from the material property table. Subtract the instantaneous polar radius parameters corresponding to adjacent sampling points, divide the difference by the physical arc length distance between sampling points to obtain the tangent slope of the sampling point, calculate the tangent direction vector based on the tangent slope, and rotate it 90 degrees to obtain the normal vector parameter. Retrieve the standard light source incident direction vector, perform a dot product operation between the normal vector and the incident direction vector, and divide by the product of their magnitudes to obtain the normal vector deflection cosine value. Retrieve the air refractive index reference value, multiply the air refractive index by the sine value of the angle corresponding to the normal vector deflection cosine value, and divide by the glass phase matrix refractive index parameter to obtain the refraction angle sine value. Calculate the optical path offset caused by the glass phase matrix refractive index and assign it to the discrete compensation variable. Retrieve the basic calibration homogeneous transformation matrix, add the corresponding elements of the translation component to the discrete compensation variable, and output the dynamic compensation mapping matrix.

[0026] The spatial coordinate topology mapping submodule calls the dynamic compensation mapping matrix to perform matrix linear transformation on the instantaneous radial quantity set, mapping the pixel coordinates to three-dimensional Euclidean space, performing point cloud density normalization and establishing a spatial topology set including normal vector information, and outputting the physical topology array of the umbrella skirt glaze layer spatial morphology. The system receives a dynamic compensation mapping matrix and performs matrix multiplication with the homogeneous pixel coordinate vectors corresponding to the polar radius parameters in the instantaneous radial vector set. Pixel coordinates in the 2D image coordinate system are converted to physical coordinates in 3D Euclidean space through a multiplication-addition logic between matrix row elements and vector column elements. A spatial data structure node containing coordinate parameters and normal vector parameters is established, and the converted 3D spatial coordinates are written to the corresponding fields. A 3D voxel mesh is built in 3D Euclidean space, and the total number of 3D spatial coordinate points falling within each voxel mesh is counted. The total number of coordinate points within each voxel is compared with a set value for the standard point cloud of a single voxel, and coordinate points exceeding the set value are removed, followed by point cloud density normalization. For example, if a voxel contains 45 coordinate points and the preset standard number of single voxels is 30, 15 coordinate points exceeding the limit are removed. The 3D spatial position parameters of the remaining coordinate points are extracted and combined with the normal vector parameters. A dataset containing vertex sequences and edge relationships is constructed in memory, using 3D coordinates as vertices and normal vectors as connection attributes. This dataset is then output as a physical topology array of the umbrella-shaped glaze layer spatial morphology.

[0027] Specifically, such as Figure 2 , 5 As shown, the defect feature screening module includes: The deep fault threshold locking submodule, based on the physical topology array of the spatial morphology of the umbrella skirt glaze layer, compares the gradient difference in the depth direction of adjacent points to determine candidate spatial coordinate points that exceed the threshold. By evaluating the connected component characteristics of the coordinate distribution in the abnormal area, it records the spatial location index and obtains the suspected deep fault coordinate points. The three-dimensional spatial coordinate data of each vertex is read from the physical topology array of the umbrella-shaped glaze layer. Depth values ​​of adjacent physical coordinate points are extracted along the Z-axis. The depth gradient difference between adjacent points is obtained by subtracting the depth value of the preceding point from the depth value of the subsequent point. Depth abrupt change threshold parameters related to material stress variation are retrieved from the configuration library. The absolute value of the depth gradient difference between adjacent points is compared with the threshold parameter. Points with an absolute difference greater than the threshold parameter are marked as candidate spatial coordinate points. For example, if the absolute depth difference between the preceding and succeeding points is 2.8 mm, which is greater than the abrupt change threshold parameter of 1.5 mm, the succeeding point is marked. All candidate spatial coordinate points are extracted. The Euclidean distance between any two candidate spatial coordinate points is calculated in the coordinate set formed by these points. Points with a distance less than 3 mm are grouped into the same data cluster to form a connected domain of coordinate distribution within the anomalous region. Candidate spatial coordinate points belonging to the same connected domain are extracted, and their array index in the physical topology array is used as a spatial location index and stored in a memory list. The entire set of coordinates in the indexed list is defined as a suspected depth fault coordinate point.

[0028] The geometric centroid physical positioning submodule retrieves the coordinates of suspected depth faults, calculates the arithmetic mean of the coordinate cluster to determine the geometric centroid, and performs depth compensation operation in combination with the glaze layer thickness constant to obtain the coordinates of the defect geometric center. The system retrieves the coordinates of suspected depth fault points and their associated connected data clusters. It extracts the X, Y, and Z-axis coordinates of all points within a data cluster. The values ​​for each of the three axes are summed, and the arithmetic mean of each sum is calculated by dividing the total sum by the total number of points in the data cluster. These three arithmetic means are combined to form the initial coordinates of the geometric centroid of the data cluster. For example, if a data cluster contains four points, the sum of the X-axis values ​​is 48 mm. Dividing this by the total number of points (4) yields an X-axis arithmetic mean of 12 mm. Similarly, the Y-axis arithmetic mean is 20 mm, and the Z-axis arithmetic mean is 16 mm. The system then reads the preset glaze thickness constant from the equipment parameter table and subtracts this constant from the Z-axis arithmetic mean in the initial geometric centroid coordinates to perform depth compensation. For example, if the glaze layer thickness constant is set to 2.5 mm, the initial Z-axis coordinate of the geometric centroid is 16 mm minus 2.5 mm to obtain a compensated Z-axis coordinate of 13.5 mm. The three-axis coordinate parameters updated after depth compensation are packaged and output as the geometric center coordinates of the defect.

[0029] The radiation path energy detection submodule maps the coordinates of the geometric center of the defect, extracts the gray values ​​along the eight-directional radiation path in the energy response matrix of the reflected light intensity of the glaze surface, and combines the reflection intensity and transmittance parameters of the glaze grain boundary to calculate the light intensity energy attenuation at the beginning and end of the radiation path, thus obtaining the feature set of light intensity attenuation of the radiation path at the defect edge. The system receives the coordinate parameters of the defect's geometric center and returns the addressing and matching row and column index nodes of the glaze reflection light intensity energy response matrix generated by the light intensity matrix centroid mapping submodule based on the X and Y axis values. Starting from the addressing and matching node, eight ray traversal paths with an included angle of 45 degrees are generated in the two-dimensional matrix plane along the horizontal, vertical, and diagonal directions. The system extracts the corrected grayscale values ​​stored within the response matrix cell by cell along each traversal path, retrieves the glaze grain boundary reflection intensity parameters and transmittance parameters from the insulating material database, subtracts the grain boundary reflection intensity parameter from the extracted traversal path's beginning corrected grayscale value, and multiplies it with the transmittance parameter to obtain the theoretical emitted light intensity energy. Finally, the system subtracts the path's end corrected grayscale value from the beginning corrected grayscale value and divides the result by the theoretical emitted light intensity energy to calculate the light intensity energy attenuation ratio of the radiation path. For example, the gray value at the beginning of the path is corrected to 180. Subtracting the reflection intensity of 20 and multiplying by the transmittance of 0.9 yields a radiation energy of 144. The gray value difference of 60 divided by 144 gives an attenuation ratio of 0.4167. The light intensity energy attenuation ratio in each direction is bound to the corresponding angle parameters to generate a feature set of light intensity attenuation of the radiation path at the defect edge.

[0030] Specifically, such as Figure 2 , 6 As shown, the glaze light loss quantification module includes: The attenuation difference calculation submodule calculates the unidirectional light intensity attenuation rate and filters out extreme values ​​based on the light intensity attenuation feature set of the radiation path at the defect edge. It determines the isotropic level of light transmission inside the glaze and associates it with the unidirectional absorption path length to obtain the directional attenuation rate sequence. Unidirectional light intensity attenuation rate and extreme value screening refers to extracting the incident and outgoing energy of the sampling path of the light intensity attenuation feature set of the radiation path at the defect edge, calculating the logarithmic loss value of light intensity per unit length of the sampling path, and locking the maximum and minimum values ​​of energy attenuation by comparing the loss values ​​under different spatial dimensions. The system reads data items from the light intensity attenuation feature set of the radiation path at the defect edge, extracts the incident and outgoing energies on each path, performs a natural logarithmic operation on the quotient obtained by dividing the incident energy by the outgoing energy, and divides the logarithmic result by the pixel distance length of the sampling path in the matrix to calculate the logarithmic light intensity loss per unit length of the sampling path. For example, the path incident energy of 180 divided by the outgoing energy of 120 yields a quotient of 1.5, which, after performing a natural logarithmic operation, becomes 0.4. Dividing this by the pixel distance length of 5 yields a light intensity logarithmic loss of 0.08, which is stored as the unidirectional light intensity attenuation rate. The system iterates and compares the unidirectional light intensity attenuation rate array in each direction, extracting the largest and smallest elements as extrema and removing them from the remaining values. The remaining unidirectional light intensity attenuation rate after the removal process is divided by the corresponding unidirectional absorption path length parameter, i.e., the physical distance from the center to the edge of the background region. The calculation results are then arranged according to the original angular attributes to generate a directional attenuation rate sequence.

[0031] The scattering per-directional difference molecular module, based on the directional attenuation rate sequence, introduces the scattering coefficient of the glass phase matrix and the diffuse reflection constant of the micropores of the glaze surface, calculates the isotropic offset and performs weight correction in combination with the lattice arrangement direction, and generates the attenuation anisotropy coefficient. The directional attenuation rate sequence is retrieved, and the maximum and minimum directional attenuation rates are extracted. The difference between the maximum and minimum values ​​is assigned to the isotropic offset. For example, subtracting the minimum directional attenuation rate of 0.015 from the maximum directional attenuation rate of 0.022 yields an isotropic offset of 0.007. The scattering coefficient of the glass phase matrix and the diffuse reflection constant of the micropores in the glaze are imported from the material analysis file. The isotropic offset is multiplied by the scattering coefficient of the glass phase matrix and then added to the diffuse reflection constant of the micropores to obtain the basic anisotropic component. A convolutional neural network architecture including convolutional layers and fully connected layers is established. The input layer receives the image matrix of the local defect region, the convolutional layer extracts the angular gradient features of the lattice arrangement direction, and the fully connected layer uses an activation function to output the convergence probability distribution value. A weight correction parameter is generated based on the lattice arrangement direction order probability value output by the network. For example, the lattice arrangement direction order probability of 0.8 output by the fully connected layer is set as the weight correction parameter 0.8. The basic anisotropic component of 0.013 is multiplied by the weight correction parameter 0.8 to calculate the final value of 0.0104 as the attenuation anisotropic coefficient.

[0032] The energy absorption calculation submodule, based on the attenuation anisotropy coefficient, accumulates the unidirectional light intensity attenuation rate that meets the conditions, obtains the total attenuation accumulation value, compares it with the preset absorption constant, calculates the energy absorption ratio per unit volume of the transparent medium, matches the preset threshold range of bubble characteristics, and performs nonlinear fitting in combination with the local region refractive index fluctuation component to obtain the non-uniformity detection result of enamel light loss. The process begins by obtaining the attenuation anisotropy coefficient. The directional attenuation rate is then compared sequentially with this coefficient, and values ​​greater than the coefficient are summed to obtain the total accumulated attenuation value. Next, a preset absorption constant parameter stored in memory is retrieved. The total accumulated attenuation value is subtracted from the preset absorption constant, and then divided by the unit calibrated volume of the transparent medium to calculate the energy absorption ratio per unit volume. For example, a total accumulated attenuation value of 0.05 minus the preset absorption constant of 0.01 yields 0.04, which, when divided by the unit calibrated volume of 2 cubic millimeters, results in an energy absorption ratio of 0.02 per cubic millimeter. Finally, a preset threshold range parameter for bubble features is retrieved. The calculated energy absorption ratio is substituted into this range for comparison to determine the matching feature's status. For example, an energy absorption ratio of 0.02, falling within the range's lower limit of 0.01 to its upper limit of 0.03, is matched to indicate the presence of a bubble feature. The refractive index fluctuation component of the local area is extracted, and a nonlinear fitting operation is performed by multiplying the fluctuation component with the energy absorption ratio using a quadratic polynomial function. The calculation results are summarized and output as the glaze gloss non-uniformity detection result including the absorption ratio and bubble identification.

[0033] Specifically, such as Figure 2 , 7 As shown, the image detection and evaluation module includes: The feature threshold filtering submodule, based on the non-uniformity detection results of enamel optical damage, starts image detection, compares the bubble feature interval with the preset threshold and performs spatial domain filtering, performs coordinate inverse transformation and calculates the geometric area-to-depth ratio to obtain the coordinate point of the center of the range. Extract the glaze gloss non-uniformity detection result record, read the energy absorption ratio value of the bubble feature state, and perform a double subtraction comparison with the bubble feature interval preset threshold to remove boundary range data. Based on the retained two-dimensional coordinate parameters, construct a median filter window, and calculate the median value of the pixel grayscale value within the matrix coordinate neighborhood to replace the original grayscale value of the center point coordinate. Retrieve the camera intrinsic perspective transformation matrix, and perform a multiplication operation between the two-dimensional coordinates and the inverse of the perspective transformation matrix to obtain the physical plane coordinate set. Count the total number of coordinate sets and multiply by the physical area of ​​a single pixel to calculate the geometric area parameter. Retrieve the corresponding geometric centroid Z-axis depth parameter, and divide the geometric area parameter by the square of the Z-axis depth parameter to obtain the geometric area-depth ratio. For example, the geometric area parameter 15 divided by the square of the depth parameter 5 (25) yields a ratio of 0.6. Extract the set of coordinate points with ratios within the normal feature range, calculate the average value of the horizontal and vertical coordinates, and output the center coordinate point within the range.

[0034] The defect topology redrawing submodule performs image detection and comparison with preset defect space templates based on the center coordinates of the conforming range, fuses and classifies the discrete point set of defect space, calibrates the distribution trajectory of internal microbubbles and analyzes the connectivity features of the point cloud, and generates a defect topology structure set. The system receives coordinates of the center point of the conforming range and retrieves a preset defect space template from the local database. This template includes the side lengths and voxel distribution density parameters of a standard microbubble 3D bounding box. Using the center point of the conforming range as the origin, a measured defect point cloud bounding box is generated in 3D virtual space. The side lengths of each dimension of the measured bounding box are subtracted from the side lengths of the preset defect space template to obtain the size difference value. Measured bounding boxes with size differences less than a threshold are extracted, and their internal discrete point sets are added to a classified defect set, merging to generate a classified defect spatial discrete point set. Using each independent discrete point in the classified defect spatial discrete point set as a node element, the 3D spatial linear Euclidean distance parameter between any two node elements is calculated. Nodes with linear Euclidean distance parameters less than a connectivity threshold are connected by line segments, and the internal microbubble distribution trajectory is calibrated based on the link structure formed by the connections. The total number of connecting line segments for each node is counted, and nodes with a number of connections greater than a set value are defined as core connected nodes. The connectivity characteristics of the point cloud composed of core connected nodes are analyzed, and the corresponding connecting line segments and nodes are aggregated in memory to generate a defect topology set.

[0035] The distribution map integration submodule extracts the image detection index, summarizes the polar diameter parameters and image detection index of the defect topology set, combines the three-dimensional surface mesh parameters in the physical topology array of the umbrella skirt glaze layer spatial morphology, performs a mapping transformation from spatial coordinates to surface unfolding, aggregates the microbubble spatial trajectories, and generates a spatial distribution trajectory map of microbubble defects in insulators. The system acquires a set of defect topologies, extracts the coordinates of core connected nodes within the set and their polar radius parameters mapped to the polar coordinate system, and extracts the associated image detection index number parameters recorded by the feature threshold filtering submodule. All three are then written into a single database record for aggregation. The system calls the umbrella-skirt glaze layer spatial morphology physical topology array data generated by the spatial coordinate topology mapping submodule, extracting the 3D surface mesh parameters recorded within the array. These mesh parameters include the vertex coordinates and normal angles of each mesh triangle face. Based on the cylindrical surface unfolding transformation logic, the X-axis and Y-axis values ​​of the mesh vertex coordinates are substituted into inverse trigonometric functions to obtain the circumferential angle. This circumferential angle is multiplied by the reference cylinder radius parameter to convert it into a planar abscissa position, and the Z-axis value is translated to obtain the planar ordinate position. A spatial coordinate to surface unfolding mapping transformation is then performed. For example, a mesh vertex with X-axis 30 and Y-axis 40 yields a circumferential angle of 53 degrees, which, multiplied by the reference cylinder radius parameter 100, results in a planar abscissa of 92.5, with the Z-axis directly used as the ordinate. Using the same transformation logic, the coordinates of nodes within the defect topology set are transformed to a plane. Based on the connected line segments, the spatial trajectory of microbubbles is aggregated to generate a spatial distribution trajectory map of insulator microbubble defects.

[0036] The above embodiments illustrate preferred embodiments of the present invention. Any equivalent adjustments to the technical solution based on software engineering methods are within the scope of protection, including but not limited to: implementing algorithm logic using different programming languages, refactoring functional modules into services, adjusting data interaction protocols, and optimizing resource scheduling strategies. Any implementation scheme derived from reasonable modifications to the data processing flow, service call chain, or system architecture layer without departing from the core technology of the present invention should be considered within the protection scope defined by the technical solution of the present invention.

Claims

1. A laser detection and evaluation system for the glaze quality of porcelain insulators, characterized in that, The system includes: The glaze photoelectric acquisition module captures the rising edge of the pulse level signal of the photoelectric rotary encoder, matches the laser emitter of the preset trigger reference trigger line structure, and works with the industrial area array camera to capture the reflected laser light strip on the umbrella skirt arc surface. The effective component after extracting the reflectivity correction factor of the insulating layer medium and removing the background energy interference is extracted. The centroid of the scanning line is extracted to generate the glaze reflected light intensity energy response matrix. The umbrella skirt morphology reconstruction module calculates the extreme values ​​of pixel brightness sequence gradient based on the glaze surface reflected light intensity energy response matrix, constructs instantaneous radial quantity, extracts instantaneous normal vector by fusing the curvature radius of the umbrella skirt arc surface and the refractive index of the glass phase matrix, updates the dynamic compensation matrix by matching the cosine value of the reference normal vector deflection, and outputs the physical topology array of the umbrella skirt glaze layer spatial morphology. The defect feature screening module is based on the physical topology array of the spatial morphology of the umbrella skirt glaze layer. It compares the gradient difference in the depth direction of adjacent spatial points, locks the geometric center of the defect outside the preset threshold of the depth fault, extracts the pixel grayscale of the radiation path, and calculates the energy attenuation by combining the reflection intensity and transmittance of the glaze grain boundary to obtain the light intensity attenuation feature set of the radiation path of the defect edge. The glaze optical loss quantification module calculates the optical loss sweep attenuation rate sequence that varies with direction based on the light intensity attenuation feature set of the radiation path at the defect edge, integrates the scattering coefficient of the glass substrate and the scattering constant of the micropores of the glaze surface, extracts the isotropic difference value that satisfies the deformation tolerance, calculates the medium absorption ratio, and outputs the non-uniformity detection result of the glaze optical loss.

2. The laser detection and evaluation system for the glaze quality of porcelain insulators according to claim 1, characterized in that: The glaze surface reflected light intensity energy response matrix includes normalized pixel gray levels, synthetic energy response distribution weights, and pulse level amplitudes after background suppression. The umbrella-skirt glaze layer spatial morphology physical topology array includes point cloud three-dimensional vertex coordinates, local surface normal feature vectors, and spatial topological adjacency weight matrix. The defect edge radiation path light intensity attenuation feature set includes radiation direction energy attenuation rate, defect boundary light intensity contrast, and unit path dielectric loss constant. The non-uniformity detection results of glaze light loss include anisotropic attenuation deviation coefficient, dielectric layer energy absorption rate, and isotropic correction value caused by diffuse reflection.

3. The laser detection and evaluation system for the glaze quality of porcelain insulators according to claim 1, characterized in that, The glaze surface photoelectric acquisition module includes: The rotor circumferential phase synchronization capture submodule counts the rising edge of the output pulse level of the insulator rotating shaft, calculates the time difference with the preset trigger reference, sends the start command to the linear structure laser emitter, and obtains the synchronous control trigger pulse quantity through circumferential phase angle modulus calculation. The arc surface light strip exposure response submodule receives the synchronous control trigger pulse quantity, drives the camera to capture the reflected laser light strip on the arc surface of the umbrella skirt, performs nonlinear mapping by combining the reflectivity correction factor of the insulating layer medium, extracts the two-dimensional coordinates corresponding to the gray-scale extreme points in the width direction of the light strip, and obtains the original pixel set of the arc surface of the umbrella skirt. The light intensity matrix centroid mapping submodule loads the original pixel set of the umbrella skirt arc surface, performs scan line centroid extraction, corrects the slope offset of the light intensity response curve, arranges the grayscale data of multiple frames of exposed pixels into a two-dimensional spatial array, and generates the glaze surface reflective light intensity energy response matrix.

4. The laser detection and evaluation system for the glaze quality of porcelain insulators according to claim 3, characterized in that, The umbrella skirt morphology reconstruction module includes: The brightness gradient extreme radius extraction submodule performs adjacent point operations on the brightness sequence based on the glaze surface reflected light intensity energy response matrix, locks the pixel horizontal coordinates corresponding to the positive and negative extreme values ​​of the gradient, extracts and calculates the sub-pixel center, and merges the geometric center to form the glaze surface contour feature chain to obtain the instantaneous extreme radial quantity set. The refractive index dynamic compensation submodule introduces the instantaneous polar radial quantity set, integrates the curvature radius of the umbrella skirt arc surface design with the refractive index of the glass phase matrix, extracts discrete compensation variables by comparing the cosine value of the normal vector deflection, updates the basic calibration matrix, and obtains the dynamic compensation mapping matrix. The spatial coordinate topology mapping submodule calls the dynamic compensation mapping matrix to perform matrix linear transformation on the instantaneous radial quantity set, maps the pixel coordinates to three-dimensional Euclidean space, performs point cloud density normalization and establishes a spatial topology set including normal vector information, and outputs the physical topology array of the umbrella skirt glaze layer spatial morphology.

5. The laser detection and evaluation system for the glaze quality of porcelain insulators according to claim 4, characterized in that: The discrete compensation variable refers to the optical path offset caused by the refractive index of the glass phase matrix, calculated based on the slope of the tangent at the sampling point of the instantaneous radial quantity set, combined with the curvature radius of the umbrella skirt arc surface design.

6. The laser detection and evaluation system for the glaze quality of porcelain insulators according to claim 4, characterized in that, The defect feature screening module includes: The depth fault threshold locking submodule, based on the physical topology array of the spatial morphology of the umbrella skirt glaze layer, compares the gradient difference in the depth direction of adjacent points to determine candidate spatial coordinate points that exceed the threshold. By evaluating the connected domain characteristics of the coordinate distribution in the abnormal area, it records the spatial location index and obtains the suspected depth fault coordinate points. The geometric centroid physical positioning submodule retrieves the coordinates of the suspected depth fault, calculates the arithmetic mean of the coordinate cluster to determine the geometric centroid, and performs a depth compensation operation in combination with the glaze layer thickness constant to obtain the coordinates of the defect geometric center. The radiation path energy detection submodule maps the coordinates of the geometric center of the defect, extracts the gray values ​​along the eight-directional radiation path in the glaze reflection light intensity energy response matrix, and combines the glaze grain boundary reflection intensity and transmittance parameters to calculate the light intensity energy attenuation at the beginning and end of the radiation path, thus obtaining the feature set of light intensity attenuation of the radiation path at the defect edge.

7. The laser detection and evaluation system for the glaze quality of porcelain insulators according to claim 6, characterized in that, The glaze optical loss quantification module includes: The attenuation difference calculation submodule calculates the unidirectional light intensity attenuation rate and filters out extreme values ​​based on the light intensity attenuation feature set of the radiation path at the defect edge, determines the isotropic level of light transmission inside the glaze, and associates it with the unidirectional absorption path length to obtain the directional attenuation rate sequence. The scattering per-directional difference molecular module, based on the directional attenuation rate sequence, introduces the scattering coefficient of the glass phase matrix and the diffuse reflection constant of the glaze micropores, calculates the isotropic offset and performs weight correction in combination with the lattice arrangement direction, and generates the attenuation anisotropy coefficient. The energy absorption calculation submodule, based on the attenuation anisotropy coefficient, accumulates the unidirectional light intensity attenuation rate that meets the conditions, obtains the total attenuation accumulation value, compares it with the preset absorption constant, calculates the energy absorption ratio per unit volume of the transparent medium, matches the preset threshold interval of bubble characteristics, and performs nonlinear fitting in combination with the refractive index fluctuation component of the local area to obtain the non-uniformity detection result of enamel optical loss.

8. The laser detection and evaluation system for the glaze quality of porcelain insulators according to claim 7, characterized in that, The unidirectional light intensity attenuation rate and extreme value screening refers to extracting the incident and outgoing energy of the sampled path from the light intensity attenuation feature set of the radiation path at the defect edge, calculating the logarithmic loss value of light intensity per unit length of the sampled path, and locking the maximum and minimum values ​​of energy attenuation by comparing the loss values ​​under different spatial dimensions.

9. The laser detection and evaluation system for the glaze quality of porcelain insulators according to claim 1, characterized in that, The system also includes an image detection and evaluation module: Based on the non-uniformity detection results of the glaze optical loss, the image detection and evaluation module extracts the bubble feature interval and performs spatial filtering on the glaze grain boundary morphology. By mapping the umbrella skirt glaze layer topology array, it calibrates the discrete points of microbubbles and generates a spatial distribution trajectory map of microbubble defects in the insulator. The spatial distribution trajectory map of microbubble defects in the insulator includes the coordinates of the centroid of the three-dimensional defect cluster, the geometric connectivity features of the microbubbles, and the surface projection distribution points based on the image detection results.

10. The laser detection and evaluation system for the glaze quality of porcelain insulators according to claim 9, characterized in that, The image detection and evaluation module includes: The feature threshold filtering submodule, based on the non-uniformity detection result of the enamel light loss, starts image detection, compares the bubble feature interval with the preset threshold and performs spatial domain filtering, performs coordinate inverse transformation and calculates the geometric area-to-depth ratio to obtain the coordinate point of the center of the range. The defect topology redrawing submodule performs image detection and preset defect space template comparison based on the center coordinate point of the conforming range, fuses and classifies the discrete point set of defect space, calibrates the distribution trajectory of internal microbubbles and analyzes the connectivity features of point cloud, and generates a defect topology structure set. The distribution map integration submodule extracts the image detection index, summarizes the polar diameter parameters and image detection index of the defect topology set, combines the three-dimensional surface mesh parameters in the physical topology array of the umbrella skirt glaze layer spatial morphology, performs a mapping transformation from spatial coordinates to surface unfolding, aggregates the microbubble spatial trajectories, and generates a spatial distribution trajectory map of insulator microbubble defects.

Images