Porcelain insulator automatic quality inspection line based on visual defect identification

Abstract

The invention relates to the technical field of visual inspection, in particular to a porcelain insulator automatic quality inspection line based on visual flaw recognition, which comprises an automatic quality inspection center, a quality inspection planning unit, a precondition unit, a quality inspection collaborative visual unit, a repair quantification unit and a quality inspection response unit, according to the method, analysis is preliminarily performed from the angle of historical quality inspection of the quality inspection line, two-point analysis including preliminary efficiency analysis and deep stability and accuracy analysis is performed, and whether the comprehensive quality inspection quality of the quality inspection line is qualified or not is judged, so that the qualification of the comprehensive quality inspection quality of the whole quality inspection line is ensured; according to the method, analysis is carried out from the perspective of prerequisite requirement conditions of a quality inspection line so as to improve the fitness of the prerequisite conditions before quality inspection of the quality inspection line, meanwhile, surface defect analysis is carried out on the porcelain insulator from different stations in an information layer-by-layer progressive mode so as to obtain a defective porcelain insulator model, and deep analysis is carried out based on the defective porcelain insulator model, so that the quality inspection accuracy is improved. And potential safety hazards caused by blind repair are avoided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of visual detection, and particularly relates to a porcelain insulator automatic quality inspection line based on visual flaw identification. BACKGROUND

[0002] Porcelain insulators are important components in power systems, and the quality thereof directly relates to the safe and stable operation of the power system; in the production process of the porcelain insulators, some flaws, such as cracks, glaze defects and spots, are inevitably generated; if the flaws cannot be found and removed in time, the safety hazard will be caused in the use process;

[0003] With the continuous development of machine vision technology, more and more fields begin to use visual detection technology to replace manual inspection, however, the existing porcelain insulator quality inspection based on visual detection still has some problems: it is difficult to analyze the potential quality inspection influence of the quality inspection line, so that the quality inspection quality of the quality inspection line cannot be guaranteed, it is difficult to accurately identify the surface flaws of the porcelain insulator, and it is difficult to accurately position and visualize the existing flaws, so that the sorting and management efficiency of the porcelain insulator is reduced, and it is difficult to quantitatively evaluate the surface repair feasibility of the porcelain insulator, and the blind repair problem exists;

[0004] In view of the above technical defects, a solution is proposed. SUMMARY

[0005] The purpose of the present application is to provide a porcelain insulator automatic quality inspection line based on visual flaw identification to solve the above technical defects, the present application preliminarily analyzes the historical quality inspection of the quality inspection line, and analyzes the preliminary efficiency, the in-depth stability and the accuracy, judges whether the comprehensive quality inspection quality of the quality inspection line is qualified, so as to ensure that the comprehensive quality inspection quality of the whole quality inspection line is qualified, and based on the qualification of the quality inspection line, the prerequisite requirement condition of the quality inspection line is analyzed to judge whether the quality inspection condition of the porcelain insulator is up to standard, so as to adjust the parameters of the quality inspection line according to the information feedback, improve the fitting degree of the prerequisite before the quality inspection of the quality inspection line, and analyze the surface flaws of the porcelain insulator from different stations through the information layer-by-layer progressive mode, obtain a flaw porcelain insulator model, and analyze the porcelain insulator in depth based on the flaw porcelain insulator model, judge whether the porcelain insulator needs to be repaired, avoid the safety hazard caused by blind repair, and realize the optimized use of resources under the premise of ensuring reliability.

[0006] The purpose of the present application can be realized by the following technical scheme: a porcelain insulator automatic quality inspection line based on visual flaw identification, comprising an automatic quality inspection center, a quality inspection planning unit, a prerequisite unit, a quality inspection collaborative visualization unit, a repair quantization unit and a quality inspection response unit.

[0007] An automatic quality inspection center is configured to retrieve historical quality inspection information of the quality inspection line;

[0008] A quality inspection planning unit is configured to analyze the retrieved historical quality inspection information, determine whether the comprehensive quality inspection quality of the quality inspection line is qualified, and obtain a quality inspection management signal or a normal signal;

[0009] A precondition unit is configured to analyze the collected quality inspection basic information of the quality inspection line, determine whether the quality inspection condition of the porcelain insulator is qualified, and obtain a condition compliance signal or a condition defect signal;

[0010] A quality inspection collaborative visual unit is configured to analyze the collected surface image information at different stations, determine whether the porcelain insulator surface has defects, and obtain a defective porcelain insulator model;

[0011] A repair quantification unit is configured to analyze the defective porcelain insulator model, determine whether the porcelain insulator needs to be repaired, and obtain a repair signal or a discard signal.

[0012] Preferably, the analysis process of the quality inspection planning unit is as follows:

[0013] The historical quality inspection information of the quality inspection line is obtained, and the historical quality inspection information includes a quality inspection efficiency average value, a unit time quality inspection quantity curve, and a quality inspection error rate average value;

[0014] The quality inspection efficiency average value is preliminarily discriminated to obtain a warning signal or a feedback signal. When the feedback signal is generated, the difference between the maximum unit time quality inspection quantity and the minimum unit time quality inspection quantity is obtained based on the unit time quality inspection quantity curve, and is set as a regular fluctuation value. The regular fluctuation value and the quality inspection error rate average value are discriminated to obtain a quality inspection management signal or a normal signal.

[0015] Preferably, the analysis process of the precondition unit is as follows:

[0016] The quality inspection basic information of the porcelain insulator is collected, and the quality inspection basic information includes loading data and shooting data;

[0017] The parameters in the quality inspection basic information are discriminated one by one to obtain discrimination output results of each parameter in the quality inspection basic information, and the discrimination output results include qualified and unqualified. The total number of parameters in the quality inspection basic information is set as g, g is a natural number greater than zero, the total number of discrimination output results corresponding to qualified in the quality inspection basic information is obtained, and the total number of discrimination output results corresponding to qualified in the quality inspection basic information is set as a quality inspection condition coefficient. The quality inspection condition coefficient is compared and analyzed with g. If the quality inspection condition coefficient is equal to g, a condition compliance signal is generated. If the quality inspection condition coefficient is not equal to g, a condition defect signal is generated.

[0018] Preferably, the analysis process of the quality inspection collaborative visual unit is as follows:

[0019] S1: The quality inspection period of the porcelain insulator is collected, and the surface image information of the porcelain insulator at different workstations during the quality inspection period is obtained through industrial cameras and light sources set at multiple angles;

[0020] S2: Perform global coordinate calibration on the cameras at each station through the calibration plate;

[0021] S3: Surface defect recognition and analysis is performed on the surface image information at different workstations based on the global coordinate calibration of the workstation camera to obtain the defect type and defect pixel coordinates or sorting signals;

[0022] S4: Based on the generation of the cross-station coordinate conversion matrix, all defective pixel coordinates are converted to the world coordinate system to obtain the world coordinates of the defective pixels;

[0023] S5: Build a 3D model of the porcelain insulator using laser scanning technology or OpenGL, and overlay the 3D model of the insulator with the world coordinates of the defective pixels;

[0024] S6: assigning colors to the defect types, marking the colors corresponding to the defect types on the porcelain insulator 3D model, and thus obtaining a defective porcelain insulator model.

[0025] Preferably, the different workstations include a top inspection workstation, a side inspection workstation, and a bottom inspection workstation.

[0026] Preferably, the specific surface defect identification and analysis process is as follows:

[0027] A convolutional neural network (CNN) algorithm is used for defect recognition. A defect recognition model is obtained by training a large number of porcelain insulator defect images.

[0028] The surface image information is identified by a pre-set defect recognition model to obtain the defect detection results of each workstation. The defect detection results include the presence and absence of defects. The defect detection results are analyzed to obtain the defect type and defect pixel coordinates or sorting signals.

[0029] Preferably, the analysis process of the repair quantification unit is as follows: based on the defective porcelain insulator model, the total number of colors and the area of ​​the color corresponding area are obtained, and the total number of colors and the area of ​​the color corresponding area are set as the defect index and defect repair difficulty respectively, and the total repair cost of the porcelain insulator is obtained at the same time, and the weight factor coefficients of the defect index, defect repair difficulty and total repair cost are pre-set respectively, and the sum of the defect index, defect repair difficulty and total repair cost multiplied by the corresponding weight factor coefficients is set as the defect repair demand.

[0030] Preferably, the flaw repair demand degree is determined, if the flaw repair demand degree is less than a preset flaw repair demand degree threshold, a repair signal is generated, and if the flaw repair demand degree is greater than or equal to the preset flaw repair demand degree threshold, an elimination signal is generated.

[0031] The beneficial effects of the present application are as follows:

[0032] (1) The present application preliminarily analyzes from the historical quality inspection angle of the quality inspection line, is divided into two points of preliminary efficiency and in-depth stability and accuracy analysis, judges whether the comprehensive quality inspection quality of the quality inspection line is qualified, to ensure that the comprehensive quality inspection quality of the whole quality inspection line is qualified, and based on the quality inspection line qualified, analyzes from the prerequisite requirement angle of the quality inspection line, to judge whether the quality inspection condition of the porcelain insulator meets the standard, so as to adjust the parameters of the quality inspection line according to the information feedback, to improve the fitting degree of the prerequisite before the quality inspection of the quality inspection line;

[0033] (2) The present application analyzes the surface flaw of the porcelain insulator from different stations through the layer-by-layer progressive mode, obtains a flaw porcelain insulator model, and in-depth analyzes based on the flaw porcelain insulator model, judges whether the porcelain insulator needs to be repaired, avoids the safety hidden danger caused by blind repair, and realizes the optimized use of resources under the premise of ensuring reliability. BRIEF DESCRIPTION OF DRAWINGS

[0034] The present application will be further described below with reference to the drawings;

[0035] Fig. 1 It is a system flow chart of the present application;

[0036] Fig. 2 It is a local analysis reference diagram of the present application. DETAILED DESCRIPTION

[0037] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0038] In this document, reference to“an embodiment” means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the application. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment, nor are they necessarily mutually exclusive of one another;

[0039] Embodiment one:

[0040] Please refer to Figs. 1-2 As shown in the drawings, the application is an automatic porcelain insulator quality inspection line based on visual flaw identification, which comprises an automatic quality inspection center, a quality inspection planning unit, a pre-condition unit, a quality inspection collaborative visual unit, a repair quantization unit and a quality inspection response unit. The automatic quality inspection center is in bidirectional communication connection with the quality inspection planning unit, and in unidirectional communication connection with the pre-condition unit. The pre-condition unit is in bidirectional communication connection with the quality inspection collaborative visual unit and the repair quantization unit. The quality inspection collaborative visual unit and the repair quantization unit are in unidirectional communication connection with the quality inspection response unit.

[0041] The automatic quality inspection center is used to call the historical quality inspection information of the quality inspection line, and send the historical quality inspection information to the quality inspection planning unit for quality inspection comprehensive evaluation and planning analysis, to judge whether the comprehensive quality inspection quality of the quality inspection line is qualified. The specific quality inspection comprehensive evaluation and planning analysis process is as follows:

[0042] The historical quality inspection information of the quality inspection line is obtained, which includes the average quality inspection efficiency, the unit time quality inspection quantity curve and the average quality inspection error rate.

[0043] The average quality inspection efficiency is preliminarily discriminated. If the quality inspection efficiency is less than the preset quality inspection efficiency threshold value, a warning signal is generated. If the quality inspection efficiency is greater than or equal to the preset quality inspection efficiency threshold value, a feedback signal is generated. When the feedback signal is generated, the difference between the maximum unit time quality inspection quantity and the minimum unit time quality inspection quantity is obtained based on the unit time quality inspection quantity curve, and is set as a regular fluctuation value. The regular fluctuation value and the average quality inspection error rate are discriminated. If the regular fluctuation value is greater than the preset regular fluctuation value threshold value, or the average quality inspection error rate is greater than the preset average quality inspection error rate threshold value, a quality inspection management signal is generated. If the regular fluctuation value is less than or equal to the preset regular fluctuation value threshold value, and the average quality inspection error rate is less than or equal to the preset average quality inspection error rate threshold value, a normal signal is generated. The quality inspection response unit is used to respond to the quality inspection management signal or the normal signal, and immediately make the preset warning operation corresponding to the quality inspection management signal or the normal signal, to ensure that the comprehensive quality inspection quality of the entire quality inspection line is qualified, and to improve the subsequent quality inspection stability and accuracy of the quality inspection line.

[0044] When the normal signal is generated, the pre-condition unit is used to perform pre-condition evaluation analysis on the collected quality inspection basic information of the quality inspection line, to judge whether the quality inspection conditions of the porcelain insulator are up to standard, so as to adjust the parameters of the quality inspection line according to the information feedback, to improve the pre-condition fitting degree of the quality inspection line before quality inspection. The specific pre-condition evaluation analysis process is as follows:

[0045] The quality inspection basic information of the porcelain insulator is collected, and the quality inspection basic information includes loading data, shooting data, etc. The loading data includes loading interval time length, loading speed, etc. The shooting data includes shooting blind area, shooting upload time length, etc.

[0046] The parameters in the quality inspection basic information are processed one by one to obtain the discrimination output results of the parameters in the quality inspection basic information. The discrimination output results include qualified and unqualified. The total number of parameters in the quality inspection basic information is set as g, g is a natural number greater than zero, the total number of discrimination output results corresponding to qualified in the quality inspection basic information is obtained, and the total number of discrimination output results corresponding to qualified in the quality inspection basic information is set as a quality inspection condition coefficient. The quality inspection condition coefficient is compared and analyzed with g. If the quality inspection condition coefficient is equal to g, a condition fitting signal is generated. If the quality inspection condition coefficient is not equal to g, a condition defect signal is generated. The quality inspection response unit is used to respond to the condition fitting signal or the condition defect signal. The preset warning text corresponding to the condition fitting signal or the condition defect signal is immediately displayed on the control panel, so that the parameters of the quality inspection line can be adjusted according to the information feedback to improve the pre-condition fitting degree of the quality inspection line before quality inspection.

[0047] For example, the loading data in the quality inspection basic information is processed one by one. The loading interval time length, loading speed and other data are compared and analyzed with the corresponding preset threshold value. If the data in the loading data are all equal to the preset threshold value, the loading data is judged to be qualified. If the data in the loading data are not all equal to the preset threshold value, the loading data is judged to be unqualified. Similarly.

[0048] Embodiment two:

[0049] When the condition fitting signal is generated: the quality inspection cooperative visual unit is used for comprehensive cooperative quality inspection visual analysis of the collected surface image information at different stations. The specific comprehensive cooperative quality inspection visual analysis process is as follows:

[0050] S1: The quality inspection period of the porcelain insulator is collected. The surface image information of the porcelain insulator at different stations in the quality inspection period is obtained through the industrial cameras and light sources arranged at multiple angles;

[0051] Different stations include top detection station, side detection station, bottom detection station, etc.

[0052] S2: The global coordinate calibration of each station camera is performed through the calibration plate to ensure that the images taken at different stations are accurately spliced in spatial dimension, that is, the spatial coordinates of the multi-station images are unified, and the position consistency of the surface defects of the porcelain insulator in the full-station detection is ensured, which provides accurate data basis for subsequent defect atlas generation and full-surface quality evaluation;

[0053] S3: Based on the global coordinate calibration of the station camera, surface defect recognition and analysis are performed on the surface image information of different stations. The surface image information is recognized by a pre-set flaw recognition model to obtain flaw detection results of each station. The flaw detection results include the presence of flaws and the absence of flaws. If at least one of the flaw detection results of each station is the presence of flaws, the flaw type and flaw pixel coordinates are obtained, and the corresponding station is set as an abnormal station. If the flaw detection results of each station are all the absence of flaws, a sorting signal is generated. The quality inspection response unit responds to the sorting signal to classify and sort the porcelain insulators without flaws;

[0054] The specific surface defect recognition and analysis process is as follows:

[0055] A convolutional neural network (CNN) algorithm is used for flaw recognition. A large number of porcelain insulator flaw images are trained to improve the accuracy of flaw recognition, and a flaw recognition model is obtained;

[0056] The surface image information of different stations is input into the flaw recognition model, and the flaw type and flaw pixel coordinates of each station are obtained through the flaw recognition model. The flaw type includes glaze defects and cracks;

[0057] S4: Based on the cross-station coordinate conversion matrix, all flaw pixel coordinates are converted to the world coordinate system to obtain flaw pixel world coordinates;

[0058] S5: A three-dimensional model of the porcelain insulator is constructed by laser scanning technology or OpenGL, and the three-dimensional model of the insulator is superimposed with the flaw pixel world coordinates;

[0059] S6: Assign colors to the flaw type (e.g., red = crack, yellow = glaze defect), mark the corresponding color of the flaw type on the three-dimensional model of the porcelain insulator, and obtain the flaw porcelain insulator model;

[0060] The quality inspection response unit responds to the flaw porcelain insulator model and immediately displays the flaw porcelain insulator model to intuitively understand the flaw porcelain insulator model of the porcelain insulator. The quality inspection line realizes the upgrade from "independent detection by station" to "full-surface collaborative analysis", significantly improving the completeness and reliability of flaw recognition, and providing visual and traceable digital basis for quality evaluation of porcelain insulators;

[0061] Station camera global coordinate calibration process:

[0062] Preliminary calibration of single-station camera: in the single-station camera field of view, place the calibration board and transform at least 6 different poses (rotation, translation), take 1 image for each pose, obtain the relationship between the size of the calibration board and the camera field of view in the corresponding image of each pose, if the ratio between the size of the calibration board and the camera field of view is greater than or equal to the preset threshold, it is determined that the corresponding pose is qualified, if the ratio between the size of the calibration board and the camera field of view is less than the preset threshold, it is determined that the corresponding pose is not qualified, the number of qualified poses is obtained, if the number of qualified poses is greater than or equal to the preset number threshold, a qualified signal is generated, if the number of qualified poses is less than the preset number threshold, an unqualified signal is generated, and a quality inspection response unit is used to respond to the unqualified signal so as to perform position adjustment on the calibration board according to the unqualified signal;

[0063] When the qualified signal is generated: use OpenCV or other vision libraries to extract the corner points of the checkerboard in the image, and solve the intrinsic matrix (focal length fx, fy, optical center coordinates cx, cy) and distortion coefficient (radial distortion k1, k2, tangential distortion p1, p2) of the station camera by least squares method;

[0064] Further analysis of multi-station camera extrinsic parameters and global coordinate mapping: select the top left corner point of the checkerboard on the calibration board as the origin Ow of the world coordinate system, and take the calibration board plane as the XwYw plane, and the vertical direction as the Zw axis;

[0065] Fix the calibration board on a detection platform of a certain station, the station camera takes an image of the calibration board and calculates its extrinsic parameter matrix [Riti], where i = 1, 2, 3, …, g, thereby establishing a unified mapping conversion between all station camera coordinate systems and the world coordinate system;

[0066] Cross-station coordinate conversion matrix generation: for any pixel point pi in the i-th station image, first use the intrinsic matrix Ki to de-distort to obtain the normalized coordinate p'i, p'i=Ki -1 ×pi;

[0067] Based on p'i and the inverse transformation of the extrinsic parameter matrix, calculate its corresponding world coordinate P, P=Ri -1 ×(p'i-ti);

[0068] The extrinsic parameter matrix [Rt] is a mathematical tool for describing the relative position relationship between the camera coordinate system and the world coordinate system, which consists of two parts:

[0069] Rotation matrix R: a 3x3 orthogonal matrix, representing the rotation angle of the camera coordinate system relative to the world coordinate system, satisfying R T R=I (unit matrix);

[0070] Translation vector t: a 3x1 column vector, representing the position coordinates of the camera optical center in the world coordinate system;

[0071] Physical meaning: the formula for converting point Pw in the world coordinate system to Pc in the camera coordinate system is: Pc = R·Pw + t

[0072] That is, the world coordinates are rotated to the camera view angle by the rotation matrix R, and then offset to the camera optical center position by the translation vector t to ensure that the feature points cover the full range of the camera view.

[0073] Example three:

[0074] The repair quantization unit is used for regional repair quantization and repair evaluation analysis of the flaw porcelain insulator model to obtain the flaw repair demand degree, and the porcelain insulator is judged based on the flaw repair demand degree whether it needs to be repaired. The specific regional repair quantization and repair evaluation analysis process is as follows:

[0075] Based on the flaw porcelain insulator model, the total number of colors and the area of the color corresponding region are obtained, and the total number of colors and the area of the color corresponding region are respectively set as the flaw index and the flaw repair difficulty. At the same time, the total repair cost of the porcelain insulator is obtained, and the weight factor coefficients of the flaw index, the flaw repair difficulty and the total repair cost are respectively set in advance, and the sum of the flaw index, the flaw repair difficulty and the total repair cost multiplied by the corresponding weight factor coefficients is set as the flaw repair demand degree. The flaw repair demand degree is discriminated, if the flaw repair demand degree is less than the preset flaw repair demand degree threshold, a repair signal is generated, if the flaw repair demand degree is greater than or equal to the preset flaw repair demand degree threshold, an elimination signal is generated, and the quality inspection response unit is used to respond to the repair signal or the elimination signal, and immediately make the preset warning operation corresponding to the repair signal or the elimination signal, so as to evaluate the repair difficulty and the repair cost, so as to reasonably manage the porcelain insulator after quality inspection;

[0076] It should be noted that through the repair difficulty and repair cost evaluation, a scientific porcelain insulator repair decision system can be established to avoid safety hazards caused by blind repair, and to realize optimal use of resources under the premise of ensuring reliability; for high-voltage power system key equipment, "no repair" is often a more stable choice, and repair technology is mainly suitable for flaw treatment of low-voltage grade and non-critical parts;

[0077] To sum up, the application preliminarily analyzes the history quality inspection of the quality inspection line, and divides into two points of preliminary efficiency and deep stability and accuracy, judges whether the comprehensive quality inspection quality of the quality inspection line is qualified, to ensure that the comprehensive quality inspection quality of the whole quality inspection line is qualified, and based on the quality inspection line qualified, analyzes the premise requirement condition of the quality inspection line, to judge whether the quality inspection condition of the porcelain insulator meets the standard, so as to adjust the parameters of the quality inspection line according to the information feedback, to improve the fitting degree of the precondition before the quality inspection of the quality inspection line, and at the same time, the surface flaw of the porcelain insulator is analyzed from different stations through the information layer-by-layer progressive mode, the flaw porcelain insulator model is obtained, and the flaw porcelain insulator model is analyzed in depth, to judge whether the porcelain insulator needs to be repaired, to avoid the safety hidden danger caused by blind repair, and to realize the optimized use of resources under the premise of ensuring reliability.

[0078] The setting of the threshold value is for result comparison and analysis, so as to judge whether it is good or bad, and the size of the threshold value is set according to the large model analysis of sample data and artificial experience, and the sample data is also stored, and can be appropriately adjusted through seasonal or reasonable influence condition.

[0079] The size of the coefficient is a specific value obtained by quantifying each parameter, for subsequent comparison, and the size of the coefficient depends on how much sample data and the corresponding running coefficient preliminarily set by the person skilled in the art for each group of sample data.

[0080] The above is only the preferred specific embodiment of the application, but the protection scope of the application is not limited to this, any person skilled in the art in the technical range disclosed by the application, according to the technical scheme and the inventive concept of the application, equivalent replacement or change, should be covered in the protection scope of the application.

Claims

1. The automated quality inspection line for porcelain insulators based on visual defect recognition is characterized by: It includes automated quality inspection center, quality inspection planning unit, prerequisite unit, quality inspection collaborative visualization unit, repair quantification unit and quality inspection response unit; Automated quality inspection center: used to retrieve historical quality inspection information of the quality inspection line; Quality inspection planning unit: used to analyze the retrieved historical quality inspection information, determine whether the comprehensive quality inspection quality of the quality inspection line is qualified, and obtain quality inspection management signals or normal signals; Precondition unit: used to analyze the basic quality inspection information collected from the quality inspection line, determine whether the quality inspection conditions of the porcelain insulators meet the standards, and obtain conditional fitting signals or conditional defect signals; Quality inspection collaborative visual unit: used to analyze the surface image information collected at different workstations, determine whether there are defects on the porcelain insulator surface, and obtain the defective porcelain insulator model; Repair quantification unit: used to analyze the defective porcelain insulator model, determine whether the porcelain insulator needs to be repaired, and obtain a repair signal or an elimination signal.

2. The automated quality inspection line for porcelain insulators based on visual defect recognition according to claim 1, characterized in that: The analysis process of the quality inspection planning unit is as follows: Obtain historical quality inspection information of the quality inspection line, including the average quality inspection efficiency, the quality inspection volume curve per unit time, and the average quality inspection error rate; The mean quality inspection efficiency is preliminarily discriminated and processed to obtain a warning signal or feedback signal. When the feedback signal is generated, the difference between the maximum unit time quality inspection quantity and the minimum unit time quality inspection quantity is obtained based on the unit time quality inspection quantity curve, and it is set as the normal fluctuation value. The normal fluctuation value and the mean quality inspection error rate are discriminated and processed to obtain a quality inspection management signal or a normal signal.

3. The automated quality inspection line for porcelain insulators based on visual defect recognition according to claim 1, characterized in that: The analysis process of the prerequisite unit is as follows: Collect basic quality inspection information of porcelain insulators, including loading data and shooting data; The parameters in the quality inspection basic information are judged and processed one by one, and the judgment output results of each parameter in the quality inspection basic information are obtained. The judgment output results include qualified and unqualified. The total number of parameters in the quality inspection basic information is set to g, where g is a natural number greater than zero. The total number of qualified judgment output results in the quality inspection basic information is obtained, and the total number of qualified judgment output results in the quality inspection basic information is set as the quality inspection condition coefficient. The quality inspection condition coefficient is compared and analyzed with g. If the quality inspection condition coefficient is equal to g, a conditional fit signal is generated. If the quality inspection condition coefficient is not equal to g, a conditional defect signal is generated.

4. The automated quality inspection line for porcelain insulators based on visual defect recognition according to claim 1, characterized in that: The analysis process of the quality inspection collaborative visual unit is as follows: S1: The quality inspection period of the porcelain insulator is collected, and the surface image information of the porcelain insulator at different workstations during the quality inspection period is obtained through industrial cameras and light sources set at multiple angles; S2: Perform global coordinate calibration on the cameras at each station through the calibration plate; S3: Surface defect recognition and analysis is performed on the surface image information at different workstations based on the global coordinate calibration of the workstation camera to obtain the defect type and defect pixel coordinates or sorting signals; S4: Based on the cross-station coordinate conversion matrix generation, all defective pixel coordinates are converted to the world coordinate system to obtain the world coordinates of the defective pixels; S5: Build a 3D model of the porcelain insulator using laser scanning technology or OpenGL, and overlay the 3D model of the insulator with the world coordinates of the defective pixels; S6: assigning colors to the defect types, marking the colors corresponding to the defect types on the porcelain insulator 3D model, and thus obtaining a defective porcelain insulator model.

5. The automated quality inspection line for porcelain insulators based on visual defect recognition according to claim 4, characterized in that: The different workstations include a top inspection workstation, a side inspection workstation, and a bottom inspection workstation.

6. The automated quality inspection line for porcelain insulators based on visual defect recognition according to claim 4, characterized in that: The specific surface defect identification and analysis process is as follows: A convolutional neural network (CNN) algorithm is used for defect recognition. A defect recognition model is obtained by training a large number of porcelain insulator defect images. The surface image information is identified by a pre-set defect recognition model to obtain the defect detection results of each workstation. The defect detection results include the presence and absence of defects. The defect detection results are analyzed to obtain the defect type and defect pixel coordinates or sorting signals.

7. The automated quality inspection line for porcelain insulators based on visual defect recognition according to claim 1, characterized in that: The analysis process of the repair quantification unit is as follows: based on the defective porcelain insulator model, the total number of colors and the area of ​​the color corresponding area are obtained, and the total number of colors and the area of ​​the color corresponding area are set as the defect index and the defect repair difficulty respectively. At the same time, the total repair cost of the porcelain insulator is obtained, and the weight factor coefficients of the defect index, defect repair difficulty and total repair cost are pre-set respectively, and the sum of the defect index, defect repair difficulty and total repair cost multiplied by the corresponding weight factor coefficients is set as the defect repair demand.

8. The automated quality inspection line for porcelain insulators based on visual defect recognition according to claim 7, characterized in that: The defect repair demand is judged and processed. If the defect repair demand is less than a preset defect repair demand threshold, a repair signal is generated. If the defect repair demand is greater than or equal to the preset defect repair demand threshold, an elimination signal is generated.

Images