A method for automatically sorting and removing defects of building toughened glass raw sheet

Abstract

The present application belongs to the technical field of glass manufacturing, and provides a building toughened glass original sheet defect automatic sorting and removing method, which comprises collecting the surface image of the glass original sheet, extracting defect basic characteristic data, shape data and position data; determining the defect influence basic score according to the defect basic characteristic and position data, and calculating the defect influence comprehensive score in combination with the defect roundness shape correction coefficient; simultaneously collecting the current original sheet quantity of the pre-toughening furnace buffer area, and determining the production capacity influence coefficient; weighting and fusing the defect influence comprehensive score and the production capacity influence coefficient to obtain the removing tendency index, and accordingly executing the original sheet automatic sorting and removing; the present application realizes the dynamic optimization of the defect removing decision by fusing the defect shape, edge distance and production line buffer saturation, and maximally reduces the production capacity loss under the premise of ensuring the toughening quality.

Description

Technical Field

[0001] This invention belongs to the field of glass manufacturing technology, and in particular relates to an automatic sorting and rejection method for defects in architectural tempered glass sheets. Background Technology

[0002] During the manufacturing process, tempered glass sheets inevitably produce surface defects such as bubbles, scratches, and impurities. These defects can cause localized stress concentrations during subsequent tempering processes, significantly reducing the mechanical strength and safety performance of the glass, and even leading to accidental breakage of the finished product during use.

[0003] Currently, machine vision systems are widely used in industry for defect detection. Their core logic relies on a pre-set fixed threshold classification mechanism, primarily making rejection decisions based on basic geometric parameters such as defect area and length. While some optimization schemes attempt to introduce defect type weights or simplified location weights, they remain confined to a static rule framework and fail to deeply integrate the physical characteristics of defects with the dynamic state of the production line.

[0004] Specifically, the morphological characteristics of defects, especially their roundness, have a decisive impact on stress distribution during the tempering process. Circular defects can effectively disperse stress and reduce the risk of breakage, while defects with sharp edges are prone to becoming stress concentration points. However, existing technologies only focus on defect size, completely ignoring the dynamic correction effect of morphological factors on the severity of defects. Furthermore, at the production scheduling level, the defect rejection system and the management of the buffer area in front of the tempering furnace have long been disconnected. Rejection decisions are based solely on the static parameters of the defects themselves, failing to perceive the accumulation level in the buffer area in real time. When the buffer area is idle, excessive rejection will directly lead to production line shutdowns and idle capacity, while when the buffer area is close to saturation, insufficient rejection may accumulate quality hazards. This static decision-making model cannot dynamically balance production efficiency while ensuring product quality. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an automatic sorting and rejection method for defects in architectural tempered glass sheets, thus solving the aforementioned problems.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an automatic sorting and rejection method for defects in architectural tempered glass sheets, the method specifically comprising:

[0007] Collect surface images of the original tempered glass sheet for building, and extract basic defect feature data, defect morphology data, and defect location data;

[0008] The comprehensive score of defect impact is determined based on the defect's basic characteristics, morphology, and location data.

[0009] Collect the current quantity of raw glass sheets stored in the buffer area in front of the tempering furnace in the architectural tempered glass production line, and determine the capacity impact coefficient of the architectural tempered glass production line based on the current quantity of raw glass sheets stored in the buffer area.

[0010] The rejection tendency index of architectural tempered glass sheets is determined based on the capacity impact coefficient and defect impact comprehensive score of the architectural tempered glass production line.

[0011] Based on the rejection tendency index of architectural tempered glass sheets, defects in architectural tempered glass sheets are automatically sorted and rejected.

[0012] Based on the above technical solutions, the present invention also provides the following optional technical solutions:

[0013] Further technical solution: The determination of the comprehensive score for the impact of defects specifically includes:

[0014] Based on the defect's basic characteristics and location data, a basic score for the defect's impact is determined; whereby the defect location data refers to the distance between the defect and the edge of the original tempered glass sheet.

[0015] Determine the morphology correction coefficient based on the defect morphology data;

[0016] The comprehensive score of defect impact is determined based on the morphological correction coefficient and the basic score of defect impact.

[0017] Further technical solution: The determination of the impact of defects on the basic score specifically includes:

[0018] Through the formula: ;

[0019] Determine the impact of defects on the baseline score ;

[0020] in, This refers to the normalized value of the i-th basic feature data of the j-th defect. This refers to the weight coefficient of the i-th basic feature of the j-th defect. This refers to the weight coefficient of the j-th defect obtained based on location data, where m refers to the number of defects and n refers to the number of basic feature data of the j-th defect.

[0021] Further technical solution: The method for obtaining the weight coefficient of the j-th defect based on location data specifically includes:

[0022] Through the formula: ;

[0023] Determine the weighting coefficient of the j-th defect obtained based on location data. ;

[0024] in, This refers to the defect type flag value of the j-th defect. This refers to the distance of the j-th defect from the edge of the original tempered glass sheet. This refers to the distance threshold. This refers to the distance of the q-th defect from the edge of the original tempered glass sheet, and m refers to the number of defects.

[0025] Further technical solution: The determination of the morphology correction coefficient specifically includes:

[0026] Through the formula: ;

[0027] Determine the morphological correction coefficient ;

[0028] in, This refers to the roundness of the r-th defect. This refers to the weighting coefficient of the roundness of the r-th defect. This refers to the morphological influence strength coefficient, and m refers to the number of defects.

[0029] Further technical solution: the roundness of the r-th defect The specific methods of obtaining it include:

[0030] Through the formula: ;

[0031] The roundness of generated defects ;

[0032] in, This refers to the area of ​​the r-th defect. This refers to the perimeter of the r-th defect;

[0033] The area of ​​a defect refers to the area of ​​a two-dimensional projected cross section of the defect in the direction parallel to the upper and lower surfaces of the original tempered glass sheet.

[0034] The perimeter of a defect refers to the perimeter of the two-dimensional projection cross-section of the defect in the direction parallel to the upper and lower surfaces of the original tempered glass sheet.

[0035] Further technical solution: The determination of the comprehensive score for defect impact also includes:

[0036] Through the formula: ;

[0037] Determine the comprehensive score of the impact of defects. ;

[0038] in, This refers to the impact of defects on the base score. This refers to the morphological correction factor.

[0039] Further technical solution: The determination of the capacity impact coefficient of the architectural tempered glass production line specifically includes:

[0040] Through the formula: ;

[0041] Determine the capacity impact coefficient of the architectural tempered glass production line ;

[0042] in, This refers to the buffer saturation of the buffer area in front of the tempering furnace in the architectural tempered glass production line, based on the number of original sheets currently stored in the buffer area. This refers to the saturation sensitivity coefficient.

[0043] Further technical solution: The specific method for obtaining the cache saturation S is as follows:

[0044] Through the formula: ;

[0045] The buffer saturation S of the buffer zone in front of the tempering furnace in the architectural tempered glass production line is generated.

[0046] in, This refers to the number of original films currently stored in the cache. This refers to the maximum number of original films that can be stored in the buffer.

[0047] Further technical solution: The determination of the rejection tendency index for architectural tempered glass sheets specifically includes:

[0048] Through the formula: ;

[0049] Determine the rejection tendency index for architectural tempered glass sheets. ;

[0050] in, This refers to the capacity impact coefficient of the architectural tempered glass production line. This refers to the overall score for the impact of defects. This refers to the sensitivity coefficient to the impact of production capacity.

[0051] This invention provides an automatic sorting and rejection method for defects in architectural tempered glass sheets, which has the following advantages compared with the prior art:

[0052] This invention achieves dynamic evaluation and decision-making by comprehensively considering the morphological characteristics, spatial location, and production line buffer status of defects. It effectively solves the problems in the prior art where defect evaluation ignores the influence of morphological factors and location, and the rejection decision is separated from the production line status. It has the advantages of realizing dynamic comprehensive evaluation of defects and optimizing rejection decisions to balance product quality and production efficiency. Attached Figure Description

[0053] Figure 1 This is a flowchart illustrating an automatic sorting and rejection method for defects in architectural tempered glass sheets provided by the present invention.

[0054] Figure 2 This is a schematic diagram of the process S20 provided by the present invention. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0056] The specific implementation of the present invention will be described in detail below with reference to specific embodiments.

[0057] Please see Figure 1 This invention provides an automatic sorting and rejection method for defects in architectural tempered glass sheets, comprising the following steps:

[0058] S10: Collect surface images of the original tempered glass sheet for building, and extract basic defect feature data, defect morphology data, and defect location data;

[0059] Surface images refer to visual information about the surface of tempered glass sheets obtained through optical imaging equipment, which is used for subsequent defect detection and analysis.

[0060] Defect basic characteristic data refers to data that describes the basic attributes of defects, such as the type, size, and quantity of defects. This data is the basis for assessing the harm caused by defects.

[0061] Defect morphology data refers to data that describes the geometric characteristics of defects, such as the roundness and aspect ratio of defects. This data is used to assess the risk of stress concentration caused by defects during the tempering process.

[0062] Defect location data refers to the specific location information of defects on the original tempered glass sheet, such as the distance of the defect from the edge of the sheet. This data affects the impact of the defect on the overall structural integrity of the sheet.

[0063] S20: Determine the comprehensive score of defect impact based on defect basic characteristic data, defect morphology data, and defect location data;

[0064] S30: Collect the current quantity of raw glass sheets stored in the buffer area in front of the tempering furnace in the architectural tempered glass production line, and determine the capacity impact coefficient of the architectural tempered glass production line based on the current quantity of raw glass sheets stored in the buffer area.

[0065] The buffer zone before the tempering furnace refers to the area in the architectural tempered glass production line located before the tempering furnace, used to temporarily store raw glass sheets to be tempered. Its capacity is limited and it plays a regulating role in the continuous operation of the production line.

[0066] S40: Determine the rejection tendency index of raw architectural tempered glass sheets based on the capacity impact coefficient and defect impact comprehensive score of the architectural tempered glass production line;

[0067] S50: Based on the rejection tendency index of architectural tempered glass sheets, automatically sort and reject defects in architectural tempered glass sheets;

[0068] Specifically, in step S10, the surface image can be acquired by setting up an industrial camera on the production line to take pictures at a fixed frequency or when the original sheet passes through a specific position. The acquired images can be stored in the image processing unit. Defect basic feature data, such as the type, size, and number of defects, can be obtained through preliminary image segmentation and feature recognition algorithms. For example, a simple grayscale thresholding method can be used to identify abnormal areas in the image, and their area can be calculated as the defect size. The defect type can be roughly determined by matching a preset shape template. Defect morphology data, such as the shape and contour of the defect, can be obtained through edge detection algorithms. Defect location data, such as the coordinates of the defect on the original sheet, can be determined through the mapping relationship between the image coordinate system and the actual physical coordinate system.

[0069] In step S20, the comprehensive defect impact score aims to quantify the potential impact of defects on the quality and safety performance of the architectural tempered glass sheet. For example, fixed base weights can be assigned to different defect types (such as bubbles, scratches, and impurities), and linear weights can be applied based on the size of the defect. For defect location, the sheet can be simply divided into several regions (such as edge areas and center areas), and each region can be assigned a fixed location weight. Defect morphology data can be simplified into two states: "regular" or "irregular," and the base score can be adjusted accordingly.

[0070] In step S30, the current number of raw wafers stored in the buffer area can be obtained by setting counting sensors at the buffer area entrance and exit, or by manual periodic inspection and recording. After obtaining the current number of raw wafers, it can be compared with the maximum capacity of the buffer area. For example, when the current number exceeds a certain fixed percentage of the maximum capacity, the capacity impact coefficient is set to a lower fixed value, indicating that the raw wafer tempering operation in the production line can operate normally, and the rejection operation will not affect the raw wafer tempering operation (i.e., there is no need to wait for raw glass wafers); conversely, it is set to a higher fixed value, indicating that the raw wafer tempering operation in the production line may not operate normally, and the rejection operation will affect the raw wafer tempering operation (i.e., it may be necessary to wait for raw glass wafers).

[0071] In step S40, the architectural tempered glass sheet rejection tendency index serves as the final decision-making basis for whether to reject the sheet. For example, it can be achieved by simply taking the arithmetic average of the defect impact score and the production capacity impact coefficient, or by summing them using a fixed weighting ratio. When the index exceeds a preset fixed threshold, the sheet is considered to be rejected.

[0072] In step S50, once the rejection tendency index is determined and reaches the preset rejection criteria, the system will trigger the rejection operation. For example, a simple robotic arm or pneumatic pusher can be used to push the defective wafers from the main production line into the waste area. The movement of the robotic arm or pusher can be controlled by a simple relay; when the rejection tendency index exceeds the threshold, the relay closes, driving the actuator to move.

[0073] This invention extracts the basic features, morphological data, and location data of defects, calculates the basic score of defect impact, and introduces a roundness morphology correction coefficient to obtain a comprehensive score of defect impact. At the same time, it collects the number of original sheets in the buffer area, calculates the production capacity impact coefficient, and finally obtains the rejection tendency index by weighted fusion. This method combines the geometric characteristics of defects, edge distance sensitivity, and production line capacity dynamics, avoiding the risk of production capacity loss caused by single threshold rejection, and significantly improving the accuracy of tempered glass sheet sorting and production line operating efficiency.

[0074] For preferred options, please refer to [link / reference]. Figure 2 The present invention further proposes the aforementioned comprehensive score for determining the impact of defects, specifically including:

[0075] S21: Determine the basic score of defect impact based on the defect basic feature data and defect location data; where defect location data refers to the distance of the defect from the edge of the original tempered glass sheet.

[0076] S22: Determine the morphology correction coefficient based on the defect morphology data;

[0077] S23: Determine the comprehensive score of defect impact based on the morphology correction coefficient and the basic score of defect impact;

[0078] The determination of the defect impact baseline score aims to preliminarily quantify the inherent attributes and location sensitivity of defects. Defect basic characteristic data, such as defect type (e.g., bubbles, inclusions, scratches), defect size (e.g., length, diameter, area), and the number of each defect type, directly reflect the physical properties and severity of the defect. Defect location data, especially the distance of the defect from the edge of the tempered glass sheet, is a key factor in assessing defect risk; defects near the edge are more likely to cause stress concentration and breakage during tempering. This score can be achieved by weighting and summing different types, sizes, and numbers of defects and their location information using preset weights and scoring criteria, or by multi-factor evaluation, to obtain an initial numerical value reflecting the basic hazard level of the defect. For example, baseline scores can be set for different defect types and adjusted linearly or non-linearly according to defect size, while introducing an attenuation factor related to the distance from the defect to the edge; the closer the distance, the smaller the attenuation, and even the higher the weight. Another approach is to use a machine learning model, training historical defect data and tempering results to learn the relationship between defect basic characteristics and location and tempering failure rate, thereby predicting the defect impact baseline score.

[0079] Determining the shape correction factor aims to further refine the assessment of defect risk. Considering the impact of defect geometry on tempered glass performance, certain defect shapes, such as sharp, elongated, or irregular defects, are more likely to form stress concentration points during tempering compared to circular or elliptical defects, thus increasing the risk of breakage. The shape correction factor can be determined by analyzing defect shape data, such as geometric parameters like roundness, aspect ratio, and perimeter-to-area ratio. One approach is to set a baseline shape (e.g., circular) and calculate the correction factor based on the degree of deviation of the defect shape from the baseline shape; the greater the deviation, the larger the correction factor. For example, the lower the roundness (the less circular), the higher the correction factor. Another approach is to use image processing techniques to extract features from the defect contour, and then use a classifier to categorize the defect shape into different risk levels, with each level corresponding to a shape correction factor.

[0080] Determining the comprehensive defect impact score involves integrating the basic defect impact score with a morphological correction coefficient to obtain a final assessment value that comprehensively reflects the severity of the defect. This step can be achieved by multiplying the basic defect impact score and the morphological correction coefficient, allowing the morphological correction coefficient to directly amplify or reduce the basic score, thus reflecting the corrective effect of morphology on defect risk. For example, if both the basic score and the morphological correction coefficient are high, the comprehensive score will be significantly higher, indicating that the defect has an extremely high risk. Another approach is to use a weighted average or a more complex fusion algorithm to combine the two, where the weights can be adjusted based on actual production experience or historical data to balance the importance of basic and morphological characteristics in the comprehensive assessment.

[0081] This application's solution decomposes the determination process of the comprehensive defect impact score into a basic score, a shape correction coefficient, and a combination of both, achieving a refined and multi-dimensional assessment of the hazard level of defects in architectural tempered glass sheets. First, through the basic defect impact score, the system can perform preliminary and significant risk quantification of defects based on their inherent attributes such as type, size, and quantity, as well as their critical location on the sheet (distance from the edge). This ensures an accurate grasp of the inherent severity of defects, particularly the identification of high-risk defects in edge areas. Second, the introduction of a shape correction coefficient allows the system to further consider the impact of defect geometry on glass stress distribution. For example, even if two defects have similar basic characteristics, one being sharp and irregular while the other is circular, their risk contributions to the tempering process are drastically different. The introduction of the shape correction coefficient allows this shape difference to be quantified and incorporated into the assessment system. Finally, by combining the basic score with the shape correction coefficient, this method can generate a more comprehensive and accurate comprehensive defect impact score. This step-by-step, progressive evaluation mechanism avoids the limitations of single-indicator evaluation, making the judgment of defects more scientific and reasonable. This provides a more reliable basis for subsequent automatic sorting and rejection, effectively improving the quality control level of the entire architectural tempered glass production line.

[0082] Through the above technical solution, this application can more comprehensively and accurately assess the true degree of harm of defects on the original tempered glass sheet, avoid underestimating the risk of breakage during the tempering process due to only considering basic characteristics, provide a more scientific basis for subsequent automatic sorting and rejection, effectively reduce the tempering failure rate caused by defects, and improve product quality and production efficiency.

[0083] Preferably, the present invention further proposes the aforementioned basic score for determining the impact of defects, specifically including:

[0084] Through the formula: ;

[0085] Determine the impact of defects on the baseline score ;

[0086] in, This refers to the normalized value of the i-th basic feature data of the j-th defect. This refers to the weight coefficient of the i-th basic feature of the j-th defect. This refers to the weight coefficient of the j-th defect obtained based on location data, where m refers to the number of defects and n refers to the number of basic feature data of the j-th defect.

[0087] The Defect Impact Baseline Score aims to quantify all defects detected on the original tempered glass sheet for architectural applications, generating a numerical value that reflects the overall severity of the defects.

[0088] The above formula is used to accurately calculate the basic score of defect impact. Its function is to weight and aggregate the various characteristics and location information of multiple defects to obtain a comprehensive quantitative index.

[0089] Here, the normalized value of the i-th basic feature data of the j-th defect is used. The purpose of normalization is to eliminate the differences in the units and numerical ranges between different basic feature data, to ensure their comparability in the calculation process, and to avoid certain features with large values ​​dominating the scoring results. Normalization can be performed using various methods. For example, the Min-Max normalization method can be used to linearly scale the original data to a specific interval of [0,1]; or the Z-score normalization method can be used to convert the data into a standard normal distribution with a mean of 0 and a standard deviation of 1.

[0090] The weighting coefficient of the i-th basic feature of the j-th defect is used to characterize the relative importance of the i-th basic feature of the j-th defect (e.g., size, quantity) to the quality of the raw tempered glass sheet. Different defect features have different degrees of impact on glass performance. By assigning different weighting coefficients, their true hazards can be reflected more accurately. These weighting coefficients can be obtained by: setting and adjusting them based on the experience of domain experts; or by conducting statistical analysis on a large amount of historical production data and product quality inspection results, and using methods such as regression analysis and principal component analysis for learning and optimization.

[0091] The weighting coefficient of the j-th defect, obtained based on location data, is used to correct the degree of impact of a single defect on the overall quality. This correction is based on the specific location of the defect on the original tempered glass sheet. For example, defects near the glass edge are more likely to cause stress concentration during tempering, leading to breakage; therefore, their weighting coefficient may be higher than that of defects located in the center. This weighting coefficient can be obtained based on a preset distance attenuation function, i.e., the closer to the edge, the greater the weight; or, it can be determined by analyzing the failure probability of defects at different locations based on historical tempering failure data, thereby calibrating the corresponding weighting coefficient.

[0092] Specifically, this invention first extracts multi-dimensional basic feature data for each detected defect and performs normalization processing to eliminate dimensional differences between different features. Then, based on the relative importance of each basic feature's impact on glass quality, a corresponding weighting coefficient is assigned. Furthermore, considering the specific location of the defect on the glass sheet, the overall impact of each defect is corrected using weighting coefficients obtained from location data. Finally, the weighted impact values ​​of all defects are summed to obtain a comprehensive and detailed defect impact baseline score. This calculation method ensures the accuracy and objectivity of the comprehensive defect impact score, avoiding the bias that may arise from simple superposition or averaging, making the quality assessment of the glass sheet more scientific and reasonable. In this way, this application provides a solid data foundation for determining the subsequent rejection tendency index of architectural tempered glass sheets, thereby improving the reliability of the entire sorting and rejection method.

[0093] Through the above technical solutions, this application significantly improves the accuracy and reliability of automatic defect sorting and rejection, helps to reduce unnecessary rejections, lowers production costs, and ensures that the quality of finished products meets tempering requirements, thereby improving the overall intelligence level and economic benefits of the production line.

[0094] Preferably, the present invention further proposes a method for obtaining the weight coefficient of the j-th defect based on location data, specifically including:

[0095] Through the formula: ;

[0096] Determine the weighting coefficient of the j-th defect obtained based on location data. ;

[0097] in, This refers to the defect type flag value of the j-th defect. This refers to the distance of the j-th defect from the edge of the original tempered glass sheet. This refers to the distance threshold. This refers to the distance of the q-th defect from the edge of the original tempered glass sheet, and m refers to the number of defects;

[0098] This formula is used to calculate the weighting coefficient of the j-th defect, which takes into account both the influence of the defect type and its position on the glass substrate.

[0099] Among them, the exponential function can effectively simulate the nonlinear decay characteristic that the closer the defect is to the edge, the greater its influence weight; the summation term in the denominator is used to normalize the distance influence of all defects, ensuring the relative rationality of the weight coefficients of all defects in the overall evaluation.

[0100] The defect type flag value of the j-th defect quantifies the degree of influence of the j-th defect type on the tempering process of the architectural tempered glass sheet. Generally, the defect type with a greater impact on the tempering process has a larger flag value. For example, defect types such as cracks and bubbles, which may lead to stress concentration or breakage, will have a higher flag value than defect types with a smaller impact, such as scratches and stains. This flag value can be set according to industry standards and expert experience, or it can be automatically calibrated by statistical analysis and machine learning model training of historical tempering failure data.

[0101] The distance of the j-th defect from the edge of the tempered glass sheet refers to the straight-line distance from the position of the defect on the surface of the glass sheet to the nearest edge;

[0102] During the tempering process, the stress distribution in the glass edge area is more complex and sensitive. Therefore, defects near the edge are more likely to cause stress concentration and lead to tempering failure. The distance between the defect and the edge of the original tempered glass sheet can be calculated by image processing technology after identifying the defect location, combined with the geometric dimensions of the original glass sheet.

[0103] The distance threshold is a parameter used to adjust the rate at which the influence of a defect on the edge decays. Its value can be set according to the size characteristics of the glass sheet. For example, it can be set to the maximum distance from the center of the sheet to the edge, or it can be optimized based on actual production experience and experimental data. A larger distance threshold means that the influence of a defect on the edge decays more slowly, that is, defects that are farther away still have a certain influence. A smaller distance threshold means that the influence of a defect on the edge decays more quickly, and only defects that are very close to the edge have a significant influence.

[0104] Specifically, this invention no longer relies solely on the basic characteristics of defects, but organically combines the inherent risk of defect type (reflected by defect type indicator values) with the location risk of defects on the glass sheet (reflected by distance from the edge). Through an exponential decay function, the physical law that the closer a defect is to the edge, the greater its potential harm to the tempering process can be captured, and this location sensitivity can be flexibly adjusted through a distance threshold. Simultaneously, the introduction of a normalization term ensures that when multiple defects exist, the weight coefficient of each defect occupies a reasonable proportion in the overall defect impact, avoiding the problem of a single defect having an excessively high or low weight. This weight coefficient, which comprehensively considers defect type and location, allows for a more accurate quantification of the overall defect risk of the glass sheet when substituted into the calculation formula for the defect impact baseline score. Therefore, when this more accurate defect impact baseline score is further used to determine the comprehensive defect impact score and the rejection tendency index of architectural tempered glass sheets, the final automatic defect sorting and rejection decision will be more scientific and reasonable, effectively avoiding misjudgments or omissions caused by inaccurate defect risk assessment.

[0105] Through the above technical solutions, this application improves the reliability of automatic defect sorting and rejection decisions, effectively reduces the tempering failure rate caused by defects, and improves product quality and production efficiency.

[0106] Preferably, the present invention further proposes the determination of the morphological correction coefficient, specifically including:

[0107] Through the formula: ;

[0108] Determine the morphological correction coefficient ;

[0109] in, This refers to the roundness of the r-th defect. This refers to the weighting coefficient of the roundness of the r-th defect. This refers to the morphological influence strength coefficient, and m refers to the number of defects;

[0110] The morphology correction coefficient is a numerical value used to quantify the impact of defect morphology on the glass tempering process. It corrects the basic score of defect impact by taking into account the roundness of the defect and its weight, so as to more accurately reflect the potential risk of the defect. The determination of this coefficient helps to distinguish the degree of harm of defects with different morphologies. For example, a defect with low roundness (irregular shape) will receive a larger correction coefficient than a defect with high roundness (more round shape), thereby improving its overall defect impact score.

[0111] The roundness of the r-th defect is a geometric parameter that measures how close the defect shape is to a circle. The closer the roundness value is to 1, the closer the defect shape is to a circle; the smaller the value, the more irregular or elongated the defect shape is.

[0112] The roundness can be obtained through image processing techniques, such as calculating the area and perimeter of the defect. For example, edge detection algorithms can be used to identify the defect boundary, and then the area it encloses and the length of the boundary can be calculated; another approach is to process the defect region through morphological operations (such as opening and closing operations), and then calculate its geometric properties based on the processed region.

[0113] The weighting coefficient for the roundness of the r-th defect is used to represent the relative importance of the roundness of different defects when calculating the shape correction coefficient. For example, for certain types of defects (such as cracks and scratches), their roundness may have a more critical impact on the tempering process, and therefore can be assigned a higher weight. This weighting coefficient can be trained and calibrated based on historical data, expert experience, or through machine learning models. For example, different weights can be set for the roundness of different defects based on factors such as defect type and defect size.

[0114] The morphology influence intensity coefficient is a parameter that adjusts the degree of influence of the morphology correction coefficient on the comprehensive score of defects. The larger the coefficient, the more significant the influence of the defect morphology on the final score. The calibration of this coefficient usually requires statistical analysis and optimization in combination with a large amount of historical tempering failure data to ensure that it can accurately reflect the risk of defect morphology in actual production. For example, the reasonable range of values ​​for this coefficient can be determined by analyzing the probability of tempering failure caused by different morphological defects.

[0115] Specifically, the potential impact of defect morphology on the tempering process is quantified by calculating the roundness of each defect and combining it with its weighting coefficient. Roundness directly reflects the regularity of the defect shape; the lower the roundness, the more irregular the defect shape, and the higher the risk of stress concentration during tempering. By weighted summing of these roundness values ​​and combining them with a morphology influence intensity coefficient, a morphology correction coefficient is obtained. This morphology correction coefficient is then used to correct the basic defect impact score, resulting in a more comprehensive and accurate overall defect impact score. This mechanism ensures that when assessing defect risk, not only the size and location of the defect are considered, but also its shape characteristics, enabling the overall defect impact score to more accurately reflect the potential threat of defects to the quality and safety of tempered glass. In this way, this solution can more precisely distinguish the severity of different defects, providing a more reliable basis for subsequent automatic sorting and rejection, thus effectively solving the problem of inaccurate defect risk assessment relying solely on basic characteristics and location data.

[0116] Through the above technical solution, the present invention can more effectively identify defects that are poorly shaped and easily lead to tempering failure, thereby guiding more accurate automatic sorting and rejection of defects, reducing the scrap rate of tempered glass, and improving product quality and production efficiency.

[0117] Preferably, the present invention further proposes a roundness of the r-th defect. The specific methods of obtaining it include:

[0118] Through the formula: ;

[0119] The roundness of generated defects ;

[0120] in, This refers to the area of ​​the r-th defect. This refers to the perimeter of the r-th defect;

[0121] The area of ​​a defect refers to the area of ​​a two-dimensional projected cross section of the defect in the direction parallel to the upper and lower surfaces of the original tempered glass sheet.

[0122] The perimeter of a defect refers to the perimeter of the two-dimensional projected cross section of the defect in the direction parallel to the upper and lower surfaces of the original tempered glass sheet.

[0123] The roundness of a defect is a dimensionless parameter used to quantify how close its shape is to an ideal circle. The closer the value is to 1, the closer the shape is to a circle; the smaller the value, the further the shape deviates from a circle, such as exhibiting elongated or irregular features.

[0124] Circularity is a key indicator for assessing the potential stress concentration effect of defects, because non-circular defects (such as sharp cracks and slender inclusions) usually cause greater damage to the structural integrity of tempered glass than circular defects of the same area. Circularity can be calculated using defect area and perimeter obtained by image processing algorithms, or it can be quantitatively measured by processing defect morphology data extracted by a dedicated shape analysis module.

[0125] The area of ​​the r-th defect refers to the size of the two-dimensional projection of the defect. It is a basic feature for evaluating the overall scale of the defect. The area can be determined by calculating the number of pixels occupied by the defect in the captured image and multiplying it by the actual area represented by each pixel, or by accurately depicting the defect boundary through advanced image segmentation technology and then using numerical integration or polygon area calculation methods.

[0126] Similarly, the perimeter of the r-th defect refers to the length of the two-dimensional projection boundary of the defect. The perimeter can be calculated by tracking the boundary pixels of the defect in the image and accumulating the line segment lengths between adjacent boundary pixels, or by using an edge detection algorithm to identify the defect contour and then applying a curve length calculation algorithm to determine it.

[0127] The two-dimensional projection cross-section of the defect onto the upper and lower surfaces of the original tempered glass sheet defines the plane for measuring the defect's area and perimeter. By focusing on the projection parallel to the glass surface, this method captures the appearance of the defect when viewed from above or below, which directly relates to how the defect affects light transmission, visual quality, and surface stress distribution during the tempering process. This projection is typically obtained naturally when capturing surface images using a camera placed perpendicular to the glass surface. For defects with complex three-dimensional structures, advanced imaging techniques such as structured light or confocal microscopy can be used to reconstruct the three-dimensional shape, and then the two-dimensional projection onto the glass surface plane can be obtained through calculation.

[0128] This application's solution defines the roundness of defects using a standard mathematical formula. This formula takes the defect's area and perimeter as inputs, both parameters being derived from the two-dimensional projection of the defect onto the glass surface in a direction parallel to the glass surface. By precisely defining the measurement methods for area and perimeter, this application's solution ensures that the calculated roundness accurately reflects the visual and structural characteristics of defects related to the glass tempering process. For example, defects with lower roundness values ​​(such as elongated cracks or inclusions with sharp edges) indicate a higher likelihood of stress concentration during tempering, thus increasing the risk of glass breakage. Conversely, defects with higher roundness values ​​(close to 1) are generally less dangerous for the same area. By providing a precise and objective measure of roundness, this application's solution enables the shape correction coefficient to more accurately reflect the true risk posed by different defect shapes, thereby improving the reliability of the comprehensive defect impact score and optimizing the overall defect sorting and rejection decisions. This allows the system to distinguish between defects of similar size but with significantly different potential failure risks, achieving more effective and efficient quality control.

[0129] Through the above technical solution, the present invention can help the system to more accurately assess the impact of defects of different shapes on the performance of tempered glass, and avoid misjudgment caused by inaccurate defect shape assessment. For example, high-risk non-circular defects may be misjudged as low-risk, or low-risk circular defects may be misjudged as high-risk. This optimizes the automatic sorting and rejection decision of defects in architectural tempered glass sheets and improves the effectiveness of product quality control.

[0130] Preferably, the present invention further proposes the method of determining the comprehensive score of defect impact, which also includes:

[0131] Through the formula: ;

[0132] Determine the comprehensive score of the impact of defects. ;

[0133] in, This refers to the impact of defects on the base score. This refers to the morphological correction factor;

[0134] in, By modifying the form of this basic risk and multiplying the two, the impact of various attributes of the defect on the tempering process can be fully considered.

[0135] Furthermore, the formula introduces the min(1,) function, which means that regardless of and The magnitude of the calculated result will limit the final defect impact score to within 1. This limiting mechanism ensures that the score has an upper limit, keeping it within an acceptable range and preventing it from losing its practical reference value due to excessively high scores in extreme cases.

[0136] The Defect Impact Assessment Score is a comprehensive risk assessment of all defects on the original tempered glass sheet for architectural applications. This score considers multiple dimensions, including defect type, size, quantity, location, and shape, aiming to provide a quantitative indicator for subsequent defect sorting and rejection decisions.

[0137] Specifically, after acquiring surface images of the raw architectural tempered glass sheet and extracting defect basic feature data, defect morphology data, and defect location data, the system first calculates a basic defect impact score based on the defect basic feature data and defect location data. This score quantifies the inherent risk of the defect and its distribution risk on the raw sheet. Simultaneously, the system calculates a morphology correction coefficient based on the defect morphology data, which reflects the potential impact of the defect shape on tempering performance. Then, these two scores are multiplied to obtain a preliminary comprehensive evaluation value. To ensure the rationality and comparability of this comprehensive evaluation value in practical applications, this scheme further introduces the min(1,) function to limit the final comprehensive defect impact score to within 1. This mechanism effectively integrates multiple attribute information of defects, forming a unified and bounded risk assessment index, avoiding overall score distortion due to excessively high values ​​from a single factor, and providing a more accurate and reliable basis for subsequent automatic sorting and rejection of defects in the raw architectural tempered glass sheet. In this way, this scheme can more comprehensively and accurately assess the defect risk of the raw architectural tempered glass sheet, thereby optimizing the sorting and rejection strategy.

[0138] Through the above technical solution, this application can provide a standardized, easy-to-understand and accurate risk assessment basis for subsequent automatic sorting and rejection decisions of defects in architectural tempered glass sheets, thereby improving the accuracy and efficiency of sorting and reducing the risk of tempering failure due to defects.

[0139] Preferably, the present invention further proposes the determination of the capacity impact coefficient of the architectural tempered glass production line, specifically including:

[0140] Through the formula: ;

[0141] Determine the capacity impact coefficient of the architectural tempered glass production line ;

[0142] in, This refers to the buffer saturation of the buffer area in front of the tempering furnace in the architectural tempered glass production line, based on the number of original sheets currently stored in the buffer area. This refers to the saturation sensitivity coefficient;

[0143] The capacity impact coefficient of the architectural tempered glass production line is used to quantify the influence of defective glass sheet rejection decisions on the current production line status. When the buffer stock is tight, this coefficient can encourage the system to retain more defective sheets with less impact; conversely, when the buffer stock is plentiful, it tends to reject more defective sheets. This coefficient can be a dimensionless value between 0 and 1, where 1 represents the minimum capacity impact (i.e., ample stock, strict rejection is possible), and 0 represents the maximum capacity impact (i.e., tight stock, tendency to retain).

[0144] Buffer saturation is an indicator that measures the storage status of the buffer area in front of the tempering furnace in a tempered glass production line. It reflects the degree to which the buffer area is occupied. This indicator can standardize the comparison of buffer areas with different capacities and more accurately assess the pressure on the buffer area. Buffer saturation can be expressed as the ratio of the number of original sheets currently stored in the buffer area to the maximum number of original sheets that the buffer area can store. Alternatively, the buffer saturation S can also be obtained by performing piecewise function mapping or fuzzy logic processing on the number of original sheets currently stored in the buffer area to reflect the contribution of different quantity ranges to the saturation.

[0145] The saturation sensitivity coefficient is an adjustable parameter used to control the sensitivity of the buffer saturation to the capacity impact coefficient. This coefficient determines the rate and magnitude of change in the capacity impact coefficient when the buffer saturation S changes. The saturation sensitivity coefficient can be calibrated and optimized using historical production data, expert experience, or machine learning algorithms to adapt to the characteristics and management strategies of different production lines. Furthermore, the saturation sensitivity coefficient can be dynamically adjusted according to the specific operating mode of the production line (e.g., whether to pursue high output or high quality) to achieve a more flexible rejection strategy.

[0146] Specifically, this method calculates the capacity impact coefficient using an exponential function: when the cache saturation is high, it indicates sufficient wafer reserves in the cache, and the probability of wait time for wafers in the tempering process due to rejection decisions is low. At this time, the capacity impact coefficient approaches 0, indicating a small impact on capacity, and the system can be more inclined to strictly reject defective wafers. As the cache saturation decreases, the wafer reserves in the cache gradually decrease, and the capacity impact coefficient increases exponentially, indicating a higher probability of wait time for wafers in the tempering process due to rejection decisions. The system tends to relax rejection criteria to avoid production line stagnation or reduced efficiency. The saturation sensitivity coefficient acts as a regulating factor, and its value determines the steepness of the change in the capacity impact coefficient with cache saturation. A larger saturation sensitivity coefficient means the system is more sensitive to changes in cache saturation; even a slight increase in cache saturation will cause the capacity impact coefficient to drop rapidly, thus quickly adjusting the rejection strategy to cope with capacity pressure. A smaller saturation sensitivity coefficient indicates that the system is less sensitive to changes in cache saturation, and the change in the capacity impact coefficient is more gradual. This exponential relationship can more accurately simulate the nonlinear relationship between production capacity pressure and buffer zone status in actual production. This allows the determined production capacity impact coefficient to more precisely and realistically reflect the actual operating conditions of the production line, thus providing a more reliable basis for determining the rejection tendency index of architectural tempered glass sheets. This, in turn, optimizes automatic defect sorting and rejection decisions. In this way, the method effectively solves the technical problem that the quantity of sheets alone cannot accurately reflect production capacity pressure, making rejection decisions more intelligent and flexible.

[0147] Through the above technical solution, the present invention enables the automatic sorting and rejection method for defects in architectural tempered glass sheets to better adapt to the real-time operation of the production line, achieve a dynamic balance between production efficiency and product quality, and improve the intelligence and flexibility of sorting and rejection decisions.

[0148] Preferably, the present invention further proposes a specific method for obtaining the cache saturation S as follows:

[0149] Through the formula: ;

[0150] The buffer saturation S of the buffer zone in front of the tempering furnace in the architectural tempered glass production line is generated.

[0151] in, This refers to the number of original films currently stored in the cache. This refers to the maximum number of original films that the buffer can store;

[0152] The maximum number of raw glass sheets that the buffer zone can hold refers to the upper limit of the number of raw glass sheets that the buffer zone in front of the tempering furnace in the architectural tempered glass production line can accommodate in its design. This value is usually fixed or preset and represents the physical capacity of the buffer zone. The number can be determined directly by obtaining the design drawings or specifications of the buffer zone, or by estimating it by actually measuring the physical dimensions of the buffer zone and combining it with the standard dimensions of the raw glass sheets.

[0153] Specifically, this application provides an effective means of quantifying production line output status by clearly defining the method for obtaining buffer saturation, namely, calculating it using the ratio of the currently stored number of original wafers to the maximum number of original wafers that can be stored. When the number of original wafers in the buffer increases, the buffer saturation rises accordingly, indicating that the buffer is close to full capacity, and its capacity will not experience idle waiting time in the tempering furnace due to wafer rejection. Conversely, when the number of currently stored original wafers decreases, the buffer saturation decreases, indicating that the buffer is close to empty capacity, and its capacity will experience idle waiting time in the tempering furnace due to wafer rejection. This calculation method based on the actual quantity ratio allows the buffer saturation to directly and accurately reflect the real-time operating status of the production line, providing a reliable input for the subsequent calculation of the capacity impact coefficient. In this way, the present invention can dynamically adjust the rejection tendency for defective original wafers according to the actual buffer situation in the tempering furnace buffer, avoiding excessive rejection when the buffer wafer reserve is tight, or retaining too many defective original wafers when the buffer wafer reserve is sufficient, thereby optimizing production efficiency and product quality.

[0154] Through the above technical solution, the present invention can more accurately balance the relationship between the impact of defects and the production line capacity, avoid over-rejection or under-rejection caused by inaccurate capacity assessment, thereby optimizing the automatic sorting and rejection decision of defective originals and improving production efficiency and resource utilization.

[0155] Preferably, the present invention further proposes the method of determining the rejection tendency index of architectural tempered glass sheets, specifically including:

[0156] Through the formula: ;

[0157] Determine the rejection tendency index for architectural tempered glass sheets. ;

[0158] in, This refers to the capacity impact coefficient of the architectural tempered glass production line. This refers to the overall score for the impact of defects. This refers to the sensitivity coefficient to the impact of production capacity;

[0159] The architectural tempered glass sheet rejection tendency index is a comprehensive quantitative indicator used to measure the likelihood or priority of a sheet of architectural tempered glass being rejected by the automated sorting system. Its value directly reflects the tendency of the sheet to be rejected under current production conditions, considering the severity of its defects and the production line's capacity. This index is a dimensionless value, and its range is usually normalized to between 0 and 1, or other preset intervals, to facilitate subsequent decision-making. For example, when the index is above a certain threshold, the system will trigger a rejection operation; when it is below the threshold, the sheet is allowed to enter the subsequent production process.

[0160] The capacity impact coefficient of a building tempered glass production line reflects the degree to which the current capacity status of the production line affects the decision to reject raw glass sheets; its value is usually related to the busyness of the production line, that is, the more idle the production line is, the greater the capacity impact coefficient. The smaller the value, the lower the rejection tendency index for tempered glass sheets, indicating that the production line has a higher tolerance for rejecting sheets and is more inclined to retain them to maintain production efficiency; conversely, the busier the production line, the smaller the capacity impact coefficient. The higher the value, the greater the tendency index for rejecting original tempered glass sheets. This indicates that the production line has a lower tolerance for rejecting original sheets and is more inclined to reject defective original sheets in order to ensure product quality.

[0161] The capacity impact sensitivity coefficient is a weighted parameter used to adjust the relative importance of the capacity impact coefficient and the comprehensive score of defect impact in the architectural tempered glass sheet rejection tendency index. Its value typically ranges from 0 to 1. A larger capacity impact sensitivity coefficient indicates that the rejection decision places greater emphasis on the production line's capacity status, i.e., production efficiency takes priority; a smaller capacity impact sensitivity coefficient indicates that the severity of defects in the sheet is given greater emphasis, i.e., product quality takes priority. By adjusting this coefficient, the relationship between production efficiency and product quality can be flexibly balanced according to actual production needs.

[0162] Specifically, this method weights and integrates two key factors: the capacity impact coefficient and the comprehensive defect impact score. The capacity impact sensitivity coefficient serves as an adjustable weight, allowing the system to flexibly adjust the relative priority of production efficiency and product quality in rejection decisions based on actual production strategies. When the remaining wafers in the production line's buffer are scarce, the capacity impact coefficient may be high. In this case, a higher capacity impact sensitivity coefficient will lower the rejection tendency index, thereby reducing wafer rejections and ensuring production continuity. Conversely, when wafer defects are severe, the comprehensive defect impact score may be high. A higher value will increase the rejection tendency index, thus prioritizing the rejection of that original piece to ensure product quality. This fusion mechanism makes rejection decisions no longer a single-dimensional judgment, but rather a comprehensive consideration of the dynamic changes in the production site and the inherent requirements of product quality, thereby achieving more intelligent and optimized automated sorting and rejection.

[0163] Through the above technical solutions, the present invention significantly improves the intelligence level and rationality of automatic sorting and rejection of architectural tempered glass sheets, and optimizes resource allocation and product quality control in the production process.

[0164] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for automatically sorting and rejecting defects in architectural tempered glass sheets, characterized in that, The method specifically includes: Collect surface images of the original tempered glass sheet for building, and extract basic defect feature data, defect morphology data, and defect location data; The comprehensive score of defect impact is determined based on the defect's basic characteristics, morphology, and location data. Collect the current quantity of raw glass sheets stored in the buffer area in front of the tempering furnace in the architectural tempered glass production line, and determine the capacity impact coefficient of the architectural tempered glass production line based on the current quantity of raw glass sheets stored in the buffer area. The rejection tendency index of architectural tempered glass sheets is determined based on the capacity impact coefficient and defect impact comprehensive score of the architectural tempered glass production line. Based on the rejection tendency index of architectural tempered glass sheets, defects in architectural tempered glass sheets are automatically sorted and rejected.

2. The method for automatic sorting and rejection of defects in architectural tempered glass sheets according to claim 1, characterized in that, The determination of the comprehensive score for the impact of defects specifically includes: Based on the defect's basic characteristics and location data, a basic score for the defect's impact is determined; whereby the defect location data refers to the distance between the defect and the edge of the original tempered glass sheet. Determine the morphology correction coefficient based on the defect morphology data; The comprehensive score of defect impact is determined based on the morphological correction coefficient and the basic score of defect impact.

3. The method for automatic sorting and rejection of defects in architectural tempered glass sheets according to claim 1, characterized in that, The determination of the impact of defects on the base score specifically includes: By formula: ; Determine the impact of defects on the baseline score ; in, This refers to the normalized value of the i-th basic feature data of the j-th defect. This refers to the weight coefficient of the i-th basic feature of the j-th defect. This refers to the weight coefficient of the j-th defect obtained based on location data, where m refers to the number of defects and n refers to the number of basic feature data of the j-th defect.

4. The method for automatic sorting and rejection of defects in architectural tempered glass sheets according to claim 3, characterized in that, The method for obtaining the weight coefficient of the j-th defect based on location data specifically includes: By formula: ; Determine the weighting coefficient of the j-th defect obtained based on location data. ; in, This refers to the defect type flag value of the j-th defect. This refers to the distance of the j-th defect from the edge of the original tempered glass sheet. This refers to the distance threshold. This refers to the distance of the q-th defect from the edge of the original tempered glass sheet, and m refers to the number of defects.

5. The method for automatic sorting and rejection of defects in architectural tempered glass sheets according to claim 1, characterized in that, The determination of the morphological correction coefficient specifically includes: By formula: ; Determine the morphological correction coefficient ; in, This refers to the roundness of the r-th defect. This refers to the weighting coefficient of the roundness of the r-th defect. This refers to the morphological influence strength coefficient, and m refers to the number of defects.

6. The method for automatic sorting and rejection of defects in architectural tempered glass sheets according to claim 5, characterized in that, The roundness of the r-th defect The specific methods of obtaining it include: By formula: ; The roundness of generated defects ; in, This refers to the area of ​​the r-th defect. This refers to the perimeter of the r-th defect; The area of ​​a defect refers to the area of ​​a two-dimensional projected cross section of the defect in the direction parallel to the upper and lower surfaces of the original tempered glass sheet. The perimeter of a defect refers to the perimeter of the two-dimensional projected cross-section of the defect in the direction parallel to the upper and lower surfaces of the original tempered glass sheet.

7. The method for automatic sorting and rejection of defects in architectural tempered glass sheets according to claim 1, characterized in that, The determination of the comprehensive score for the impact of defects also includes: By formula: ; Determine the comprehensive score of the impact of defects. ; in, This refers to the impact of defects on the base score. This refers to the morphological correction factor.

8. The method for automatic sorting and rejection of defects in architectural tempered glass sheets according to claim 1, characterized in that, The determination of the capacity impact coefficient of the architectural tempered glass production line specifically includes: By formula: ; Determine the capacity impact coefficient of the architectural tempered glass production line ; in, This refers to the buffer saturation of the buffer area in front of the tempering furnace in the architectural tempered glass production line, based on the number of original sheets currently stored in the buffer area. This refers to the saturation sensitivity coefficient.

9. The method for automatic sorting and rejection of defects in architectural tempered glass sheets according to claim 8, characterized in that, The specific method for obtaining the cache saturation S is as follows: By formula: ; The buffer saturation S of the buffer zone in front of the tempering furnace in the architectural tempered glass production line is generated. in, This refers to the number of original films currently stored in the cache. This refers to the maximum number of original films that can be stored in the buffer.

10. The method for automatic sorting and rejection of defects in architectural tempered glass sheets according to claim 1, characterized in that, The determination of the rejection tendency index for architectural tempered glass sheets specifically includes: By formula: ; Determine the rejection tendency index for architectural tempered glass sheets. ; in, This refers to the capacity impact coefficient of the architectural tempered glass production line. This refers to the overall score for the impact of defects. This refers to the sensitivity coefficient to the impact of production capacity.

Images