A glass production line product quality real-time monitoring system and method
By monitoring the glass production line in stages, accurately locating the types of anomalies, calculating anomaly coefficients, and conducting difference analysis, the problem of low efficiency in traditional monitoring is solved, enabling real-time quality monitoring and rapid problem resolution of the glass production line.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- NANTONG XINZHOU GLASS CO LTD
- Filing Date
- 2025-07-04
- Publication Date
- 2026-05-29
AI Technical Summary
Traditional methods of monitoring glass production quality are inefficient, making it difficult to accurately identify the source of serious failures caused by the accumulation of minor anomalies of multiple types, and it is also difficult to accurately assess the impact of anomalies on product quality.
The glass production line is divided into multiple production stages. By comparing the data characteristics of normal and abnormal products, the location points of abnormal types are determined, the type location abnormality coefficient and the comprehensive abnormality coefficient are calculated, and the abnormality difference state analysis is carried out to obtain information on the combined impact judgment.
It enables real-time monitoring of product quality on the glass production line, quickly identifies the source of problems, improves problem-solving efficiency, quantitatively assesses the severity of anomalies, and takes timely measures to reduce quality problems and production losses, thereby improving production efficiency and product quality stability.
Smart Images
Figure CN120807446B_ABST
Abstract
Description
Technical Field
[0001] This invention proposes a real-time monitoring system and method for product quality in glass production lines, which relates to the field of data monitoring technology, specifically to the field of real-time monitoring of product quality in glass production lines. Background Technology
[0002] In the glass manufacturing industry, product quality control is of paramount importance. Traditional production quality monitoring methods are gradually revealing their limitations. For example, the glass production process is complex, and production quality monitoring often relies on manual sampling or single-indicator testing, which is not only inefficient but also prone to missing potential anomalies. Furthermore, it is difficult to efficiently and accurately pinpoint the source of serious malfunctions caused by the accumulation of minor anomalies from multiple sources, and it is challenging to accurately assess the extent of the anomalies' impact on product quality. Summary of the Invention
[0003] This invention provides a real-time product quality monitoring system and method for glass production lines to solve the above-mentioned problems:
[0004] This invention proposes a real-time product quality monitoring system and method for glass production lines, the method comprising:
[0005] S1. Divide the glass production line into multiple production stages, perform product quality difference analysis for each production stage, obtain the abnormality location point information of the stage-collected glass data for multiple abnormal types, and then obtain the stage-collected glass abnormal images.
[0006] S2. Obtain comprehensive abnormal glass images based on the abnormal images collected from the stage glass, obtain the type positioning abnormality coefficient of each abnormality type positioning point, calculate the type positioning abnormality difference, and then obtain the type positioning abnormality sequence information.
[0007] S3. Calculate the comprehensive anomaly coefficient, then obtain the anomaly difference coefficient, perform anomaly difference state analysis, obtain anomaly difference state judgment information, and obtain combined influence judgment information based on the anomaly difference state judgment information.
[0008] Furthermore, the system includes: a stage anomaly analysis module, used to divide the glass production line into multiple production stages, perform product quality difference analysis on each production stage, obtain anomaly type location point information of stage-collected glass data for multiple anomaly types, and then obtain stage-collected glass anomaly images.
[0009] The anomaly analysis module is used to obtain a comprehensive anomaly image based on the anomaly images collected from the stage glass, obtain the anomaly coefficient of each anomaly location point, calculate the anomaly difference of each anomaly location point, and then obtain the anomaly sequence information of each anomaly location point.
[0010] The comprehensive impact determination module is used to calculate the comprehensive anomaly coefficient, then obtain the anomaly difference coefficient, perform anomaly difference state analysis, obtain anomaly difference state determination information, and obtain combined impact determination information based on the anomaly difference state determination information.
[0011] The beneficial effects of this invention are: by dividing the production line into multiple stages and analyzing them, it is possible to accurately locate the types of anomalies that occur in each production stage, quickly identify the source of problems in the production process, and improve the efficiency of problem solving.
[0012] The calculation of the category-specific anomaly coefficient and the comprehensive anomaly coefficient enables a quantitative assessment of anomalies, providing a more intuitive and accurate understanding of their severity.
[0013] By calculating the anomaly difference coefficient and determining the combined effects, we can conduct an in-depth analysis of the interactions and synergistic effects between different types of anomalies. This avoids focusing on individual anomalies while ignoring the overall impact, and allows for a more comprehensive assessment of the impact of anomalies on glass quality.
[0014] It can enable real-time monitoring of product quality on glass production lines, timely detection of abnormalities and early warning, timely implementation of measures to reduce quality problems and production losses caused by abnormalities, and improve production efficiency and product quality stability.
[0015] By analyzing and mining abnormal data, weaknesses and potential problems in the production process can be discovered. Attached Figure Description
[0016] Figure 1 This is a schematic diagram of a method for real-time monitoring of product quality in a glass production line. Detailed Implementation
[0017] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.
[0018] In one embodiment of the present invention, a real-time monitoring system and method for product quality in a glass production line is provided, the method comprising:
[0019] S1. Divide the glass production line into multiple production stages, perform product quality difference analysis for each production stage, obtain the abnormality location point information of the stage-collected glass data for multiple abnormal types, and then obtain the stage-collected glass abnormal images.
[0020] S2. Obtain comprehensive abnormal glass images based on the abnormal images collected from the stage glass, obtain the type positioning abnormality coefficient of each abnormality type positioning point, calculate the type positioning abnormality difference, and then obtain the type positioning abnormality sequence information.
[0021] S3. Calculate the comprehensive anomaly coefficient, then obtain the anomaly difference coefficient, perform anomaly difference state analysis, obtain anomaly difference state judgment information, and obtain combined influence judgment information based on the anomaly difference state judgment information.
[0022] The working principle and technical effects of the above solution are as follows: The glass production line is subdivided into multiple production stages, and product quality difference analysis is carried out for each stage. By comparing the data characteristics of normal and abnormal products, the location point information of multiple abnormal types in each production stage is determined, and abnormal images of glass at each stage are acquired accordingly, realizing accurate location and visualization of abnormal situations at different stages.
[0023] Based on the abnormal images acquired by the stage glass, the type localization anomaly coefficient of each anomaly type location point is extracted, the type localization anomaly difference is calculated, and the anomaly types are sorted according to the size of the difference to generate type localization anomaly sequence information.
[0024] By combining anomaly information from each stage, a comprehensive anomaly coefficient is calculated, reflecting the overall anomaly situation across the entire production line. Further calculations of the anomaly difference coefficient analyze the anomaly difference status, and based on the judgment results, information on the combined impact is obtained. This process, through quantitative analysis, uncovers the interrelationships and combined effects between different anomaly types, comprehensively assessing the impact of anomalies on glass product quality.
[0025] By dividing the production line into multiple stages and analyzing them, it is possible to accurately locate the types of anomalies that occur in each stage of production, quickly identify the source of problems in the production process, and improve the efficiency of problem solving.
[0026] The calculation of the category-specific anomaly coefficient and the comprehensive anomaly coefficient enables a quantitative assessment of anomalies, providing a more intuitive and accurate understanding of their severity.
[0027] By calculating the anomaly difference coefficient and determining the combined effects, we can conduct an in-depth analysis of the interactions and synergistic effects between different types of anomalies. This avoids focusing on individual anomalies while ignoring the overall impact, and allows for a more comprehensive assessment of the impact of anomalies on glass quality.
[0028] It can enable real-time monitoring of product quality on glass production lines, timely detection of abnormalities and early warning, timely implementation of measures to reduce quality problems and production losses caused by abnormalities, and improve production efficiency and product quality stability.
[0029] By analyzing and mining abnormal data, weaknesses and potential problems in the production process can be discovered.
[0030] In one embodiment of the present invention, S1 includes:
[0031] The glass production line is divided into multiple stages to obtain multiple production stages;
[0032] Acquire standard glass data for each stage of the glass production line; glass data indicators include stage product images and other environmental quality data;
[0033] Product data is collected in stages from the glass production line to obtain staged glass data.
[0034] Obtain product quality difference information between standard glass data and glass data collected at different stages;
[0035] Based on the product quality difference information, the glass data collected in the stage is marked with difference points to obtain stage glass difference data with difference point marking information.
[0036] Based on the stage glass difference data, obtain the anomaly type information of the stage-collected glass data, and based on the anomaly type information, obtain the anomaly type location point information of the stage-collected glass data.
[0037] The working principle and technical effect of the above technical solution are as follows: the glass production line is divided into multiple stages according to the process flow, etc.
[0038] For each stage of production, standard glass data for that stage is determined, covering data indicators such as product images and environmental quality.
[0039] Glass data is collected in real-time at each stage of the glass production line, with the data type consistent with standard data to ensure the feasibility of comparative analysis. This primarily includes image information with quality characteristics.
[0040] By comparing the standard glass data and the glass data collected at each stage, we can identify the differences in product quality between the two and clarify the deviation between actual production and the standard.
[0041] Based on product quality difference information, the difference points of the glass data collected at each stage are marked, generating stage glass difference data with difference point marking information, thus making the difference information visible.
[0042] Based on the phased glass difference data, the analysis yielded information on the types of anomalies in the phased glass data, and further determined the location of the anomalies to accurately pinpoint the problem.
[0043] By collecting, comparing, and analyzing data in stages, we can accurately pinpoint the specific stages in the glass production process where abnormalities occur, quickly identify the root cause of the problem, and reduce investigation time.
[0044] Timely detection of product quality differences and anomalies allows for timely adjustments to production processes and parameters, enabling targeted improvement measures and thus enhancing the overall quality of glass products.
[0045] Analyzing production data at each stage can identify unreasonable aspects of the production process, thereby improving production efficiency and resource utilization.
[0046] By leveraging technologies such as data acquisition, analysis, and labeling, intelligent monitoring and management of glass production lines have been achieved, reducing manual intervention and improving the scientific nature and accuracy of management.
[0047] In one embodiment of the present invention, the step of obtaining anomaly type information of the stage-collected glass data based on the stage glass difference data, and obtaining anomaly type location point information of the node-collected glass data based on the anomaly type information, includes:
[0048] The stage glass anomaly data is classified according to the preset anomaly type information to obtain stage-type glass anomaly data;
[0049] Based on the abnormal glass data of the aforementioned stage types, the glass data collected in each stage is divided into abnormal data regions to obtain abnormal regions of each stage type.
[0050] The mass points of the abnormal region of the stage type are obtained as the abnormal type location points of the abnormal region of the stage type; the location points only represent symbolic meaning and do not have actual location meaning.
[0051] The system acquires anomaly location points for multiple stages of glass anomaly regions, and marks the stage-acquired glass data with these anomaly location points to obtain stage-acquired glass anomaly image information.
[0052] The working principle and technical effect of the above technical solution are as follows: Based on pre-defined anomaly type information, the stage glass anomaly data are classified, and data with the same or similar anomaly characteristics are grouped into one category to obtain stage-specific glass anomaly data. The scope of different anomaly types is clearly defined. Anomaly categories include scratches, cracks, streaks, bends, and edge defects, etc.
[0053] Based on stage-specific glass anomaly data, the collected glass data is divided into data regions according to anomaly types, determining the corresponding region in the glass data for each anomaly type, i.e., the stage-specific anomaly region. This allows for precise location of the specific distribution range of anomalies within the glass data.
[0054] Within each stage-type anomaly region, a mass point is selected as the anomaly type location point for that region. Mass points are usually key points that can represent the characteristics of the anomaly region. By determining the mass points, the anomaly location can be marked more concisely (without actual location reference significance).
[0055] This process involves collecting anomaly location points for multiple anomaly types across different stages of glass acquisition, and then marking these points on the acquired glass data. The result is a stage-acquired glass anomaly image with these anomaly location point annotations. This transforms abstract anomaly location information into intuitive image annotations.
[0056] By classifying, dividing, and determining the location of abnormal data, information on abnormal situations that occur during the glass production process can be accurately located, greatly improving the efficiency of problem solving.
[0057] By marking the location points of anomalies on the glass anomaly images acquired in each stage, the anomalies are presented in an intuitive image form, making it easy to quickly understand the distribution and characteristics of the anomalies and obtain key information without complex analysis.
[0058] Classifying and labeling abnormal data by type can establish a systematic abnormal data management system, which facilitates the statistics, analysis and tracing of different types of abnormalities.
[0059] Timely detection of anomalies allows for prompt adjustments to production processes and parameters, enabling targeted measures to resolve issues and thereby improve the production quality and stability of glass products.
[0060] In one embodiment of the present invention, S2 includes:
[0061] Acquire glass anomaly images from all stages, integrate the anomaly type location points from the glass anomaly images from all stages into the glass anomaly image from the final stage, and obtain a comprehensive abnormal glass image.
[0062] Obtain the stage-specific glass anomaly data corresponding to the location point information of each anomaly type in the comprehensive anomaly glass image;
[0063] Calculate the type location anomaly coefficient for each anomaly type location point based on the glass anomaly data of the aforementioned stage types.
[0064] The formula for calculating the anomaly coefficient of the aforementioned type is:
[0065]
[0066] Where loys is the category location anomaly coefficient, S si For the real-time data collected for the i-th type, S ipQ represents the average value of the stage-specific glass anomaly data for the i-th type. i The preset weight data for the i-th category;
[0067] Obtain the maximum value among the anomaly coefficients of multiple anomaly location points to obtain the maximum anomaly value.
[0068] Obtain the difference between the maximum value of the anomaly of each type and the local anomaly coefficient of each type to obtain the local anomaly difference of the type;
[0069] The difference between the location of multiple anomalies is sorted from smallest to largest to obtain the anomaly sequence information.
[0070] The working principle and technical effect of the above technical solution are as follows: Collect glass anomaly images acquired at all stages, integrate the anomaly type location points from each image into the final glass anomaly image acquired at the last stage, forming a comprehensive anomaly glass image. This process converges anomaly information from different production stages into a single image, facilitating overall observation and analysis.
[0071] For each anomaly type location point in the comprehensive anomaly glass image, find its corresponding stage-specific glass anomaly data. Establish a correlation between the location point and the original anomaly data.
[0072] Based on the glass anomaly data for each stage, calculate the type-specific anomaly coefficient for each anomaly location point. Find the maximum value among these coefficients; this maximum value represents the most severe anomaly in the entire production process.
[0073] The difference between the maximum value of each anomaly and the coefficient of each localized anomaly is calculated to obtain the localized anomaly difference. The magnitude of the difference reflects the gap between each anomaly and the most severe anomaly.
[0074] The anomaly differences of multiple anomaly location points are sorted from smallest to largest to form an anomaly sequence. This sequence visually displays the severity ranking of each anomaly point relative to the most severe anomaly.
[0075] By integrating the location points of abnormal images at each stage into a comprehensive image, we can gain a complete and intuitive understanding of the distribution and status of all abnormalities throughout the glass production process, avoiding the omission of key information.
[0076] The calculation of the anomaly coefficient for each category enables a quantitative assessment of the severity of the anomaly, allowing for a more accurate determination of its impact on product quality and the production process.
[0077] Identifying the maximum value of each type of anomaly allows for the rapid identification of the most serious anomalies in the production process, facilitating priority handling and improving problem-solving efficiency.
[0078] By calculating the difference between different types of anomalies and generating sequence information, the priority order of each anomaly point was clarified.
[0079] By combining images of abnormal glass with related anomaly analysis data, a wealth of information is provided for continuous improvement of the production process. In-depth analysis of anomalies can reveal weaknesses in the production process.
[0080] In one embodiment of the present invention, S3 includes:
[0081] The comprehensive anomaly coefficient is calculated by integrating information from the abnormal glass images.
[0082] The formula for calculating the comprehensive anomaly coefficient is as follows:
[0083]
[0084] Where zys is the comprehensive anomaly difference coefficient, N is the total number of anomaly types, and loys i The anomaly coefficient for the i-th category is determined, λ is the correction coefficient with a value range of 0.1-0.3, e is the amount of additional anomaly data detected, and ep is the existing average amount of anomaly data.
[0085] The abnormality coefficient is calculated by combining the type abnormality localization coefficient with the comprehensive abnormality coefficient.
[0086] The coefficient of variation is the sum of all types of abnormality coefficients and the sum of the abnormality coefficients.
[0087] The abnormal difference coefficient is compared with the preset abnormal difference threshold to obtain the abnormal difference comparison result;
[0088] The abnormal difference status is determined based on the abnormal difference comparison results, and abnormal difference status determination information is obtained.
[0089] Information on the combined impact is obtained based on the information on abnormal difference status.
[0090] The working principle and technical effect of the above technical solution are as follows: a comprehensive anomaly coefficient is calculated based on comprehensive abnormal glass image information. This coefficient is a quantitative expression of the overall severity of abnormal situations throughout the entire glass production process.
[0091] By combining the previously calculated anomaly localization coefficients with the comprehensive anomaly coefficient, an anomaly difference coefficient is calculated. This allows for the assessment of discrepancies between the unified data obtained from the two analytical methods when both methods are used to acquire the same data.
[0092] The calculated anomaly coefficient is compared with a pre-set anomaly threshold. The threshold is determined based on factors such as production experience. The comparison yields the anomaly comparison results.
[0093] Based on the comparison results of abnormal differences, the status of abnormal differences is determined. If the abnormal difference coefficient is greater than the threshold, it may be judged as a large abnormal difference, and it is determined whether further analysis is needed; if it is less than the threshold, it is judged as a small abnormal difference, and the production process is relatively stable.
[0094] Based on the information regarding the anomaly status, the combined effects of various anomalies are further analyzed. For example, when multiple anomalies with relatively small anomalies appear simultaneously in a specific area of the glass, they may have a synergistic effect on the overall performance of the glass, leading to more serious quality problems. The comprehensive impact of these anomaly combinations on glass quality is assessed.
[0095] By calculating the comprehensive anomaly coefficient and the anomaly difference coefficient, the abnormal situation in the glass production process can be accurately quantified, the severity and difference of the anomaly can be accurately understood, and the uncertainty of subjective judgment can be avoided.
[0096] By comparing the abnormal difference coefficient with a preset threshold and determining the status, it is possible to promptly identify situations with significant abnormal differences during the production process and provide early warnings of potential quality risks.
[0097] Based on the combined impact assessment information, and taking into account the interaction of multiple anomalies, a more reasonable production adjustment plan can be formulated, such as adjusting process parameters and strengthening quality inspection, in order to improve production efficiency and product quality.
[0098] By continuously monitoring and analyzing abnormal situations, and promptly identifying and addressing discrepancies, the stability of the production process can be maintained, production interruptions and quality fluctuations caused by abnormal situations can be reduced, and the continuity and reliability of production can be improved.
[0099] In one embodiment of the present invention, obtaining combined influence determination information based on abnormal difference state determination information includes:
[0100] When the abnormal difference status determination information is an abnormal difference status, the comprehensive abnormal coefficient is divided into multiple categories of coefficients according to the category-based abnormal sequence to obtain multiple category segmentation coefficients.
[0101] Calculate the ratio of the segmentation coefficient of each category to the corresponding category location anomaly coefficient to obtain the difference ratio within the same category;
[0102] Calculate the difference between the differences between any two species to obtain the species influence difference;
[0103] The influence difference of the aforementioned categories is compared with a preset influence difference threshold to obtain the influence difference comparison result;
[0104] Based on the impact difference comparison results, an abnormal impact is determined for each type to obtain abnormal impact determination information, and combined impact analysis is triggered based on the abnormal impact determination information.
[0105] Information on determining the combined effects is obtained from the combined effect analysis data.
[0106] The working principle and technical effect of the above technical solution are as follows: When the abnormal difference status determination information shows a difference abnormal status, the comprehensive abnormal coefficient is segmented according to the previously generated category positioning abnormal sequence. The category positioning abnormal sequence reflects the priority or severity ranking of different abnormality category positioning points. According to this sequence, the comprehensive abnormal coefficient is divided into multiple category segmentation coefficients, and each segmentation coefficient corresponds to an abnormality type.
[0107] For each category segmentation coefficient, its ratio to the corresponding category localization anomaly coefficient is calculated to obtain the category-specific difference ratio. This ratio reflects the relative contribution or difference of that category of anomaly in the overall anomaly. Categories with anomalies are then processed.
[0108] Calculate the difference between any two difference ratios of the same type, i.e., the type effect difference. By comparing the difference ratios of different types of abnormalities, we can understand the relative differences in their influence on the overall abnormality.
[0109] The difference in impact between categories is compared with a preset threshold, and the abnormal impact of each category is determined based on the comparison result. If the difference in impact between categories exceeds the threshold, it indicates that the impact of that category of anomaly on other categories of anomalies differs significantly from that of other categories of anomalies in the overall anomaly, and the corresponding category can be processed directly.
[0110] Based on the abnormal impact determination information, a combined impact analysis is triggered, which comprehensively considers the interaction and influence between multiple abnormal types, and finally obtains the combined impact determination information.
[0111] By using coefficient segmentation and the calculation of differences within the same category, we can accurately identify the types of anomalies that contribute significantly or show obvious differences in the overall anomalies. This allows us to focus on and address these key anomalies, thereby improving the efficiency of problem-solving.
[0112] The difference in impact between different types of anomalies was calculated and compared with a threshold, enabling a quantitative analysis of the differences in impact between different types of anomalies.
[0113] Triggering combined impact analysis and obtaining combined impact determination information can deeply explore the interaction and synergistic effects between multiple anomaly types, avoiding the situation of focusing only on a single anomaly while ignoring the overall impact, and can more comprehensively assess the impact of anomalies on glass quality.
[0114] Based on the information on the determination of combined effects, the adverse effects of abnormal combinations on glass quality can be reduced, thereby improving production efficiency and product quality.
[0115] In one embodiment of the present invention, the step of determining abnormal impact for each category based on the impact difference comparison result, obtaining abnormal impact determination information, and triggering combined impact analysis based on the abnormal impact determination information includes:
[0116] When the difference in influence between categories is less than or equal to a preset threshold, the category is considered to have low influence.
[0117] When the difference in influence between categories exceeds a preset threshold, the category is determined to have a high influence.
[0118] When the abnormal impact assessment information indicates a low-level impact, a combined impact analysis is triggered.
[0119] When the abnormal impact assessment information indicates a high impact of a certain type, the corresponding type of abnormality is assessed and adjusted.
[0120] When the influence of a species is low, it can be understood that the abnormal factor is not caused by a single species, but by a combination of multiple species with low influence (analyze using two species).
[0121] The working principle and technical effect of the above technical solution are as follows: Based on the comparison between the difference in influence of different types and a preset threshold for the difference in influence, the influence of each anomaly type is determined. If the difference in influence of different types is less than or equal to the preset threshold, the type is determined to have a low influence; if the difference in influence of different types is greater than the preset threshold, the type is determined to have a high influence. This rule clarifies the relative influence of different anomaly types in the overall anomaly.
[0122] When the anomaly impact assessment information shows a low-impact category, a combined impact analysis is triggered. This is because a low impact for a single category may mean that the anomaly is not caused by a single factor, but rather by the combination and synergistic effect of multiple low-impact categories. Therefore, further analysis of the combined effects of these low-impact categories is needed.
[0123] When the anomaly impact assessment information indicates a high-impact category, the corresponding category is directly assessed and adjusted. This indicates that this type of anomaly contributes significantly to the overall anomaly and is a key factor leading to the anomaly problem, requiring priority handling to reduce its adverse impact on glass quality.
[0124] When the influence of a given category is low, it is assumed that the anomalous factor is caused by a combination of multiple low-influence categories (taking two as an example). The cause of the overall anomaly is explained by analyzing the interactions and cumulative effects among these low-influence categories.
[0125] By setting an impact difference threshold and determining the impact of anomalies, high-impact categories that have a significant impact on overall anomalies can be quickly identified, thereby improving the efficiency of problem-solving.
[0126] When a type of low-impact condition is identified, a combined impact analysis is triggered. This can uncover potential problems where the individual impacts are small, but when combined, they may have a significant impact on glass quality, thus avoiding the omission of quality issues due to neglecting these low-impact types.
[0127] For high-impact categories, direct anomaly identification and adjustment enable timely and targeted measures to address the issues. For low-impact categories, combined impact analysis allows for the development of more reasonable handling plans, avoiding blind intervention and improving the accuracy and effectiveness of anomaly handling.
[0128] By classifying and processing anomalies of different degrees of impact, we can more comprehensively solve abnormal problems in the production process, reduce the interference of anomalies on the production process, and improve the stability and continuity of production.
[0129] In one embodiment of the present invention, obtaining combined impact determination information based on combined impact analysis data includes:
[0130] Obtain the sum of the category splitting coefficients for every two categories;
[0131] Calculate the ratio of the sum of the category segmentation coefficients to the sum of the category positioning anomaly coefficients to obtain the combined category difference ratio;
[0132] Calculate the difference in the difference ratio between any two combinations to obtain the combination effect difference;
[0133] The combined effect difference is compared with a preset combined effect threshold to obtain a combined effect comparison result. Based on the combined effect comparison result, the combined effect of each combination type is determined to obtain combined effect determination information.
[0134] Based on the combined influence determination information, abnormal combination types are identified and adjusted.
[0135] The working principle and technical effect of the above technical solution are as follows: obtain the sum of the category segmentation coefficients of each pair of categories, and combine and quantify the contributions of different anomaly categories in the comprehensive anomaly.
[0136] The ratio of the sum of the category segmentation coefficients to the sum of the corresponding category location anomaly coefficients is used to obtain the combined category difference ratio. This ratio reflects the relative degree of difference between different anomaly category combinations in the comprehensive anomaly, comprehensively considering the contribution of each anomaly category within the combination and their characteristics in the original anomaly location.
[0137] Calculate the difference in the difference ratio between each pair of combinations to obtain the combination effect difference. By comparing the difference ratios between different combinations, we can understand the relative impact differences between different abnormal combinations and determine which combinations may have a more significant impact on glass quality.
[0138] The combined effect difference is compared with a preset combined effect difference threshold, and the combined effect of each combination type is determined based on the comparison result. If the combined effect difference exceeds the threshold, it indicates that the combination has a significant difference in effect from other combinations, and there may be special abnormal combination situations.
[0139] Based on the combined impact assessment information, the types of abnormal combinations are determined, identifying which combinations require special attention and handling, and corresponding adjustment measures are taken, such as adjusting process parameters and optimizing production processes, to reduce the adverse effects of abnormal combinations on glass quality.
[0140] By calculating the difference ratio of different anomaly types and the difference in combined effects, we can comprehensively assess the combined impact of different anomaly types on glass quality, avoiding the problem of focusing only on a single anomaly type while ignoring the combined effect.
[0141] By comparing the combined impact difference with a preset threshold, key abnormal combinations that have a significant impact on glass quality can be accurately identified, thus improving the efficiency of problem solving.
[0142] By identifying and adjusting abnormal combinations based on combined impact assessment information, more scientific and reasonable abnormal handling plans can be formulated. Targeted measures can be taken for different abnormal combinations, improving the accuracy and effectiveness of abnormal handling and reducing unnecessary interference with the production process.
[0143] Timely detection and handling of key anomalies can reduce quality defects in glass products and improve product consistency and stability.
[0144] In one embodiment of the present invention, comparing the combined effect difference with a preset combined effect difference threshold to obtain a combined effect difference comparison result, and determining the combined effect of each combination type based on the combined effect difference comparison result to obtain combined effect determination information, includes:
[0145] When the combined effect difference is greater than the preset combined effect difference threshold, the corresponding combined type is judged to have a high combined effect.
[0146] When the combined impact difference is less than or equal to the preset combined impact difference threshold, the corresponding combined type is judged to have low combined impact.
[0147] The working principle and technical effect of the above technical solution are as follows: The calculated combination impact difference is compared numerically with a pre-set combination difference threshold. Based on the comparison result, the degree of influence of each abnormal combination type is determined. The pre-set combination difference threshold is determined comprehensively based on production experience and other factors, and is used to distinguish the high and low boundaries of the impact of different abnormal combinations on glass quality. When the combination impact difference exceeds this threshold, the combination type is determined to be a high-impact combination; when the combination impact difference does not exceed this threshold (i.e., less than or equal to), the combination type is determined to be a low-impact combination.
[0148] It can quickly identify key combinations (high-impact combinations) that may have a significant impact on glass quality from numerous abnormal combinations, improving the efficiency of problem solving and preventing the escalation of quality problems due to untimely handling.
[0149] For abnormal combinations that are determined to have low impact, relatively conventional or simplified handling measures can be taken. For abnormal combinations that have high impact, more resources and efforts should be invested in in-depth analysis and handling to achieve reasonable allocation of resources and improve the precision of production management.
[0150] Develop more targeted handling plans for abnormal combinations with different levels of impact, improve the accuracy and effectiveness of abnormal handling, and reduce unnecessary interference with normal production processes.
[0151] By accurately identifying and promptly addressing abnormal combinations, fluctuations in glass product quality caused by such combinations can be reduced, improving product consistency and stability, and meeting customer requirements for product quality.
[0152] In one embodiment of the present invention, the system includes: a stage anomaly analysis module, used to divide the glass production line into multiple production stages, perform product quality difference analysis on each production stage, obtain anomaly type location point information of stage-collected glass data for multiple anomaly types, and then obtain stage-collected glass anomaly images.
[0153] The anomaly analysis module is used to obtain a comprehensive anomaly image based on the anomaly images collected from the stage glass, obtain the anomaly coefficient of each anomaly location point, calculate the anomaly difference of each anomaly location point, and then obtain the anomaly sequence information of each anomaly location point.
[0154] The comprehensive impact determination module is used to calculate the comprehensive anomaly coefficient, then obtain the anomaly difference coefficient, perform anomaly difference state analysis, obtain anomaly difference state determination information, and obtain combined impact determination information based on the anomaly difference state determination information.
[0155] The working principle and technical effects of the above solution are as follows: The glass production line is subdivided into multiple production stages, and product quality difference analysis is carried out for each stage. By comparing the data characteristics of normal and abnormal products, the location point information of multiple abnormal types in each production stage is determined, and abnormal images of glass at each stage are acquired accordingly, realizing accurate location and visualization of abnormal situations at different stages.
[0156] Based on the abnormal images acquired by the stage glass, the type localization anomaly coefficient of each anomaly type location point is extracted, the type localization anomaly difference is calculated, and the anomaly types are sorted according to the size of the difference to generate type localization anomaly sequence information.
[0157] By combining anomaly information from each stage, a comprehensive anomaly coefficient is calculated, reflecting the overall anomaly situation across the entire production line. Further calculations of the anomaly difference coefficient analyze the anomaly difference status, and based on the judgment results, information on the combined impact is obtained. This process, through quantitative analysis, uncovers the interrelationships and combined effects between different anomaly types, comprehensively assessing the impact of anomalies on glass product quality.
[0158] By dividing the production line into multiple stages and analyzing them, it is possible to accurately locate the types of anomalies that occur in each stage of production, quickly identify the source of problems in the production process, and improve the efficiency of problem solving.
[0159] The calculation of the category-specific anomaly coefficient and the comprehensive anomaly coefficient enables a quantitative assessment of anomalies, providing a more intuitive and accurate understanding of their severity.
[0160] By calculating the anomaly difference coefficient and determining the combined effects, we can conduct an in-depth analysis of the interactions and synergistic effects between different types of anomalies. This avoids focusing on individual anomalies while ignoring the overall impact, and allows for a more comprehensive assessment of the impact of anomalies on glass quality.
[0161] It can enable real-time monitoring of product quality on glass production lines, timely detection of abnormalities and early warning, timely implementation of measures to reduce quality problems and production losses caused by abnormalities, and improve production efficiency and product quality stability.
[0162] By analyzing and mining abnormal data, weaknesses and potential problems in the production process can be discovered.
[0163] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.
Claims
1. A method for real-time monitoring of product quality in a glass production line, characterized in that, The method includes: S1. Divide the glass production line into multiple production stages, perform product quality difference analysis for each production stage, obtain the abnormality location point information of the stage-collected glass data for multiple abnormal types, and then obtain the stage-collected glass abnormality image. S2. Based on the glass anomaly images collected in stages, obtain comprehensive anomaly images, obtain the type location anomaly coefficient of each anomaly type location point, calculate the type location anomaly difference, and then obtain type location anomaly sequence information. S3. Calculate the comprehensive anomaly coefficient, then obtain the anomaly difference coefficient, conduct anomaly difference status analysis, obtain anomaly difference status judgment information, and obtain combined influence judgment information based on the anomaly difference status judgment information. Wherein, S2 includes: Acquire glass anomaly images from all stages, integrate the anomaly type location points from the glass anomaly images from all stages into the glass anomaly image from the final stage, and obtain a comprehensive abnormal glass image. Obtain the stage-specific glass anomaly data corresponding to the location point information of each anomaly type in the comprehensive anomaly glass image; Calculate the type location anomaly coefficient for each anomaly type location point based on the glass anomaly data of the aforementioned stage types. Obtain the maximum value among the anomaly coefficients of multiple anomaly location points to obtain the maximum anomaly value. Obtain the difference between the maximum value of the anomaly of each type and the local anomaly coefficient of each type to obtain the local anomaly difference of the type; The difference between the anomalies in the location of multiple anomalies is sorted from smallest to largest to obtain the anomaly sequence information. Wherein, S3 includes: The comprehensive anomaly coefficient is calculated by integrating information from the abnormal glass images. The abnormality coefficient is calculated by combining the type-specific abnormality coefficient with the comprehensive abnormality coefficient. The abnormal difference coefficient is compared with the preset abnormal difference threshold to obtain the abnormal difference comparison result; The abnormal difference status is determined based on the abnormal difference comparison results, and abnormal difference status determination information is obtained. Information on the combined impact is obtained based on the information on abnormal difference status.
2. The method for real-time monitoring of product quality in a glass production line according to claim 1, characterized in that, S1 includes: The glass production line is divided into multiple stages to obtain multiple production stages; Obtain the standard glass data for each stage of the glass production line; Product data is collected in stages from the glass production line to obtain staged glass data. Obtain product quality difference information between standard glass data and glass data collected at different stages; Based on the product quality difference information, the glass data collected in the stage is marked with difference points to obtain stage glass difference data with difference point marking information. Based on the stage glass difference data, obtain the anomaly type information of the stage-collected glass data, and based on the anomaly type information, obtain the anomaly type location point information of the stage-collected glass data.
3. The method for real-time monitoring of product quality in a glass production line according to claim 2, characterized in that, The step of obtaining anomaly type information of the stage-collected glass data based on the stage glass difference data, and obtaining anomaly type location point information of the node-collected glass data based on the anomaly type information, includes: The stage glass difference data are classified according to the preset anomaly type information to obtain stage-type glass anomaly data; Based on the abnormal glass data of the aforementioned stage types, the glass data collected in each stage is divided into abnormal data regions to obtain abnormal regions of each stage type. The particles in the abnormal region of the stage type are used as the abnormal type location points of the abnormal region of the stage type. The system acquires anomaly location points for multiple stages of glass anomaly regions, and marks the stage-acquired glass data with these anomaly location points to obtain stage-acquired glass anomaly image information.
4. The method for real-time monitoring of product quality in a glass production line according to claim 1, characterized in that, The step of obtaining combined influence determination information based on abnormal difference state determination information includes: When the abnormal difference status determination information is an abnormal difference status, the comprehensive abnormal coefficient is divided into multiple categories of coefficients according to the category-based abnormal sequence to obtain multiple category segmentation coefficients. Calculate the ratio of the segmentation coefficient of each category to the corresponding category location anomaly coefficient to obtain the difference ratio within the same category; Calculate the difference between the differences between any two species to obtain the species influence difference; The influence difference of the aforementioned categories is compared with a preset influence difference threshold to obtain the influence difference comparison result; Based on the impact difference comparison results, an abnormal impact is determined for each type to obtain abnormal impact determination information, and combined impact analysis is triggered based on the abnormal impact determination information. Information on determining the combined effects is obtained from the combined effect analysis data.
5. The method for real-time monitoring of product quality in a glass production line according to claim 4, characterized in that, The step of determining abnormal impact for each category based on the impact difference comparison results, obtaining abnormal impact determination information, and triggering combined impact analysis based on the abnormal impact determination information includes: When the difference in influence between categories is less than or equal to a preset threshold, the category is considered to have low influence. When the difference in influence between categories exceeds a preset threshold, the category is determined to have a high influence. When the abnormal impact assessment information indicates a low-level impact, a combined impact analysis is triggered. When the abnormal impact assessment information indicates a high impact of a certain type, the corresponding type of abnormality is assessed and adjusted.
6. The method for real-time monitoring of product quality in a glass production line according to claim 4, characterized in that, The step of obtaining combined impact determination information based on combined impact analysis data includes: Obtain the sum of the category splitting coefficients for every two categories; Calculate the ratio of the sum of the category segmentation coefficients to the sum of the category positioning anomaly coefficients to obtain the combined category difference ratio; Calculate the difference in the difference ratio between any two combinations to obtain the combination effect difference; The combined effect difference is compared with a preset combined effect threshold to obtain a combined effect comparison result. Based on the combined effect comparison result, the combined effect of each combination type is determined to obtain combined effect determination information. Based on the combined influence determination information, abnormal combination types are identified and adjusted.
7. The method for real-time monitoring of product quality in a glass production line according to claim 6, characterized in that, The step involves comparing the combined impact difference with a preset combined impact threshold to obtain a combined impact comparison result, and then determining the combined impact of each combination type based on the combined impact comparison result to obtain combined impact determination information, including: When the combined effect difference is greater than the preset combined effect difference threshold, the corresponding combined type is judged to have a high combined effect. When the combined impact difference is less than or equal to the preset combined impact difference threshold, the corresponding combined type is judged to have low combined impact.
8. A real-time product quality monitoring system for a glass production line, characterized in that, The system includes: The stage anomaly analysis module is used to divide the glass production line into multiple production stages, perform product quality difference analysis for each production stage, obtain anomaly type location point information of stage-collected glass data for multiple anomaly types, and then obtain stage-collected glass anomaly images. The anomaly analysis module is used to obtain comprehensive anomaly glass images based on the glass anomaly images collected in stages, obtain the anomaly coefficient of each anomaly type location point, calculate the anomaly difference of each anomaly type, and then obtain the anomaly sequence information of each anomaly type. The comprehensive impact determination module is used to calculate the comprehensive anomaly coefficient, then obtain the anomaly difference coefficient, perform anomaly difference state analysis, obtain anomaly difference state determination information, and obtain combined impact determination information based on the anomaly difference state determination information. The anomaly analysis module includes: Acquire glass anomaly images from all stages, integrate the anomaly type location points from the glass anomaly images from all stages into the glass anomaly image from the final stage, and obtain a comprehensive abnormal glass image. Obtain the stage-specific glass anomaly data corresponding to the location point information of each anomaly type in the comprehensive anomaly glass image; Calculate the type location anomaly coefficient for each anomaly type location point based on the glass anomaly data of the aforementioned stage types. Obtain the maximum value among the anomaly coefficients of multiple anomaly location points to obtain the maximum anomaly value. Obtain the difference between the maximum value of the anomaly of each type and the local anomaly coefficient of each type to obtain the local anomaly difference of the type; The difference between the anomalies in the location of multiple anomalies is sorted from smallest to largest to obtain the anomaly sequence information. The comprehensive impact determination module includes: The comprehensive anomaly coefficient is calculated by integrating information from the abnormal glass images. The abnormality coefficient is calculated by combining the type-specific abnormality coefficient with the comprehensive abnormality coefficient. The abnormal difference coefficient is compared with the preset abnormal difference threshold to obtain the abnormal difference comparison result; The abnormal difference status is determined based on the abnormal difference comparison results, and abnormal difference status determination information is obtained. Information on the combined impact is obtained based on the information on abnormal difference status.