Liquid crystal display module production line whole-process quality traceability method and system

Abstract

The application provides a liquid crystal display module production line whole-process quality traceability method and system, relates to the technical field of data processing, and comprises the following steps: step 1, collecting multi-source heterogeneous data, and carrying out cleaning and format unification processing on the multi-source heterogeneous data, establishing the association between product serial numbers and various parameter information, and forming structured whole-process data records; step 2, based on the structured whole-process data records, real-time monitoring is carried out on a core process parameter set, when any parameter is detected to be out of a preset specification range, the product serial number produced in the corresponding abnormal period is marked as a suspicious product, and a warning signal is generated. Through whole-process data structured integration, real-time monitoring and early warning, five-edge topology monitoring and multi-parameter interactive analysis, the application establishes the association between process parameters and display unevenness and light leakage defects, efficiently deduces process optimization compensation, and realizes accurate traceability of the quality of the liquid crystal display module production line, defect risk prevention and control and product quality stability improvement.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and system for tracing the quality of the entire process of a liquid crystal display module production line. Background Technology

[0002] In the field of LCD module manufacturing, quality traceability methods are of great significance for improving product yield and reliability. In existing technologies, quality traceability usually relies on discrete monitoring and analysis of some key parameters in the production process. However, this method may have certain limitations. For example, due to the diverse sources of production data, including equipment sensors, manual records, and other aspects, the data formats and structures are often inconsistent, making it difficult to integrate data throughout the entire process and form a complete product quality traceability chain. In addition, existing monitoring systems mostly focus on threshold alarms for individual parameters. When there are complex interactions between multiple process parameters, the system may not be able to accurately identify the root cause of quality defects, thereby affecting the accuracy and efficiency of process optimization.

[0003] For example, in LCD module production lines, when defects such as bright spots or dark spots appear, existing methods may only be able to adjust individual parameters, but cannot fully analyze the synergistic changes between multiple parameters. This makes the optimization process dependent on experience and the compensation effect unstable. Therefore, existing technologies still need improvement in handling the integration of multi-source heterogeneous data and the analysis of multi-parameter interactions. Summary of the Invention

[0004] The technical problem to be solved by this invention is to provide a method and system for full-process quality traceability of liquid crystal display module production lines, so as to realize dynamic adjustment of process parameters and effectively reduce quality defects such as uneven display and light leakage.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0006] Firstly, a method for full-process quality traceability in a liquid crystal display module production line, the method comprising:

[0007] Step 1: Collect multi-source heterogeneous data, clean and unify the format of the multi-source heterogeneous data, establish the association between product serial number and various parameter information, and form a structured full-process data record;

[0008] Step 2: Based on structured full-process data recording, the core process parameter set is monitored in real time. When any parameter is detected to exceed the preset specification range, the product serial number produced in the corresponding abnormal period is marked as a suspicious product, and an early warning signal is generated.

[0009] Step 3: Based on the full-process data records of suspicious items and historical defective items, configure the first to fifth process units and construct a pentagonal topology monitoring structure;

[0010] Step 4: Based on the core process parameter set data collected by the pentagonal topology monitoring structure, establish the correspondence between process parameter changes and the occurrence of quality defects, and determine the benchmark parameter points based on the distribution characteristics of the monitoring data of the five process units to establish the benchmark parameter status.

[0011] Step 5: Using the baseline parameter state as a reference point, construct two process parameter transmission links, calculate the corresponding orientation difference values, and define the multi-parameter interaction domain.

[0012] Step 6: Based on the multi-parameter interaction domain, analyze the monitoring data of each process unit in the pentagonal topology monitoring structure, construct a closed-loop feedback path, and extract the variation law of the core process parameter set by calculating the intersection characteristics of the closed-loop feedback path and the multi-parameter interaction domain; derive the process parameter optimization compensation amount according to the variation law to adjust the core process parameter set.

[0013] Secondly, the full-process quality traceability system for LCD module production lines includes:

[0014] The data acquisition module is used to collect multi-source heterogeneous data, clean and unify the format of the multi-source heterogeneous data, establish the association between product serial number and various parameter information, and form a structured full-process data record.

[0015] The marking module is used to monitor the core process parameter set in real time based on structured full-process data records. When any parameter is detected to be outside the preset specification range, the product serial number produced in the corresponding abnormal period is marked as a suspicious product and an early warning signal is generated.

[0016] The configuration module is used to configure the first to fifth process units and construct a pentagonal topology monitoring structure based on the full-process data recording of suspicious items and historical defective items.

[0017] A module is established to collect core process parameter set data based on the pentagonal topology monitoring structure, establish the correspondence between process parameter changes and quality defect occurrence, and determine the baseline parameter points based on the distribution characteristics of the monitoring data of the five process units to establish the baseline parameter status.

[0018] The definition module is used to construct the transmission link between two process parameters with the baseline parameter state as a reference point, calculate the corresponding orientation difference value, and define the interaction domain of multiple parameters.

[0019] The adjustment module is used to analyze the monitoring data of each process unit in the pentagonal topology monitoring structure based on the multi-parameter interaction domain, construct a closed-loop feedback path, and extract the variation law of the core process parameter set by calculating the intersection characteristics of the closed-loop feedback path and the multi-parameter interaction domain; and derive the process parameter optimization compensation amount based on the variation law to adjust the core process parameter set.

[0020] Thirdly, a computing device includes:

[0021] One or more processors;

[0022] A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method.

[0023] Fourthly, a computer-readable storage medium storing a program that, when executed by a processor, implements the method.

[0024] The above-described solution of the present invention has at least the following beneficial effects:

[0025] By establishing the correlation between product serial numbers and various parameters and using a unified data format, a complete end-to-end data record is formed, breaking down data silos and building a coherent quality traceability chain. It overcomes the limitations of single-parameter monitoring by monitoring core process parameters in real time and marking suspicious products during abnormal periods, enabling early warning of quality risks, timely curbing the mass production of defective products, and reducing resource waste. The five-sided topology monitoring structure covers key process units, achieving systematic monitoring of critical links throughout the entire production process, rather than isolated monitoring of individual workstations or parameters, thus more comprehensively capturing parameter change characteristics during the process. It establishes a correspondence between process parameter changes and quality defects. By establishing relationships and baseline parameter states, the correlation logic between parameter fluctuations and quality issues is clarified, changing the current situation where it is difficult to locate the root cause of defects in existing technologies and improving the accuracy of quality traceability. By constructing process parameter transmission links and defining the interaction domain of multiple parameters, the problem of analyzing complex interactions of multiple parameters is effectively solved, enabling the identification of synergistic influence patterns between parameters and avoiding the one-sidedness caused by relying solely on experience. Based on the interaction domain of multiple parameters and closed-loop feedback path, the parameter change patterns are extracted, and the process parameter optimization compensation amount is derived to achieve dynamic adjustment of process parameters, effectively reducing quality defects such as uneven display and light leakage, and improving product yield and reliability. Attached Figure Description

[0026] Figure 1 This is a flowchart illustrating the full-process quality traceability method for liquid crystal display module production lines provided in an embodiment of the present invention.

[0027] Figure 2 This is a schematic diagram of a full-process quality traceability system for a liquid crystal display module production line provided in an embodiment of the present invention. Detailed Implementation

[0028] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0029] like Figure 1 As shown, embodiments of the present invention propose a method for full-process quality traceability in liquid crystal display module production lines, the method comprising the following steps:

[0030] Step 1: Collect multi-source heterogeneous data, clean and unify the format of the multi-source heterogeneous data, establish the association between product serial number and various parameter information, and form a structured full-process data record;

[0031] Step 2: Based on structured full-process data recording, the core process parameter set is monitored in real time. When any parameter is detected to exceed the preset specification range, the product serial number produced in the corresponding abnormal period is marked as a suspicious product, and an early warning signal is generated.

[0032] Step 3: Based on the full-process data records of suspicious items and historical defective items, configure the first to fifth process units and construct a pentagonal topology monitoring structure;

[0033] Step 4: Based on the core process parameter set data collected by the pentagonal topology monitoring structure, establish the correspondence between process parameter changes and the occurrence of quality defects, and determine the benchmark parameter points based on the distribution characteristics of the monitoring data of the five process units to establish the benchmark parameter status.

[0034] Step 5: Using the baseline parameter state as a reference point, construct two process parameter transmission links, calculate the corresponding orientation difference values, and define the multi-parameter interaction domain.

[0035] Step 6: Based on the multi-parameter interaction domain, analyze the monitoring data of each process unit in the pentagonal topology monitoring structure, construct a closed-loop feedback path, and extract the variation law of the core process parameter set by calculating the intersection characteristics of the closed-loop feedback path and the multi-parameter interaction domain; derive the process parameter optimization compensation amount according to the variation law to adjust the core process parameter set.

[0036] In this embodiment of the invention, by establishing the association between product serial numbers and various parameters and using a unified data format, a complete end-to-end data record is formed, breaking down data silos and constructing a coherent quality traceability chain. It overcomes the limitations of single-parameter monitoring by real-time monitoring of core process parameter sets and marking suspicious products during abnormal periods, achieving early warning of quality risks, timely curbing the mass production of defective products, and reducing resource waste. The five-sided topology monitoring structure covers key process units, enabling systematic monitoring of key links in the entire production process, rather than isolated monitoring of a single workstation or parameter, thus more comprehensively capturing parameter change characteristics during the process. It establishes a correlation between process parameter changes and quality defects. The study clarifies the correlation between parameters and quality issues by establishing the corresponding relationships and baseline parameter states, thereby changing the current situation where it is difficult to locate the root cause of defects in existing technologies and improving the accuracy of quality traceability. By constructing process parameter transmission links and defining the interaction domain of multiple parameters, the study effectively solves the problem of analyzing complex interactions between multiple parameters, enabling the identification of synergistic influence patterns between parameters and avoiding the one-sidedness caused by relying solely on experience. Based on the interaction domain of multiple parameters and closed-loop feedback paths, the study extracts the parameter change patterns, derives the process parameter optimization compensation amount, and realizes the dynamic adjustment of process parameters, effectively reducing quality defects such as uneven display and light leakage, and improving product yield and reliability.

[0037] In a preferred embodiment of the present invention, step 1 includes:

[0038] Step 100: Obtain the product serial number from the production order source, the core process parameter set from the process parameter source, the material batch information from the material source, and the quality defect information from the quality inspection source, including uneven display defect information and light leakage defect information; wherein the core process parameter set includes chip glass bonding temperature, chip glass bonding pressure, chip glass bonding time, cell thickness parameters, liquid crystal volume parameters, and environmental temperature and humidity parameters, specifically including:

[0039] Through the order association query function of the production order management system, based on the order number of the current production task, the system retrieves the product serial numbers of all products to be produced and already produced under that order. The system automatically generates unique product serial numbers without duplication or missing numbers according to preset coding rules. Each product corresponds to an independent code, ensuring no duplication or missing numbers. Next, the core process parameter set is collected. The chip glass bonding temperature is collected in real time by the temperature sensor built into the bonding equipment. The sensor collects data at fixed intervals and transmits it to the process control module. The chip glass bonding pressure is collected synchronously by the pressure sensor of the bonding equipment, providing real-time feedback of the actual pressure data during the bonding process. The chip glass bonding time is recorded by the process control module as the cumulative time from the start of the bonding action to the completion of the bonding. The cell thickness parameter is measured by a laser thickness gauge in the liquid crystal filling process. After scanning the corresponding area of ​​the product, the instrument outputs the actual cell thickness data. The liquid crystal volume parameter is collected by the metering module of the filling equipment, recording the actual liquid crystal volume injected into the product. Temperature and humidity sensors are deployed at each production station to collect the ambient temperature and humidity data of the area in real time. The sensor data is directly synchronized to the data acquisition terminal to ensure that every parameter in the core process parameter set is collected completely. Subsequently, through the batch association function of the material management system, the material list number corresponding to the production task is entered, and basic attribute information such as the warehousing batch number, supplier information, and warehousing inspection records of each material in the list is retrieved to clarify the batch-related information corresponding to each material, ensuring accurate traceability of material batches. Finally, through the AOI inspection equipment and manual re-inspection record system in the quality inspection process, product defect information is collected. The AOI equipment automatically scans the product display area, identifies the specific manifestations of uneven display and the coordinates of the defects, and detects the presence of light leakage defects and the approximate range of the light leakage area. The manual re-inspection record supplements the defect details not identified by the equipment, ensuring complete coverage of uneven display defect information and light leakage defect information.

[0040] Step 101 involves performing integrity checks and outlier handling on the product serial number, core process parameter set, material batch information, uneven display defect information, and light leakage defect information. Using the product serial number as the primary key, a correspondence is established between the product serial number and the core process parameter set, material batch information, and quality defect information, generating structured full-process data records, specifically including:

[0041] First, a data integrity check is performed, verifying each product serial number one by one. First, confirm that each product serial number is associated with the corresponding core process parameter set, checking that chip glass bonding temperature, chip glass bonding pressure, chip glass bonding time, cell thickness, liquid crystal volume, ambient temperature, and ambient humidity parameters all have corresponding data, with no blank or unrecorded items. Next, verify the material batch information, confirming that the batch number and basic attribute information for each material are complete, with no missing key fields. Finally, verify the quality defect information, confirming that any product defects have been recorded. If uneven display or light leakage defects exist, verify that their characteristic descriptions and location information are clear and complete. If any of the above three types of information corresponding to any product serial number is missing, the system automatically marks the data as incomplete and triggers a supplementary data collection command. The data acquisition terminal sends reminders to the responsible positions at the corresponding data sources. After the information is completed, it is resubmitted for verification until all product serial numbers are complete. Then, outlier handling is carried out. For each parameter in the core process parameter set, the historical data (i.e., valid operating data) collected from the historical production database during the production process of batches that have been judged to be qualified by quality inspection in the past are retrieved. These data are free of missing fields, logical contradictions, and conform to the parameter specifications and production process execution standards set by the process preset. After removing invalid records that are missing, contradictory, or exceed the parameter specifications set by the process preset, the total number of remaining valid data is counted, the historical mean of the parameter is calculated, the variance is calculated, and then the square root of the calculated variance is taken to obtain the standard deviation of the parameter.

[0042] Using the historical mean as a benchmark, the lower limit of the normal range is obtained by subtracting three standard deviations from the historical mean, and the upper limit is obtained by adding three standard deviations to the historical mean. This interval represents the normal data range for the parameter. The collected parameter data is compared with the normal range. If the data exceeds the lower or upper limit, it is identified as an outlier. After the outlier is marked by the system, the average of the historical normal production data for the parameter is retrieved and used to replace the outlier, ensuring that the parameter data conforms to the reasonable range of normal production and guaranteeing data validity. After data cleaning, the product serial number is used as the unique primary key, and the association mapping function of the data integration module is used to associate each product serial number with... The product precisely binds the complete set of core process parameters, all batch information of materials used, and quality defect information obtained during the production process, establishing a one-to-one correspondence. This ensures that each product's data is uniquely associated with its own serial number, without any cross-matching, many-to-one, or one-to-many situations. Finally, according to the preset structured data format specifications, all associated data are classified by product serial number, with each product corresponding to a complete record. The record is divided into four modules: product serial number, core process parameter set, material batch information, and quality defect information. Under each module, the corresponding data items are arranged in a fixed order, forming a structured full-process data record with a unified format and clear association.

[0043] This embodiment comprehensively covers four key data sources: production orders, process parameters, materials, and quality inspection. It effectively solves the problems of scattered and incomplete coverage of multi-source data in existing technologies, ensuring that no core data required for quality traceability is omitted. Through systematic integrity checks and scientific outlier handling processes, invalid data is eliminated, missing data is supplemented, and abnormal data is corrected, improving the integrity and accuracy of the data and avoiding deviations in subsequent traceability analysis due to data quality issues. By establishing data associations using product serial numbers as a unified primary key, the barriers between data from different sources are broken down, achieving standardized integration of multi-source heterogeneous data and forming a coherent and complete data chain for the entire product process.

[0044] In a preferred embodiment of the present invention, step 2 includes:

[0045] Based on structured end-to-end data recording, the core process parameter set is monitored in real time. When any parameter, including chip glass bonding temperature, chip glass bonding pressure, chip glass bonding time, cell thickness, liquid crystal volume, or ambient temperature and humidity, exceeds the preset specification range, the start and end times of the parameter anomaly are automatically recorded. All product serial numbers produced within the corresponding time period are marked as suspicious items, and an early warning signal is generated, specifically including:

[0046] After completing step 101 to generate structured full-process data records, immediately initiate the real-time monitoring process for core process parameters. First, determine the preset specification range for each core process parameter. The specific process is as follows: For chip glass bonding temperature, set the process baseline value to 100℃; extract historical chip glass bonding temperature data for all qualified products from the past year from the historical production database, calculate the difference between each historical data point and 100℃, and take the absolute value of all differences. Assume the largest absolute value is 3℃, which is the historical maximum fluctuation value of this parameter; at the same time, determine the allowable deviation value of this parameter to be 4℃. Compare 4℃ and 3℃, and take the larger 4℃ as the final allowable maximum deviation value; determine the preset specification range through calculation, with the upper limit = 100℃ + 4℃ and the lower limit = 100℃ - 4℃, that is, the preset specification range for chip glass bonding temperature is 96℃ to 104℃.

[0047] For the chip glass bonding pressure, the process baseline value is set to 0.5MPa. Historical data of this parameter corresponding to qualified products in the past year are extracted, and the difference between each historical data point and 0.5MPa is calculated and the absolute value is taken. The maximum absolute value is 0.1MPa, which is the maximum historical fluctuation value. The allowable deviation value of this parameter is 0.15MPa. Comparing 0.15MPa and 0.1MPa, 0.15MPa is taken as the maximum allowable deviation value. The upper limit value is 0.5MPa + 0.15MPa, and the lower limit value is 0.5MPa - 0.15MPa, forming a preset specification range of 0.35MPa to 0.65MPa.

[0048] For chip glass bonding time, the process baseline is set to 20 seconds; the absolute value of the difference between historical qualified data and 20 seconds is calculated, with a maximum absolute value of 2 seconds, which is the historical maximum fluctuation value; the allowable deviation is 3 seconds, which is taken as the maximum allowable deviation value; the upper limit is 20 seconds + 3 seconds, the lower limit is 20 seconds - 3 seconds, and the preset specification range is 17 seconds to 23 seconds; for cell thickness parameters, the process baseline is set to 4μm; the maximum absolute value of the difference between historical qualified data and 4μm is 0.3μm, which is the historical maximum fluctuation value; the allowable deviation is 0.4μm, which is taken as 0. 4μm is the maximum allowable deviation value; the upper limit is 4μm + 0.4μm, the lower limit is 4μm - 0.4μm, and the preset specification range is 3.6μm to 4.4μm; for the liquid crystal amount parameter, the process reference value is set to 5mg; the maximum absolute value of the difference between historical qualified data and 5mg is 0.5mg, which is the historical maximum fluctuation value; the allowable deviation value is 0.6mg, and 0.6mg is taken as the maximum allowable deviation value; the upper limit is 5mg + 0.6mg, the lower limit is 5mg - 0.6mg, and the preset specification range is 4.4mg to 5.6mg.

[0049] For the ambient temperature parameter, the process baseline value is set to 23℃; the maximum absolute value of the difference between historical qualified data and 23℃ is 2℃, which is the historical maximum fluctuation value; the allowable deviation value is 3℃, which is taken as the maximum allowable deviation value; the upper limit value is 23℃ + 3℃, the lower limit value is 23℃ - 3℃, and the preset specification range is 20℃ to 26℃. For the ambient humidity parameter, the process baseline value is set to 45%RH; the maximum absolute value of the difference between historical qualified data and 45%RH is 5%RH, which is the historical maximum fluctuation value; the allowable deviation value is 6%RH, which is taken as the maximum allowable deviation value; the upper limit value is 45%RH + 6%RH, the lower limit value is 45%RH - 6%RH, and the preset specification range is 39%RH to 51%RH.

[0050] After the preset specification ranges of each core process parameter are determined, the latest core process parameter set data is continuously extracted from the structured full-process data records at preset cycles. Each parameter data is bound to the corresponding product serial number, collection time, and production station information. During the monitoring process, for each extracted parameter data, it is compared in real time with the upper and lower limits of the preset specification range corresponding to that parameter to determine whether the data is between the upper and lower limits. If the chip glass bonding temperature is detected to be higher than 104℃ or lower than 96℃, the chip glass bonding pressure is higher than 0.65MPa or lower than 0.35MPa, the chip glass bonding time is longer than 23 seconds or shorter than 17 seconds, or the cell thickness parameter is greater than 4.4... If any of the following conditions are met, the parameter is immediately determined to be abnormal: μm or less than 3.6μm; liquid crystal quantity parameter is more than 5.6mg or less than 4.4mg; ambient temperature is higher than 26℃ or lower than 20℃; or ambient humidity is higher than 51%RH or lower than 39%RH. The current precise time is recorded by the system time synchronization module as the start time of the parameter abnormality, and the parameter is continuously monitored at high frequency. During the period of abnormality, the parameter data is compared with the preset range in each acquisition cycle until the parameter data is stable within the corresponding preset specification range for three consecutive acquisition cycles. At this time, the current time is recorded as the end time of the abnormality, thus defining the abnormal period.

[0051] Once the abnormal period is identified, the system automatically retrieves the production time information of all products from the structured full-process data records, including the start and end times of production for each product. For each product, suspicious items are filtered through logical judgment. If the product's production start time is earlier than the abnormal end time and the product's production end time is later than the abnormal start time, it is determined that the product's production process overlaps with the abnormal period and belongs to the products produced within the abnormal period. The system automatically marks its product serial number as a suspicious item and adds a suspicious item identification field to the product in the structured full-process data records. At the same time, it associates the corresponding abnormal parameter name, abnormal start time, and abnormal end time to ensure data traceability. Finally, the system automatically generates an early warning signal. The early warning signal includes key information such as the name of the abnormal core process parameter, the start and end times of the parameter abnormality, the production station involved, the list of suspicious item serial numbers, and the direction in which the abnormal parameter deviates from the preset range. The early warning signal is issued through the audible and visual alarm devices on the production site and is also sent to the work terminals of quality control personnel, process technicians, and production management personnel in the form of push notifications to ensure that relevant personnel receive abnormal information as soon as possible.

[0052] This embodiment, by combining process requirements and historical data to set preset specification ranges for each core process parameter, avoids the one-sidedness of relying solely on experience to set thresholds, making the monitoring standards more aligned with actual production needs and improving the rationality and accuracy of parameter monitoring. It achieves comprehensive real-time monitoring of all core process parameters, breaking through the limitations of discrete monitoring and missing key areas in existing technologies. This ensures that fluctuations in key process parameters during production can be captured promptly, with no monitoring blind spots. Recording the start and end times of parameter anomalies and filtering suspicious products through overlapping production periods allows for rapid identification of products with potential quality risks, preventing defective products from being transferred to subsequent processes in batches and reducing unnecessary waste of production resources. Automatically generating early warning signals containing key information enables rapid response and notification of quality anomalies, shortening the response cycle for anomaly handling and facilitating timely intervention by staff to investigate the causes of anomalies and take control measures. Monitoring and marking based on structured full-process data establishes a complete association between suspicious products, abnormal parameters, and production periods, providing clear data support for subsequent quality traceability and defect cause analysis, and improving the systematicness and consistency of quality control.

[0053] In a preferred embodiment of the present invention, step 3 includes:

[0054] Step 300: Based on the full-process data records of display unevenness defects and light leakage defects in suspected and historical defective products, a first process unit is configured at the chip glass bonding station, a second process unit at the light guide plate assembly station, a third process unit at the liquid crystal filling station, a fourth process unit at the polarizer attachment station, and a fifth process unit at the backlight module installation station. Specifically, the first process unit corresponds to the chip glass bonding temperature, chip glass bonding pressure, and chip glass bonding time; the second process unit corresponds to the ambient temperature and ambient humidity parameters; the third process unit corresponds to the cell thickness and liquid crystal volume parameters; the fourth process unit corresponds to the ambient temperature and ambient humidity parameters; and the fifth process unit corresponds to the ambient temperature and ambient humidity parameters.

[0055] First, we analyzed the correlation between parameter fluctuations and defect occurrences at each workstation in the entire production process for suspected and historically defective products. We identified the chip glass bonding workstation, light guide plate assembly workstation, liquid crystal filling workstation, polarizer attachment workstation, and backlight module installation workstation as the key workstations affecting uneven display and light leakage defects. Then, we carried out targeted configuration of the five process units.

[0056] At the chip glass bonding station, an integrated high-precision monitoring module is installed as the first process unit. This module is synchronized with the bonding equipment at the station and collects three core process parameters: chip glass bonding temperature, chip glass bonding pressure, and chip glass bonding time. The collection frequency is consistent with the monitoring cycle in step 2 to ensure that parameter collection and process execution are synchronized in real time without any time difference. At the light guide plate assembly station, an environmental monitoring component is deployed as the second process unit. The component has built-in temperature and humidity sensors to collect the ambient temperature and humidity parameters of the station. The monitoring range of the sensors covers the entire light guide plate assembly operation area to ensure the comprehensiveness of environmental parameter collection. At the liquid crystal filling station, a dedicated process monitoring device is configured as the third process unit. This device is linked with the metering and thickness measurement modules of the filling equipment to collect cell thickness and liquid crystal volume parameters, capturing the dynamic changes of these two key parameters in real time during the filling process. At the polarizer attachment station, an environmental monitoring component of the same model as the second process unit is installed as the fourth process unit to collect the ambient temperature and humidity parameters of this station, ensuring accurate monitoring of environmental conditions during polarizer attachment. At the backlight module installation station, the same environmental monitoring component is deployed as the fifth process unit to collect the ambient temperature and humidity parameters of this station, realizing continuous monitoring of environmental parameters during the backlight module installation process. All five process units establish data transmission channels with the structured full-process data recording system to ensure that the collected parameter data can be synchronized to the system in real time.

[0057] Step 301 involves establishing a correlation between the monitoring data of the five process units, constructing a pentagonal topology monitoring structure, and dynamically updating the pentagonal topology monitoring structure based on real-time monitoring data. Specifically, this includes: firstly, establishing a correlation between the monitoring data of the five process units, using the product serial number as the core correlation link, and combining it with the production timestamp to achieve accurate data matching. For each product serial number, extracting its chip glass bonding temperature, chip glass bonding pressure, and chip glass bonding time data in the first process unit, and corresponding them one by one with the ambient temperature and humidity parameters in the second process unit, the cell thickness and liquid crystal quantity parameters in the third process unit, the ambient temperature and humidity parameters in the fourth process unit, and the ambient temperature and humidity parameters in the fifth process unit. This ensures that the monitoring data of the same product in different process units are uniquely correlated through the serial number and timestamp, without mismatch or omission.

[0058] The first to fifth process units are respectively regarded as five fixed nodes in the topology structure. Each node stores the monitoring parameter dataset of the corresponding workstation and the process attribute information of the unit. The nodes are connected through data association links to form a closed pentagonal topology monitoring structure. The first process unit node establishes a connection link with the second and fifth process unit nodes respectively. The second process unit node establishes a connection link with the first and third process unit nodes. The third process unit node establishes a connection link with the second and fourth process unit nodes. The fourth process unit node establishes a connection link with the third and fifth process unit nodes. The fifth process unit node establishes a connection link with the fourth and first process unit nodes. The links carry parameter association data and time connection information of the same product serial number in adjacent process units.

[0059] After the topology is constructed, the system acquires the latest real-time monitoring data from the five process units at a preset cycle (consistent with the high-frequency monitoring cycle in step 2). First, the newly acquired data is validated to ensure that the data format meets the requirements and there are no obvious anomalies (the validation standard is consistent with the anomaly handling standard in step 101). After the validation is passed, the new data is updated to the node storage of the corresponding process unit. At the same time, the associated data of the product serial number corresponding to the data is checked in other process unit nodes. If it is found that the associated data needs to be updated (such as the latest parameters of the product collected by the subsequent process unit), the association information in the connection link between nodes is updated synchronously. When a new suspicious product mark, parameter anomaly warning or process adjustment occurs, the system automatically triggers the emergency update process of the topology, prioritizes the processing of the monitoring data association relationship of related products, and recalibrates the link data between nodes to ensure that the five-sided topology monitoring structure is always synchronized with the actual production situation and data changes, and accurately reflects the parameter association status of each process unit.

[0060] This embodiment, by configuring process units at key workstations that affect uneven display and light leakage defects, achieves targeted and precise monitoring of core process parameters and environmental parameters. This avoids the problem of indiscriminate monitoring throughout the entire process, ensuring that no key parameters are missed. By establishing a correlation between the monitoring data of the five process units using product serial numbers and timestamps, the barrier of isolated data at each workstation is broken down, forming an organic whole from scattered parameter data. This provides a data foundation for analyzing the synergistic effects of multiple workstation parameters on defects. The five-sided topology monitoring structure uses key workstations as nodes and data associations as links, intuitively presenting the transmission and association logic of process parameters throughout the entire process. This solves the problem of existing technologies struggling to systematically analyze the interaction of multiple workstation parameters. The dynamic update mechanism ensures that the topology structure can respond to data changes and production anomalies in real time, guaranteeing the timeliness and accuracy of the monitoring structure.

[0061] In a preferred embodiment of the present invention, step 4 includes:

[0062] Step 400: Based on the core process parameter set data collected by the pentagonal topology monitoring structure, a seven-dimensional parameter space is constructed using chip glass bonding temperature, chip glass bonding pressure, chip glass bonding time, cell thickness parameter, liquid crystal quantity parameter, ambient temperature parameter, and ambient humidity parameter as coordinate axes. Specifically, this includes setting chip glass bonding temperature coordinate axis, chip glass bonding pressure coordinate axis, chip glass bonding time coordinate axis, cell thickness parameter coordinate axis, liquid crystal quantity parameter coordinate axis, ambient temperature parameter coordinate axis, and ambient humidity parameter coordinate axis as independent coordinate axes. The value range of each coordinate axis is consistent with the preset specification range of the corresponding parameter determined in Step 2, ensuring that the range of the coordinate axis can fully cover the parameter fluctuation range in actual production. The seven coordinate axes are perpendicular to each other and do not intersect, together forming a complete seven-dimensional parameter space.

[0063] Step 401: In the seven-dimensional parameter space, for each process unit, obtain all core process parameter values ​​of the corresponding product at the corresponding workstation from the full-process data record, and convert the monitoring data of the five process units into five data points. Each data point is composed of the coordinates of all core process parameter values ​​of the corresponding process unit in the seven-dimensional parameter space. Connect the five data points in the order of the process units to form a pentagon. Specifically, this includes: for each marked suspicious product and historical defective product, extract its complete core process parameter values ​​at the production stages of the five process units from the structured full-process data record. For the first process unit (chip glass)... For the bonding station, extract the real-time data of chip glass bonding temperature, chip glass bonding pressure, and chip glass bonding time during the production of the product at that station. Simultaneously extract the real-time data of cell thickness, liquid crystal volume, ambient temperature, and ambient humidity collected synchronously during the production of the product at that station. These seven parameters together constitute the data point coordinates of the first process unit in a seven-dimensional parameter space. For the second process unit (light guide plate assembly station), extract the real-time data of ambient temperature and humidity during the production of the product at that station, as well as the synchronously recorded chip glass bonding temperature. Historical data points (final values ​​for this product in the first process unit), historical data points for chip glass bonding pressure (final values ​​for the first process unit), historical data points for chip glass bonding time (final values ​​for the first process unit), real-time data points for cell thickness parameters (synchronized data from the infusion equipment), and real-time data points for liquid crystal volume parameters (synchronized data from the infusion equipment) form the seven-dimensional coordinates of the data points in the second process unit. The data point coordinates for the third process unit (liquid crystal infusion station) are formed by combining the real-time data points for cell thickness and liquid crystal volume for this product at that station with the synchronously recorded historical values ​​for chip glass bonding temperature and pressure, and environmental data. The data points of the fourth process unit (polarizer attachment station) and the fifth process unit (backlight module installation station) are composed of real-time temperature and humidity values ​​collected at the corresponding workstations, combined with the synchronously recorded values ​​of the other five parameters, following the logic of the second process unit. After the data points of the five process units have completed seven-dimensional coordinate mapping, they are connected sequentially to the data points of adjacent process units in the order of the first, second, third, fourth, and fifth process units, forming a closed pentagonal structure. The link automatically records the difference in the changes of parameters in each dimension between adjacent data points.

[0064] Step 402: Calculate the minimum area enclosing rectangle of the pentagon. The minimum area enclosing rectangle is the rectangle that contains all vertices of the pentagon and has the smallest area. The center point of the corresponding minimum area enclosing rectangle is used as the reference parameter point. Specifically, this includes: First, decompose the five data points of the pentagon into dimensions. For each coordinate axis of the seven-dimensional parameter space, extract all coordinate values ​​of the five data points on that coordinate axis. After arranging these coordinate values ​​in ascending order, filter out the maximum and minimum values ​​to determine the range of the interval enclosing all data points on that coordinate axis. For example, on the chip glass bonding temperature coordinate axis, extract the temperature coordinate values ​​of the five data points, sort them, and obtain the maximum and minimum values ​​of that dimension. This interval is the minimum area enclosing rectangle. The boundary range of the rectangle enclosing the pentagon in this dimension is determined; the boundary ranges of the other six coordinate axes are determined one by one in the same way. After the boundary ranges of all seven coordinate axes are determined, a super rectangle containing all vertices of the pentagon and having the smallest area (i.e., the rectangle with the smallest area) is formed. Then, the center point of this rectangle is calculated. For each coordinate axis, the maximum value of that dimension is added to the minimum value, and the sum is divided by two to obtain the coordinate value of the center point on that coordinate axis. The center point calculations of the seven coordinate axes of chip glass bonding temperature, chip glass bonding pressure, chip glass bonding time, cell thickness parameter, liquid crystal quantity parameter, ambient temperature parameter, and ambient humidity parameter are completed in sequence. The center point coordinate values ​​of the seven dimensions are integrated to form the reference parameter points in the seven-dimensional parameter space.

[0065] Step 403: Establish the correspondence between process parameter changes and the occurrence of display unevenness defects or light leakage defects based on the reference parameter points. The projection values ​​of the reference parameter points on the chip glass bonding temperature coordinate axis are used as temperature reference values, on the chip glass bonding pressure coordinate axis as pressure reference values, on the chip glass bonding time coordinate axis as time reference values, on the cell thickness parameter coordinate axis as cell thickness reference values, on the liquid crystal quantity parameter coordinate axis as liquid crystal quantity reference values, on the ambient temperature parameter coordinate axis as ambient temperature reference values, and on the ambient humidity parameter coordinate axis as ambient humidity reference values. The reference parameter state is composed of the temperature reference value, pressure reference value, time reference value, cell thickness reference value, liquid crystal quantity reference value, ambient temperature reference value, and ambient humidity reference value, specifically including:

[0066] First, the product classification and screening process is initiated. All marked suspicious products, historical defective products, and corresponding production batches' product data are retrieved from the structured full-process data records. Classification is based on the quality defect information field, classifying products with clearly recorded uneven display or light leakage defects as defective products. Simultaneously, products from the same production batch as the defective products, whose quality defect information field is marked as defect-free, are classified as qualified products. This ensures consistency in the production background (e.g., material batches, production equipment, basic process settings) between the two categories, guaranteeing the effectiveness of the comparative analysis. After classification, the data from both categories is extracted. The associated data, namely the pentagonal structures corresponding to defective and qualified products, all come from the closed pentagons formed by connecting the data points of the five process units in step 401 (including the seven-dimensional coordinates and edge link information of each data point); the minimum area bounding rectangles of the two types of products come from the super rectangles calculated for their respective pentagons in step 402 (including the maximum, minimum, and range values ​​on the seven coordinate axes); the reference parameter points of the two types of products are the center points of their respective minimum area bounding rectangles calculated in step 402 (including coordinate values ​​on the seven coordinate axes), and all extracted data are bound one-to-one with the product serial number.

[0067] For each coordinate axis, the coordinate values ​​of the five data points of the defective product pentagon are extracted one by one and compared with the coordinate values ​​of the corresponding data points of the qualified product pentagon. The difference between each data point on that coordinate axis is calculated as (defective product data point coordinate value - qualified product data point coordinate value). If the difference between most data points on the same coordinate axis is positive, the dimension is determined to be positively offset; if the difference is mostly negative, it is negatively offset, thus clarifying the overall offset trend of the pentagon. For each corresponding side of the defective and qualified product pentagons (such as the side connecting the first and second process units), the side length is calculated using the seven-dimensional spatial distance formula. First, the coordinate difference between two adjacent data points on the seven coordinate axes is calculated. Each difference is squared and then summed. Finally, the square root of the sum is taken. Calculate the length of each side; then subtract the length of the corresponding side of the qualified product from the length of the side of the defective product to obtain the length difference value of each side, and determine whether there is a significant difference in the length of a certain side; for the minimum area bounding rectangle of the two types of products, calculate the difference between the maximum and minimum values ​​on each coordinate axis (i.e., the side length of that dimension), and multiply the side lengths of the seven coordinate axes in turn to obtain the volume of the rectangle; analyze the difference in parameter distribution range by the difference between the volume of the rectangle of the defective product and the volume of the rectangle of the qualified product; for each coordinate axis, subtract the coordinate value of the reference parameter point of the qualified product from the coordinate value of the reference parameter point of the defective product, and then take the absolute value of the difference to obtain the deviation range of that dimension, and record the deviation range data of the seven coordinate axes one by one.

[0068] Based on the above comparison results, the changing trends of core process parameter sets are analyzed. For example, if over 80% of the data points for a certain type of parameter show positive or negative shifts, or if the amplitude of coordinated fluctuations between parameters (such as simultaneous shifts in ambient temperature and humidity in the same direction, or synchronous fluctuations in chip glass bonding temperature, pressure, and time parameters) are correlated with defect types and defect occurrence probabilities, then the following logic applies: If over 80% of the data points in a dimension corresponding to a certain core process parameter maintain a unified shift direction, and the deviation amplitude of this dimension (the absolute value of the difference between the baseline parameter point of defective products and the baseline parameter point of qualified products) exceeds twice the standard deviation of the corresponding parameter of qualified products, and simultaneously, the difference in the length of the pentagonal side of this dimension (the absolute value of the difference between the side length of defective products and the side length of qualified products) exceeds 30% of the side length of qualified products, and the volume difference of the rectangle enclosed by the smallest area (the rectangle of defective products) is also considered. If the difference between the volume of the product and the rectangular volume of the qualified product exceeds 50% of the rectangular volume of the qualified product due to the increase in the difference in the side length of this dimension, then the abnormal change of this parameter is directly related to a certain type of defect (display unevenness or light leakage). If multiple parameters (3 or more) show coordinated fluctuations, such as the cell thickness parameter and liquid crystal quantity parameter shifting positively at the same time, and the ambient temperature and ambient humidity shifting negatively at the same time, causing the pentagon of the defective product to exceed the coverage of the pentagon of the qualified product, or the deviation of the reference parameter point on multiple coordinate axes (the deviation of each dimension exceeds 1.5 times the standard deviation of the corresponding parameter of the qualified product) and the sum exceeds 1.5 times the sum of the standard deviations of the qualified product in each dimension, then it is determined that the coordinated change of multiple parameters will significantly increase the risk of the corresponding defect. In this way, a complete and clear correspondence between the changes in process parameters and the occurrence of display unevenness defects and light leakage defects is established.

[0069] Based on this, the projection values ​​of the reference parameter points are extracted. The reference parameter points are the center points of the seven-dimensional parameter space calculated in step 402. Their coordinate values ​​on the chip glass bonding temperature coordinate axis are directly used as the temperature reference value; their coordinate values ​​on the chip glass bonding pressure coordinate axis are used as the pressure reference value; their coordinate values ​​on the chip glass bonding time coordinate axis are used as the time reference value; their coordinate values ​​on the cell thickness parameter coordinate axis are used as the cell thickness reference value; their coordinate values ​​on the liquid crystal quantity parameter coordinate axis are used as the liquid crystal quantity reference value; and their coordinate values ​​on the ambient temperature parameter coordinate axis are used as the ambient temperature reference value. The coordinate values ​​on the environmental humidity parameter coordinate axis are used as environmental humidity reference values. Each reference value is directly taken from the calculation result of step 402 to ensure accurate matching with the distribution characteristics of the minimum area enclosing rectangle. Finally, the temperature reference value, pressure reference value, time reference value, cell thickness reference value, liquid crystal volume reference value, environmental temperature reference value, and environmental humidity reference value are classified and organized according to the category of core process parameters (chip glass bonding related parameters, liquid crystal filling related parameters, and environmental related parameters), and duplicate related data are eliminated to form a unified reference parameter state with a clear structure and complete parameters.

[0070] This embodiment, by constructing a seven-dimensional parameter space, transforms dispersed core process parameters into visualized spatial coordinate data, solving the problem of difficult systematic integration and analysis of multi-source parameters, and making the synergistic change characteristics of multiple parameters more intuitive. Mapping the monitoring data of five process units into a pentagonal structure clearly presents the transmission path and correlation logic of process parameters throughout the entire process, breaking the limitations of isolated analysis of parameters at each workstation and providing a structured carrier for capturing the interaction of multiple parameters. The calculation of the minimum area enclosing rectangle and the benchmark parameter point extracts the core distribution characteristics of the entire process parameters, clarifies the parameter benchmark under normal production conditions, and provides an objective basis for judging whether parameter fluctuations are abnormal. Establishing a correspondence between process parameters and defects based on benchmark parameter points changes the current situation where existing technologies rely on experience to judge the cause of defects. The formation of benchmark parameter states avoids the blindness of parameter adjustments.

[0071] In a preferred embodiment of the present invention, step 5 includes:

[0072] Step 500: Starting from the reference parameter point, construct a first process parameter transmission link in the parameter space along a first direction and a second process parameter transmission link along a second direction. The first direction is determined by historical data statistics of chip glass bonding temperature, cell thickness, and liquid crystal volume. The second direction is determined by historical data statistics of chip glass bonding pressure, chip glass bonding time, ambient temperature, and ambient humidity. Specifically, this includes: first, clarifying the parameter composition of the two links; the first process parameter transmission link is associated with chip glass bonding temperature, cell thickness, and liquid crystal volume; the second process parameter transmission link is associated with chip glass bonding pressure, chip glass bonding time, ambient temperature, and ambient humidity. To determine the first direction, extract historical valid data of chip glass bonding temperature, cell thickness, and liquid crystal volume corresponding to all qualified products in the past year from the structured full-process data records. Data (excluding outliers processed in step 101); For each parameter, calculate its historical statistical value, sum all historical valid data for that parameter, and then divide by the total number of historical valid data to obtain the historical average value of that parameter. This average value is the historical data statistical value of that parameter. Taking the reference parameter point as the starting position, the logic for determining the first direction is as follows: calculate the difference between the historical average value of the chip glass bonding temperature and the temperature reference value at the reference parameter point, the difference between the historical average value of the cell thickness parameter and the cell thickness reference value, and the difference between the historical average value of the liquid crystal quantity parameter and the liquid crystal quantity reference value. These three differences together constitute the core trend basis of the first direction. Extend along this trend in the seven-dimensional parameter space to form the first process parameter transmission link. The coordinate value of each node in the link is based on the coordinates corresponding to the reference parameter point, and gradually increases or decreases according to the above difference ratio to ensure that the link direction is consistent with the historical parameter change trend.

[0073] For the second direction, historical valid data on chip glass bonding pressure, chip glass bonding time, ambient temperature parameter, and ambient humidity parameter corresponding to all qualified products in the past year are extracted from the structured full-process data records. The historical mean of each parameter is calculated one by one (calculated by summing all historical valid data and dividing by the total number of data points) as the historical data statistical value of each parameter. The differences between the historical mean of chip glass bonding pressure and the pressure benchmark value, the historical mean of chip glass bonding time and the time benchmark value, the historical mean of ambient temperature parameter and the ambient temperature benchmark value, and the historical mean of ambient humidity parameter and the ambient humidity benchmark value are calculated. These four differences constitute the trend basis for the second direction. Starting from the benchmark parameter point, the second process parameter transmission link is constructed by extending along this trend in the seven-dimensional parameter space. The coordinates of the link nodes increase or decrease according to the corresponding difference ratio to ensure that they match the historical change trend of the corresponding parameters. Both links extend to cover the maximum distribution range of monitoring data of five process units to ensure that the parameter transmission logic can be fully reflected.

[0074] Step 501: The orientation difference value is obtained by calculating the angle between the direction vectors of the first process parameter transmission link and the second process parameter transmission link. Specifically, this includes: firstly, converting the two process parameter transmission links into direction vectors in seven-dimensional space; the direction vector of the first process parameter transmission link is composed of the difference between the historical average and the reference value of the chip glass bonding temperature, cell thickness parameter, and liquid crystal quantity parameter, while the components of the other four unrelated parameters are set to 0; the direction vector of the second process parameter transmission link is composed of the difference between the historical average and the reference value of the chip glass bonding pressure, chip glass bonding time, ambient temperature parameter, and ambient humidity parameter, while the components of the other three unrelated parameters are set to 0.

[0075] The angle between the two vectors is then calculated using the dot product formula. The first step involves calculating the dot product of the two direction vectors by multiplying the corresponding components of each vector and summing the results. The second step calculates the magnitude of the first direction vector by squaring its components and summing the sums, then taking the square root of the sum. The third step calculates the magnitude of the second direction vector, following the same logic as the first direction vector. The fourth step calculates the azimuth difference by dividing the dot product by the product of the magnitudes of the first and second direction vectors, obtaining the cosine of the angle between the two vectors. This cosine value is then converted to the corresponding angle using the inverse cosine function. This angle is the azimuth difference (ranging from 0 to 90 degrees; if the angle calculated using the inverse cosine function is less than 0 degrees, it is directly defined as 0 degrees; if the calculated angle is greater than 90 degrees, it is directly defined as 90 degrees, ensuring that the azimuth difference value remains within the preset range).

[0076] Step 502: Based on the azimuth difference value, and taking the reference parameter point as the center, combined with the maximum distribution distance of the monitoring data of the five process units in the parameter space, determine the boundary range of the multi-parameter interaction domain. The multi-parameter interaction domain is a fan-shaped area centered on the reference parameter point, with the azimuth difference value as the expansion angle, and the maximum distribution distance of the real-time monitoring data as the radius. Specifically, it includes: firstly, calculating the maximum distribution distance of the monitoring data of the five process units in the seven-dimensional parameter space; for all monitoring data points of the first to fifth process units, calculating the distance one by one; taking a single monitoring data point as an example, first extracting the seven coordinate values ​​of the data point in the seven-dimensional parameter space (chip glass bonding temperature, chip glass bonding pressure, chip glass bonding time, cell thickness parameter, liquid crystal quantity parameter). The monitoring data point is analyzed by first extracting the ambient temperature and humidity parameters, then extracting seven reference values ​​corresponding to the reference parameter points (temperature reference value, pressure reference value, time reference value, cell thickness reference value, liquid crystal volume reference value, ambient temperature reference value, and ambient humidity reference value). The difference between each coordinate value of the data point and the corresponding reference value is calculated, and the seven differences are squared. All the squared results are then summed, and the square root of the sum is taken to obtain the Euclidean distance between the monitoring data point and the reference parameter point. Following the same logic, all monitoring data points of the five process units are traversed. After calculating the Euclidean distance between each data point and the reference parameter point, all the calculated distance values ​​are compared and filtered to determine the largest value, which is the maximum distribution distance of the real-time monitoring data.

[0077] First, the reference for defining the sector area is clearly defined. The reference parameter point calculated in step 402 is directly used as the unique center of the sector. The seven-dimensional coordinates of this center are composed of the reference values ​​for chip glass bonding temperature, chip glass bonding pressure, chip glass bonding time, cell thickness, liquid crystal volume, ambient temperature, and ambient humidity. These coordinates are completely consistent with the seven-dimensional coordinates of the reference parameter point, ensuring that all subsequent boundary delineation revolves around a unified reference without any deviation. Next, the expansion angle of the sector is set. The orientation difference value calculated in step 501 using the vector dot product formula is directly used as the expansion angle of the sector. The starting edge of the expansion angle is completely aligned with the direction of the first process parameter transmission link, that is, it extends along the historical trend of chip glass bonding temperature, cell thickness, and liquid crystal volume. The ending edge is completely aligned with the direction of the second process parameter transmission link, that is, it extends along the historical trend of chip glass bonding pressure, chip glass bonding time, ambient temperature, and ambient humidity. This ensures that the two process parameter transmission links fit exactly against the two sides of the sector area, forming an opening with a fixed angle.

[0078] Then, the radius of the sector and the arc boundary are determined. The radius is taken as the maximum distribution distance of the real-time monitoring data. The calculation process is as follows: for each monitoring data point of the five process units, calculate the seven-dimensional Euclidean distance between it and the reference parameter point. Specifically, first calculate the difference of each dimension coordinate, square all the differences and sum them, then take the square root of the sum. After traversing all data points, select the largest Euclidean distance as the maximum distribution distance of the real-time monitoring data. Then, starting from the center of the circle, divide the unfolding angle into several subdivision directions at intervals of 1 degree. For each subdivision direction, first determine the seven-dimensional Euclidean distance of that direction. A unit vector is defined as follows: each dimension component represents the projection percentage of that direction onto the corresponding dimension, and the magnitude of the unit vector is 1. The calculation method is to take the square root of the sum of the squares of the projection values ​​of each dimension, which equals 1. Then, the coordinates of the boundary points of each dimension are calculated. Specifically, the coordinates are the baseline value of that dimension plus the product of the unit vector component of the corresponding dimension of that subdivision direction and the maximum distribution distance of the real-time monitoring data. This ensures that the seven-dimensional Euclidean distance from each boundary point to the center of the circle is equal to the radius. Finally, the boundary points of all subdivision directions are connected sequentially to form a fan-shaped arc boundary, which, together with the starting edge and the ending edge, closes to form a complete fan-shaped region. This region is the multi-parameter interaction domain.

[0079] Finally, the system automatically verifies the parameters. The first step is to verify the coverage of key parameter combination points. For each core process parameter, the upper limit of the preset specification range determined in step 2 is subtracted from the lower limit. The result is then multiplied by 1% to obtain the minimum variation range of the parameter. Next, following the logic of gradually increasing the minimum variation range from the lower limit to the upper limit for each parameter, seven-dimensional coordinate points corresponding to all possible parameter combinations between the two process parameter transmission links are generated. Then, key parameter combination points that are clearly associated with uneven display defects and light leakage defects are selected. Each key parameter combination point is checked to see if it is within the fan-shaped area. If there are uncovered key combination points, the starting or ending edge of the unfolding angle is adjusted in small increments of 1 degree, with a maximum adjustment range of no more than 5 degrees. After adjustment, the check is repeated until all key parameter combination points are included in the fan-shaped area. The second step is to verify the coverage of the main distribution range of data points. In a seven-dimensional parameter space, a fixed volume unit space is defined, with each dimension ranging from ±1% of the baseline value. The number of monitoring data points in each unit space is counted to obtain the distribution density of each region. Then, the initial volume of the sector is calculated, which is the seven-dimensional sector volume calculated based on the initial radius and the unfolding angle. The average distribution density is obtained by dividing the total number of monitoring data points of the five process units by the initial volume. The region with a distribution density greater than the average distribution density × 3 is determined as the main distribution range of the data points. It is checked whether all data points within the main distribution range are within the sector region. If there are data points that exceed the range, the seven-dimensional Euclidean distance from the data point to the center of the circle is calculated. The maximum distance of the data point that exceeds the range is used as the new radius. The difference between the new radius and the original radius is calculated to ensure that (new radius - original radius) ÷ original radius ≤ 20%. The sector radius is updated according to the new radius, while the unfolding angle remains unchanged, until all data points within the main distribution range are included in the sector region.

[0080] This embodiment constructs two process parameter transmission links by splitting them according to parameter functional attributes, clarifying the transmission trends of different types of parameters, breaking the analysis dilemma of chaotic multi-parameters, and making the parameter transmission logic more systematic. The calculation of directional difference values ​​quantifies the degree of difference in the transmission directions of the two types of parameters, providing an objective basis for defining the core range of parameter interaction and avoiding range deviations caused by subjective judgment. The construction of the multi-parameter interaction domain transforms the abstract multi-parameter interaction relationship into a concrete spatial region, clearly defining the effective range of parameter synergy and solving the problem of difficulty in capturing the complex interaction effects of multiple parameters. The fan-shaped region design centered on the benchmark parameter point and combined with the directional difference value and the maximum distribution distance ensures the specificity of the interaction domain and can fully cover the key range of parameter changes, improving the specificity of process optimization.

[0081] In a preferred embodiment of the present invention, step 6 includes:

[0082] Step 600: Based on the boundary range of the multi-parameter interaction domain, the monitoring data points of the five process units are sequentially connected to form a closed-loop feedback path according to the time sequence. Specifically, this includes: first, extracting all monitoring data points of the five process units from the structured full-process data record. Each data point carries a corresponding acquisition timestamp, product serial number, and seven-dimensional coordinate information. The data points are sorted according to the time sequence of the production process, starting with the acquisition time when the product enters the first process unit (chip glass bonding station), and then sequentially connecting the data points in the second process unit (light guide plate assembly station), the third process unit (liquid crystal filling station), and the fourth process unit (chip glass bonding station). The acquisition time of the process unit (polarizer attachment station) and the fifth process unit (backlight module installation station) ensures that the data points of the same product in each process unit are arranged according to the actual production sequence. Then, according to the sorting result, the first data point of the first process unit is connected to the corresponding data point of the second process unit with a data link. The data point of the second process unit is then connected to the corresponding data point of the third process unit, and so on, until the last data point of the fifth process unit is connected to the first data point of the first process unit, forming a closed loop feedback path. The time interval between adjacent data points and the difference in changes of parameters in each dimension are automatically recorded in the path.

[0083] Step 601: Analyze the intersection characteristics of the circular closed-loop feedback path and the boundary of the multi-parameter interaction domain, including the length of the intersecting line segments, the number of intersection points, and the distribution density of intersection points. Based on the intersection characteristics, extract the variation law of the core process parameter set, including parameter fluctuation amplitude, fluctuation frequency, and correlation strength between parameters. Specifically, this includes: First, analyzing the intersection characteristics of the circular closed-loop feedback path and the boundary of the multi-parameter interaction domain, first determining the intersection points, and then checking each data point and path segment on the circular closed-loop feedback path one by one. If the seven-dimensional coordinates of the data point fall exactly on the boundary of the multi-parameter interaction domain, or if the path segment crosses the boundary... (That is, if the endpoints of the line segment are located inside and outside the interaction domain respectively), then it is determined that the data point or line segment intersects the boundary, and the intersection point of the data point or line segment with the boundary is recorded simultaneously as the intersection point; then the length of the intersecting line segment is calculated. For each path line segment that has been determined to intersect the boundary, first identify the two intersection points of the line segment with the boundary, and then calculate the distance between these two intersection points according to the seven-dimensional Euclidean distance formula. The specific process is to calculate the coordinate difference of the two intersection points in seven dimensions, square each difference separately, sum them up, and then take the square root of the sum to finally obtain the length of the intersecting line segment; then the intersection is statistically analyzed. To determine the number of intersection points, iterate through all segments of the closed-loop feedback path, accumulate the total number of identified intersection points, and finally calculate the intersection point distribution density. First, extract the range of each dimension of the multi-parameter interaction domain boundary (chip-glass bonding temperature dimension: 104℃ - 96℃ = 8℃; chip-glass bonding pressure dimension: 0.65MPa - 0.35MPa = 0.3MPa; chip-glass bonding time dimension: 23 seconds - 17 seconds = 6 seconds; cell thickness parameter dimension: 4.4μm - 3.6μm = 0.8μm; liquid crystal quantity parameter dimension: 5.6mg - 4.4mg = 1.2mg; ambient temperature...). The degree parameter dimension is 26℃-20℃=6℃, and the environmental humidity parameter dimension is 51%RH-39%RH=12%RH, both of which are the differences between the upper and lower limits of the boundary corresponding to each core process parameter. Then, according to the seven-dimensional sector surface area formula, the range values ​​of the seven dimensions are multiplied in turn to obtain the basic value of the equivalent surface area of ​​the seven-dimensional sphere. Then, the basic value is multiplied by the ratio of the orientation difference value to 360 degrees (i.e., the proportion of the sector angle to the full circumference angle) to derive the seven-dimensional surface area of ​​the boundary of the multi-parameter interaction domain. Finally, the number of intersection points obtained by statistics is divided by the seven-dimensional surface area to obtain the intersection point distribution density.

[0084] Based on the intersection features obtained above, the variation patterns of the core process parameter set are further extracted. For each core process parameter, when calculating its fluctuation amplitude, the projection value of the length of all intersecting line segments on that dimension is first calculated (i.e., the length of the intersecting line segment multiplied by the direction cosine of that dimension). Then, the absolute value of all projection values ​​is taken and the average value is calculated. This average value is the fluctuation amplitude of the parameter. The fluctuation frequency is calculated by dividing the number of intersection points obtained by statistics by the total production time corresponding to the circular closed-loop feedback path, where the total production time is the difference between the acquisition time of the first data point and the acquisition time of the last data point of the path. The calculation result is the parameter fluctuation frequency per unit time. The correlation strength between parameters is obtained by calculating the product of the distribution density of the intersection points corresponding to any two parameters, and then dividing the product by the product of the fluctuation amplitudes of the two parameters to obtain the correlation strength between the two parameters. This logic is used to traverse all parameter combinations to form a complete parameter correlation strength matrix.

[0085] Step 602: Based on the variation patterns, calculate the compensation amounts for chip glass bonding temperature, chip glass bonding pressure, chip glass bonding time, cell thickness parameters, liquid crystal volume, ambient temperature, and ambient humidity to form a complete process parameter optimization and compensation scheme. This scheme is used to adjust the core process parameter set to correct process deviations and eliminate quality defects. Specifically, this includes: calculating the optimization compensation amount for each process parameter based on the extracted variation patterns to form a complete scheme. For the chip glass bonding temperature compensation amount, use the formula: Temperature Compensation Amount = Temperature Reference Value - (Current Chip Glass Bonding Temperature Compensation Amount) / ... The calculation is as follows: (Average temperature + Temperature fluctuation amplitude × Average correlation strength between temperature and the other six core process parameters (chip glass bonding pressure, chip glass bonding time, cell thickness, liquid crystal volume, ambient temperature, and ambient humidity)). A positive result indicates the current temperature is too high and needs to be reduced; a negative result indicates the current temperature is too low and needs to be increased; a zero result indicates no adjustment is needed. The chip glass bonding pressure compensation is calculated using the formula: Pressure Compensation = Pressure Baseline Value - (Average current chip glass bonding pressure + Pressure fluctuation amplitude × Average correlation strength between pressure and the other six core process parameters (chip glass bonding pressure, chip glass bonding time, cell thickness, liquid crystal volume, ambient temperature, and ambient humidity)). The average value of the correlation strength between bonding temperature, chip glass bonding time, cell thickness parameter, liquid crystal volume parameter, ambient temperature parameter, and ambient humidity parameter is calculated. The direction of pressure adjustment is determined based on the sign of the result (positive value requires pressure reduction, negative value requires pressure increase, zero value requires no adjustment). The chip glass bonding time compensation is calculated using the formula: Time Compensation = Time Base Value - (Average value of current chip glass bonding time + Time Fluctuation Amplitude × Time) / (Each of the other six core process parameters (chip glass bonding temperature, chip glass bonding pressure, cell thickness parameter, liquid crystal volume parameter, ambient temperature parameter, and ambient humidity parameter)). The cell thickness compensation is calculated using the average value of the correlation strength of the cell thickness parameters. The bonding time is adjusted based on the result (positive values ​​require shortening the time, negative values ​​require extending the time, and zero values ​​do not require adjustment). The cell thickness compensation is calculated using the formula: Cell Thickness Compensation = Cell Thickness Base Value - (Current Cell Thickness Parameter Average Value + Cell Thickness Fluctuation Amplitude × Average Value of the Correlation Strength between Cell Thickness and the Other Six Core Process Parameters (Chip Glass Bonding Temperature, Chip Glass Bonding Pressure, Chip Glass Bonding Time, Liquid Crystal Quantity Parameter, Ambient Temperature Parameter, Ambient Humidity Parameter). The cell thickness is corrected based on the result (positive values ​​require decreasing the cell thickness, negative values ​​require increasing the cell thickness, and zero values ​​do not require adjustment).

[0086] Liquid crystal volume compensation is calculated using the formula: Liquid Crystal Volume Compensation = Liquid Crystal Volume Base Value - (Average value of current liquid crystal volume parameters + Liquid Crystal Volume Fluctuation Amplitude × Average value of the correlation strength between liquid crystal volume and the other six core process parameters (chip glass bonding temperature, chip glass bonding pressure, chip glass bonding time, cell thickness parameter, ambient temperature parameter, and ambient humidity parameter)). The liquid crystal injection amount is adjusted based on the result (positive values ​​require a decrease in injection amount, negative values ​​require an increase in injection amount, and zero values ​​require no adjustment). Ambient temperature compensation is calculated using the formula: Ambient Temperature Compensation = Ambient Temperature Base Value - (Average value of current ambient temperature + Ambient Temperature Fluctuation Amplitude × Average value of ambient temperature and the other six core process parameters (chip glass bonding temperature, chip glass bonding pressure)). The production environment temperature is adjusted based on the average correlation strength between the chip glass bonding time, cell thickness parameter, liquid crystal quantity parameter, and ambient humidity parameter. Positive values ​​require a decrease in temperature, negative values ​​require an increase in temperature, and zero values ​​require no adjustment. The ambient humidity compensation is calculated using the formula: Ambient Humidity Compensation = Ambient Humidity Base Value - (Average Ambient Humidity + Ambient Humidity Fluctuation Amplitude × Average Correlation Strength Between Ambient Humidity and the Other Six Core Process Parameters (Chip Glass Bonding Temperature, Chip Glass Bonding Pressure, Chip Glass Bonding Time, Cell Thickness Parameter, Liquid Crystal Quantity Parameter, and Ambient Temperature Parameter). The production environment humidity is adjusted based on the results. Positive values ​​require a decrease in humidity, negative values ​​require an increase in humidity, and zero values ​​require no adjustment.

[0087] The compensation amounts of the above seven process parameters are organized according to the core process parameter categories (chip glass bonding related parameters, i.e., chip glass bonding temperature, pressure, and time compensation amounts; liquid crystal filling related parameters, i.e. cell thickness parameters and liquid crystal volume compensation amounts; environmental related parameters, i.e. environmental temperature and humidity compensation amounts). The adjustment direction (increase / decrease, lengthen / shorten, increase / decrease, etc.) and specific values ​​of each compensation amount are clearly defined to form a complete process parameter optimization compensation scheme. The scheme simultaneously marks the basis for the change pattern of each compensation amount, including the fluctuation range data of the parameter, the correlation strength details with the other six parameters, and the average value, to ensure that the logic of parameter adjustment is traceable and there are no ambiguities.

[0088] This embodiment constructs a closed-loop feedback path in chronological order, fully reconstructing the transmission trajectory of core process parameters throughout the entire production process. This allows for a direct presentation of the continuity and correlation of parameter changes, solving the problem of fragmented parameter analysis. By quantitatively analyzing intersection features, abstract parameter changes are transformed into calculable specific indicators, providing objective data support for extracting parameter change patterns and avoiding the subjectivity of relying on experience-based judgments. Based on intersection features, the embodiment accurately extracts the variation patterns of parameter fluctuation amplitude, fluctuation frequency, and the strength of correlation between parameters, clearly defining the core sources of process deviations and laying a scientific foundation for compensation calculation. The embodiment calculates the optimized compensation amount of each parameter according to a unified logic, forming a complete compensation scheme to ensure the pertinence and consistency of parameter adjustments, effectively correcting process deviations and eliminating quality defects such as uneven display and light leakage at their root.

[0089] like Figure 2 As shown, embodiments of the present invention also provide a full-process quality traceability system for liquid crystal display module production lines, including:

[0090] The data acquisition module is used to collect multi-source heterogeneous data, clean and unify the format of the multi-source heterogeneous data, establish the association between product serial number and various parameter information, and form a structured full-process data record.

[0091] The marking module is used to monitor the core process parameter set in real time based on structured full-process data records. When any parameter is detected to be outside the preset specification range, the product serial number produced in the corresponding abnormal period is marked as a suspicious product and an early warning signal is generated.

[0092] The configuration module is used to configure the first to fifth process units and construct a pentagonal topology monitoring structure based on the full-process data recording of suspicious items and historical defective items.

[0093] A module is established to collect core process parameter set data based on the pentagonal topology monitoring structure, establish the correspondence between process parameter changes and quality defect occurrence, and determine the baseline parameter points based on the distribution characteristics of the monitoring data of the five process units to establish the baseline parameter status.

[0094] The definition module is used to construct the transmission link between two process parameters with the baseline parameter state as a reference point, calculate the corresponding orientation difference value, and define the interaction domain of multiple parameters.

[0095] The adjustment module is used to analyze the monitoring data of each process unit in the pentagonal topology monitoring structure based on the multi-parameter interaction domain, construct a closed-loop feedback path, and extract the variation law of the core process parameter set by calculating the intersection characteristics of the closed-loop feedback path and the multi-parameter interaction domain; and derive the process parameter optimization compensation amount based on the variation law to adjust the core process parameter set.

[0096] It should be noted that this system is a system corresponding to the above method. All implementation methods in the above method embodiments are applicable to this embodiment and can achieve the same technical effect.

[0097] Embodiments of the present invention also provide a computing device, including: a processor and a memory storing a computer program, wherein the computer program, when executed by the processor, performs the method described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0098] Embodiments of the present invention also provide a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the method described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0099] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for quality tracing of a whole process of a liquid crystal display module production line, characterized in that, The method comprises: Step 1, collecting multi-source heterogeneous data, and performing cleaning and format unification processing on the multi-source heterogeneous data, establishing an association between product serial numbers and parameter information, and forming structured full-process data records; Step 2, based on the structured full-process data records, real-time monitoring is performed on the core process parameter set, when any parameter is detected to be out of the preset specification range, the product serial number of the product produced in the corresponding abnormal period is marked as a suspect product, and a warning signal is generated; Step 3, based on the full-process data records of the suspect product and the historical defective product with the display uneven defect and the light leakage defect, a first process unit is configured at a chip glass bonding station, a second process unit is configured at a light guide plate assembly station, a third process unit is configured at a liquid crystal filling station, a fourth process unit is configured at a polarizing plate attaching station, and a fifth process unit is configured at a backlight module installation station; wherein the first process unit corresponds to the chip glass bonding temperature, the chip glass bonding pressure and the chip glass bonding time, the second process unit corresponds to the environmental temperature parameter and the environmental humidity parameter, the third process unit corresponds to the box thickness parameter and the liquid crystal amount parameter, the fourth process unit corresponds to the environmental temperature parameter and the environmental humidity parameter, and the fifth process unit corresponds to the environmental temperature parameter and the environmental humidity parameter; the monitoring data of the five process units are associated to construct a five-edge topology monitoring structure, and the five-edge topology monitoring structure is dynamically updated according to the real-time monitoring data; Step 4, based on the core process parameter set data collected by the five-edge topology monitoring structure, a corresponding relationship between process parameter changes and quality defect occurrences is established, and a reference parameter point is determined based on the distribution characteristics of the monitoring data of the five process units to establish a reference parameter state; Step 5, taking the reference parameter point as the starting position, a first process parameter conduction link is constructed in the parameter space along a first direction, and a second process parameter conduction link is constructed along a second direction, wherein the first direction is determined by the historical data statistical values of the chip glass bonding temperature, the box thickness parameter and the liquid crystal amount parameter, and the second direction is determined by the historical data statistical values of the chip glass bonding pressure, the chip glass bonding time, the environmental temperature parameter and the environmental humidity parameter; the azimuth difference value is obtained by calculating the included angle between the direction vector of the first process parameter conduction link and the direction vector of the second process parameter conduction link; based on the azimuth difference value, the boundary range of the multi-parameter interaction scope is determined based on the maximum distribution distance of the monitoring data of the five process units in the parameter space with the reference parameter point as the center and the maximum distribution distance of the real-time monitoring data as the radius; the multi-parameter interaction scope is a sector region with the reference parameter point as the center, the azimuth difference value as the expansion angle, and the maximum distribution distance of the real-time monitoring data as the radius; Step 6, based on the multi-parameter interaction scope, the monitoring data of each process unit in the five-edge topology monitoring structure is analyzed to construct a closed-loop feedback path, and the core process parameter set change rule is extracted by calculating the intersection characteristics of the closed-loop feedback path and the multi-parameter interaction scope; the process parameter optimization compensation amount is derived according to the change rule to adjust the core process parameter set.

2. The liquid crystal display module production line full-process quality traceability method according to claim 1, characterized in that, The step 1 comprises: The product serial number is obtained from a production order source, the core process parameter set is obtained from a process parameter source, the material batch information is obtained from a material source, and the quality defect information including display uneven defect information and light leakage defect information is obtained from a quality detection source; wherein the core process parameter set includes chip glass bonding temperature, chip glass bonding pressure, chip glass bonding time, box thickness parameter, liquid crystal amount parameter, and environmental temperature and humidity parameter; The product serial number, the core process parameter set, the material batch information, the display uneven defect information, and the light leakage defect information are subjected to integrity check and abnormal value processing, a corresponding relationship between the product serial number and the core process parameter set, the material batch information, and the quality defect information is established, and a structured full-process data record is generated.

3. The liquid crystal display module production line full-process quality traceability method according to claim 2, characterized in that, The step 2 comprises: Based on the structured full-process data record, the core process parameter set is monitored in real time, and when any one of the chip glass bonding temperature, the chip glass bonding pressure, the chip glass bonding time, the box thickness parameter, the liquid crystal amount parameter, or the environmental temperature and humidity parameter is detected to be out of a preset specification range, the start time and the end time of the parameter abnormality are recorded automatically, all product serial numbers produced in the corresponding time period are marked as suspicious products, and a warning signal is generated.

4. The liquid crystal display module production line full-process quality traceability method according to claim 3, characterized in that, The step 4 comprises: Based on the core process parameter set data collected by the five-edge topology monitoring structure, the chip glass bonding temperature, the chip glass bonding pressure, the chip glass bonding time, the box thickness parameter, the liquid crystal amount parameter, the environmental temperature parameter, and the environmental humidity parameter are taken as coordinate axes to construct a seven-dimensional parameter space; In the seven-dimensional parameter space, for each process unit, all core process parameter values of the corresponding product at the corresponding station are obtained from the full-process data record, and the monitoring data of the five process units are converted into five data points, each data point being composed of the coordinates of all core process parameter values of the corresponding process unit in the seven-dimensional parameter space, and the five data points are connected in the order of the process units to form a five-edge polygon; The minimum area enclosing rectangle of the five-edge polygon is calculated, the minimum area enclosing rectangle being a rectangle containing all vertices of the five-edge polygon and having the minimum area, and the center point of the corresponding minimum area enclosing rectangle is taken as a reference parameter point; Based on the reference parameter point, a corresponding relationship between process parameter changes and the occurrence of display uneven defects or light leakage defects is established, and the projection value of the reference parameter point on the chip glass bonding temperature coordinate axis is taken as a temperature reference value, the projection value on the chip glass bonding pressure coordinate axis is taken as a pressure reference value, the projection value on the chip glass bonding time coordinate axis is taken as a time reference value, the projection value on the box thickness parameter coordinate axis is taken as a box thickness reference value, the projection value on the liquid crystal amount parameter coordinate axis is taken as a liquid crystal amount reference value, the projection value on the environmental temperature parameter coordinate axis is taken as an environmental temperature reference value, and the projection value on the environmental humidity parameter coordinate axis is taken as an environmental humidity reference value; the temperature reference value, the pressure reference value, the time reference value, the box thickness reference value, the liquid crystal amount reference value, the environmental temperature reference value, and the environmental humidity reference value constitute a reference parameter state.

5. The liquid crystal display module production line full-process quality traceability method according to claim 4, characterized in that, The step 6 comprises: According to the time sequence, the monitoring data points of the five process units are sequentially connected to form a ring-shaped closed-loop feedback path based on the boundary range of the multi-parameter interaction scope; The intersection characteristics of the ring-shaped closed-loop feedback path and the boundary of the multi-parameter interaction scope are analyzed, including the intersection line length, the number of intersection points, and the intersection point distribution density. The variation law of the core process parameter set is extracted according to the intersection characteristics, including the parameter fluctuation amplitude, the fluctuation frequency, and the correlation strength between parameters. Based on the variation law, the chip glass bonding temperature compensation amount, the chip glass bonding pressure compensation amount, the chip glass bonding time compensation amount, the box thickness parameter compensation amount, the liquid crystal amount compensation amount, the environmental temperature compensation amount, and the environmental humidity compensation amount are calculated to form a complete process parameter optimization compensation scheme for adjusting the core process parameter set to correct process deviations and eliminate quality defects.

6. A liquid crystal display module production line full-process quality traceability system, the system implements the method according to any one of claims 1 to 5, characterized in that, It includes: The acquisition module is used for collecting multi-source heterogeneous data, cleaning and formatting the multi-source heterogeneous data, establishing the association between the product serial number and each parameter information, and forming a structured full-process data record; The marking module is used for real-time monitoring of the core process parameter set based on the structured full-process data record. When any parameter is detected to be out of the preset specification range, the product serial number produced in the corresponding abnormal period is marked as a suspect product, and a warning signal is generated; The configuration module is used for configuring the first to fifth process units based on the full-process data record of the suspect product and the historical defective product, and constructing a five-edge topology monitoring structure; The establishment module is used for establishing the correspondence between the process parameter variation and the quality defect occurrence based on the core process parameter set data collected by the five-edge topology monitoring structure, and determining the reference parameter point based on the distribution characteristics of the monitoring data of the five process units to establish the reference parameter state; The delimitation module is used to take the reference parameter state as a reference point to construct two process parameter transmission links, calculate the corresponding azimuth difference value, and delimit the multi-parameter interaction scope; The adjustment module is used for analyzing the monitoring data of each process unit in the five-edge topology monitoring structure based on the multi-parameter interaction scope, constructing a closed-loop feedback path, and extracting the variation law of the core process parameter set by calculating the intersection characteristics of the closed-loop feedback path and the multi-parameter interaction scope. According to the variation law, the process parameter optimization compensation amount is derived to adjust the core process parameter set.

7. A computing device, comprising: It includes: One or more processors; Storage device for storing one or more programs, when the one or more programs are executed by the one or more processors, so that the one or more processors implement the method of any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a program which is executed by a processor to implement the method of any one of claims 1 to 5.

Images