Production control method and system of liquid crystal display screen, medium and product

Abstract

The invention discloses a production control method and system of a liquid crystal display screen, a medium and a product, and relates to the field of liquid crystal optical display. Extracting a reference product parameter and a reference process parameter corresponding to the product identification number from a process formula library; according to a difference coefficient between the display unit parameter and the reference product parameter, correcting the reference process parameter as an initial process parameter; collecting liquid crystal spreading state data in real time in the process of executing the liquid crystal instillation flow according to the initial process parameters, generating a process stability index according to the liquid crystal spreading state data, and performing parameter correction based on a process parameter compensation model when the process stability index deviates from a preset stability range to obtain process correction parameters and corresponding product characteristics; and executing a liquid crystal instillation process according to the process correction parameters, and storing the process correction parameters and the product characteristics in a process formula library. According to the invention, rapid adaptation and intelligent optimization of the production process can be realized, and the mixed line production efficiency is improved.

Description

Technical Field

[0001] This application relates to the field of liquid crystal optical displays, and particularly to a production control method, system, medium and product for a liquid crystal display screen. Background Art

[0002] With the rapid development of display technology, liquid crystal display panels are evolving towards large size and high resolution. In the manufacturing process of liquid crystal displays, liquid crystal injection is a key process that determines the product quality. To improve production efficiency, the current production mode of liquid crystal panels generally uses a single glass substrate to manufacture multiple display units simultaneously, which poses higher requirements for the precise control of the liquid crystal injection process.

[0003] In related technologies, the mainstream liquid crystal panel production line uses the ODF (One Drop Fill) process for liquid crystal injection and panel lamination. This process establishes a standard process recipe library and selects corresponding process parameters for production according to the product model. The process recipe library stores fixed parameter combinations such as the dispensing volume, dispensing position, and lamination pressure corresponding to different product models. During the production process, the equipment executes production operations according to the selected process recipe and tracks the changes in key process parameters through a real-time monitoring system.

[0004] However, in the mixed-line production mode, due to the frequent switching of products in different batches and of different sizes, the fixed process recipe is difficult to quickly adapt to product changes. When quality anomalies occur, the problem location and solution cycle in the fixed recipe mode are relatively long, affecting production continuity. Summary of the Invention

[0005] This application provides a production control method, system, medium and product for a liquid crystal display screen, which is used to achieve rapid adaptation and intelligent optimization of the production process and improve the mixed-line production efficiency.

[0006] In a first aspect, the present application provides a production control method for a liquid crystal display screen, which is applied to a control system. The method includes: obtaining display unit parameters of a current production task; the display unit parameters include the size of the effective display area, the electrode spacing value, and the product identification number; extracting corresponding reference product parameters and reference process parameters from a process recipe library; the reference process parameters include the liquid crystal dispensing amount, the dispensing position, and the lamination pressure; correcting the reference process parameters to initial process parameters according to the difference coefficient between the display unit parameters and the reference product parameters; during the process of performing the liquid crystal dropping process with the initial process parameters, real-time collecting liquid crystal spreading state data, and generating a process stability index according to the liquid crystal spreading state data; the liquid crystal spreading state data includes the spreading area ratio and the bubble distribution information; when the process stability index deviates from a preset stable range, performing parameter correction based on a process parameter compensation model to obtain process correction parameters and their corresponding product characteristics; performing the liquid crystal dropping process with the process correction parameters, and storing the process correction parameters and the product characteristics in the process recipe library.

[0007] In the above embodiment, the control system realizes the adaptive adjustment and optimization of process parameters through the process of extracting reference process parameters, determining the difference coefficient and correcting it, real-time collecting the spreading state and performing parameter correction, can intelligently match appropriate process parameters according to different product characteristics, and dynamically adjust based on real-time monitoring data to ensure the stability of the production process, and improve the mixed-line production efficiency and product yield.

[0008] Combined with some embodiments of the first aspect, in some embodiments, the step of correcting the reference process parameters to initial process parameters according to the difference coefficient between the display unit parameters and the reference product parameters specifically includes: determining a size difference coefficient, a spacing difference coefficient, and a structure difference coefficient based on the display unit parameters and the reference product parameters; calculating a liquid crystal usage correction coefficient, a dispensing position offset amount, and a pressure compensation value according to the weighted combination value of the size difference coefficient, the spacing difference coefficient, and the structure difference coefficient; multiplying the liquid crystal dispensing amount in the reference process parameters by the liquid crystal usage correction coefficient to obtain the initial dispensing amount; performing coordinate transformation on the dispensing position in the reference process parameters based on the dispensing position offset amount to obtain the initial dispensing position; adding the lamination pressure in the reference process parameters and the pressure compensation value to obtain the initial lamination pressure; integrating the initial dispensing amount, the initial dispensing position, and the initial lamination pressure into the initial process parameters.

[0009] In the above embodiment, the control system calculates the correction coefficient and corrects the parameters, establishing a mapping relationship between product characteristics and process parameters, can accurately quantify the difference degree between different products, and adjust the process parameters accordingly in a targeted manner, making the process parameters more in line with the actual production requirements and improving the accuracy of parameter adaptation.

[0010] In some embodiments in combination with some embodiments of the first aspect, during the process of performing the liquid crystal dropping process with initial process parameters, the steps of collecting liquid crystal spreading state data in real time and generating a process stability index based on the liquid crystal spreading state data specifically include: collecting an image sequence of the liquid crystal spreading process in real time, performing image segmentation and feature extraction on the image sequence to obtain the dynamic spreading contour of the liquid crystal material; determining the spreading area change rate and the bubble number change rate per unit time according to the dynamic spreading contour; and determining the process stability index according to the spreading area change rate and the bubble number change rate.

[0011] In the above embodiments, the control system realizes precise monitoring of the liquid crystal spreading process by collecting and analyzing the liquid crystal spreading image sequence in real time, extracting the dynamic spreading contour features, and calculating the key index change rates, can timely detect abnormal conditions, and ensures the stability of the production process and the product quality.

[0012] In some embodiments in combination with some embodiments of the first aspect, before the step of obtaining the display unit parameters of the current production task, the method further includes: cleaning and annotating the historical production data to construct a training sample set; performing model training based on the training sample set to obtain an initial process parameter model; collecting environmental parameters including temperature, humidity, and air pressure of the production equipment, and establishing an environmental mapping relationship between the environmental parameters and the process parameters; and performing environmental compensation on the initial process parameter model based on the environmental mapping relationship to obtain a process parameter compensation model.

[0013] In the above embodiments, the control system considers the influence of environmental factors on the process by cleaning and annotating historical data, model training, and environmental compensation, and improves the adaptability and accuracy of the model.

[0014] In some embodiments in combination with some embodiments of the first aspect, the steps of cleaning and annotating the historical production data to construct a training sample set specifically include: determining the dispensing historical data of the nozzle pressure, dispensing speed, and liquid crystal material temperature of the liquid crystal dispensing equipment; determining the bonding historical records of the pressure curve, bonding speed, and vacuum degree of the liquid crystal bonding equipment; determining the quality inspection data including alignment accuracy, dead pixel defects, and foreign object contamination; associating and matching the dispensing historical data and the bonding historical records with the quality inspection data based on the time stamp to obtain historical associated data; and classifying and annotating the historical associated data according to the product model and defect type to obtain a training sample set including process parameter sequences and quality ratings.

[0015] In the above embodiments, the control system constructs a high-quality training sample set by collecting multi-dimensional historical data, and improves the accuracy and reliability of process parameter optimization based on the complete data processing flow.

[0016] In some embodiments in combination with some embodiments of the first aspect, after the steps of performing the liquid crystal dropping process with process correction parameters and storing the process correction parameters and product characteristics in the process recipe library, the method further includes: determining a predicted product yield rate according to the process correction parameters; collecting in real time production status data including dispensing uniformity, bubble distribution, spreading rate, and mixed-line switching frequency; determining a process anomaly risk probability based on the production status data; and generating a production warning message when the predicted product yield rate is lower than a preset yield rate threshold or when the process anomaly risk probability is greater than a preset probability threshold.

[0017] In the above embodiments, the control system realizes the full - range monitoring of the production process through means such as yield prediction, status monitoring, and anomaly warning, can timely discover potential problems and give warnings, provides a decision - making basis for process parameter optimization, and improves the initiative and predictability of production management.

[0018] In some embodiments in combination with some embodiments of the first aspect, after the step of generating a production warning message when the predicted product yield rate is lower than a preset yield rate threshold or when the process anomaly risk probability is greater than a preset probability threshold, the method further includes: extracting the parameter deviation trend and quality impact assessment in the production warning message to generate a process parameter optimization strategy; determining multiple groups of candidate process parameter combinations in the process parameter optimization strategy and performing simulation verification on each group of candidate process parameters to obtain corresponding simulation results; the simulation results include liquid crystal spreading uniformity, bubble residue rate, and expected product yield rate; performing a simulation score on the simulation results and selecting the candidate process parameter with the highest simulation score as the optimized process parameter; using the optimized process parameter for production verification and collecting in real time process data including liquid crystal injection volume, spreading speed, bubble size, and distribution position; calculating the liquid crystal material utilization rate, product qualification rate, and equipment operation rate according to the process data to generate an optimization evaluation report including cost - benefit analysis.

[0019] In the above embodiments, the control system can scientifically evaluate the optimization effect and continuously improve the process level through production verification and effect evaluation, realizing the continuous optimization of the production process and the improvement of benefits.

[0020] In a second aspect, an embodiment of the present application provides a control system, which includes: one or more processors and a memory; the memory is coupled to the one or more processors, and the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the control system to execute the method described in the first aspect and any possible implementation manner in the first aspect.

[0021] In a third aspect, an embodiment of the present application provides a computer program product including instructions. When the computer program product runs on a control system, the control system is caused to execute the method described in the first aspect and any possible implementation manner in the first aspect.

[0022] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium including instructions. When the instructions run on a control system, the control system is caused to execute the method described in the first aspect and any possible implementation manner in the first aspect.

[0023] It can be understood that the control system provided in the second aspect, the computer program product provided in the third aspect, and the computer storage medium provided in the fourth aspect are all used to execute the method provided in the embodiments of the present application. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding method, and will not be elaborated here.

[0024] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: 1. Due to the adoption of the difference analysis and adaptive adjustment mechanism based on the display unit parameters and reference process parameters, combined with the method of real-time liquid crystal spreading state monitoring and dynamic correction of process parameters, it is possible to achieve precise matching and dynamic optimization of process parameters, effectively solve the problems in the prior art that fixed process recipes are difficult to quickly adapt to product switching and the parameter adjustment efficiency is low, and thus achieve rapid adaptation and intelligent optimization of the production process, significantly improving the mixed-line production efficiency and product yield.

[0025] 2. Due to the adoption of the systematic method of cleaning and labeling, model training, and environmental compensation based on historical production data, and integrating the influence of environmental parameters of production equipment, a process parameter compensation model with environmental adaptability is established, effectively solving the problems in the prior art that environmental factors are not considered in process parameter optimization and the model generalization ability is poor, and thus achieving precise compensation and intelligent adjustment of process parameters.

[0026] 3. Due to the adoption of the dual monitoring mechanism based on the yield prediction model and the anomaly warning model, combined with the real-time analysis of multi-dimensional production status data, a comprehensive production process monitoring and warning system is constructed, effectively solving the problems in the prior art that production anomalies are discovered late and problem handling is passive, and thus achieving active monitoring and predictive management of the production process. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 is a flowchart of a method for controlling the production of a liquid crystal display screen in an embodiment of the present application; Figure 2 is another flowchart of a method for controlling the production of a liquid crystal display screen in an embodiment of the present application; Figure 3 It is a schematic structural diagram of an entity device in the control system in the embodiments of the present application. Specific embodiments

[0028] The terms used in the following embodiments of the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in the specification of the present application, the singular forms "a", "an", "above-mentioned", "the", and "this" are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used in the present application refers to any or all possible combinations including one or more of the listed items.

[0029] Hereinafter, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as implying or suggesting relative importance or implicitly indicating the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the embodiments of the present application, unless otherwise specified, the meaning of "a plurality" is two or more.

[0030] For ease of understanding, the application scenarios of the embodiments of the present application are introduced below.

[0031] A certain LCD manufacturer needs to switch the production of multiple types of display panels on the same production line every day. For example, in the morning, it produces a 10.4-inch in-vehicle central control display screen, switches to a 15.6-inch laptop computer panel at noon, and then produces a 12.1-inch industrial control touch screen in the afternoon. Products in these different application scenarios have special requirements for liquid crystal spreading uniformity, viewing angle characteristics, and anti-seismic performance. Each time the product model is switched, process parameters such as liquid crystal dispensing volume, dispensing position, and lamination pressure need to be adjusted. The traditional method is for engineers to manually adjust the parameters based on experience, which is not only time-consuming but also error-prone. Especially during the peak demand season for laptop computer panels, frequent product switching leads to large fluctuations in the yield rate, causing significant losses.

[0032] In the related art, the mixed-line production control of liquid crystal display panels can be achieved by adopting the standard process recipe library method, and presetting fixed process parameter combinations for each product model, including parameters such as dispensing volume, dispensing position, and lamination pressure. The scenario of using the production control method of the liquid crystal display screen in the related art is introduced below.

[0033] A certain panel factory adopts a process recipe library management method, establishing fixed process parameter recipes for each product model. When it is necessary to switch products, the system calls the parameter combinations in the corresponding recipe. However, problems occur when producing in-vehicle display screens: Since in-vehicle products have high requirements for temperature adaptability, it is difficult for fixed recipes to effectively adjust parameters automatically according to the ambient temperature; for industrial control touch screen products, due to the complex structure, equipment differences between multiple production lines will result in inconsistent effects for the same recipe. Once, due to an abnormal increase in the workshop temperature, the fixed recipe failed to be adjusted in time, resulting in the yield of a whole batch of laptop computer panels being lower than the normal level.

[0034] By adopting the production control method of the liquid crystal display screen in the embodiment of the present application, through establishing the mapping relationship between product characteristics and process parameters, combining real-time status monitoring and dynamic compensation mechanisms, the adaptive adjustment of process parameters is realized. It can not only quickly adapt to new products, but also automatically optimize parameters according to environmental changes. The following introduces the scenarios where the production control method of the liquid crystal display screen in the present application is used.

[0035] After introducing this solution, the production line of this panel factory has achieved intelligent control. When switching to a new product, the system automatically analyzes the differences from existing products and intelligently derives the initial process parameters. By real-time monitoring the liquid crystal spreading state, the system can dynamically optimize the parameter settings. For example, when it detects that the temperature of the industrial control terminal production line rises, it automatically adjusts the temperature of the liquid crystal material and the dispensing pressure to ensure the stable quality of touch screen products. When switching from in-vehicle display screen production to laptop computer panel production once, the system completed parameter optimization in only 30 minutes, and the yield reached 98.5%.

[0036] It can be seen that by adopting the production control method of the liquid crystal display screen in the embodiment of the present application, while realizing rapid product switching, it can also effectively solve the problem that fixed process recipes cannot cope with environmental changes and equipment differences, and realizes the improvement of production efficiency and yield.

[0037] For the convenience of understanding, the method provided in this embodiment will be described in terms of its process in combination with the above scenarios. Please refer to Figure 1 , which is a schematic flow diagram of the production control method of the liquid crystal display screen in the embodiment of the present application.

[0038] S101. Obtain the display unit parameters of the current production task.

[0039] Among them, the display unit parameters represent the key technical indicators in the production process of the liquid crystal display panel, including the size of the effective display area, the electrode spacing value, and the product identification number; among them, the size of the effective display area refers to the size of the area where the liquid crystal display panel can actually display images, usually expressed in millimeters for the length and width dimensions; the electrode spacing value is used to represent the gap distance between the array substrate and the color filter substrate in the liquid crystal display panel, and this parameter directly affects the arrangement effect of liquid crystal molecules; the product identification number is the code used to uniquely identify different product models, including information such as product series and size.

[0040] The control system needs to execute this step when receiving a new production task to obtain the basic parameter information required for subsequent production. Specifically, the control system first obtains the product information to be produced currently through the production management system interface, and then extracts the technical parameter information of this product from the product design database, including physical parameters such as the size of the effective display area and the electrode spacing value of the product, and at the same time obtains the product identification number for subsequent process parameter matching. The system normalizes and validates these parameters to ensure the integrity and accuracy of the data.

[0041] In some embodiments, the display unit parameters can be obtained in multiple ways: Optionally, it can be achieved through the following steps: Connect to the manufacturing execution system (MES) to obtain the current production task information; query the product database according to the task information to obtain detailed parameters; perform format conversion and unit unification on the parameters; perform parameter integrity and validity verification; generate a parameter data packet in a standard format. Optionally, it can be achieved through the following steps: Read the production task configuration file to obtain product information; call the parameter management system API to obtain technical parameters; parse the parameter data to extract key indicators; perform parameter range checks and outlier processing; integrate to form a parameter dataset. It can be understood that other methods can also be used to obtain and process the display unit parameters, such as manual entry, scanning product barcodes, etc., which are not limited here.

[0042] S102. Extract the reference product parameters and reference process parameters corresponding to the product identification number from the process recipe library.

[0043] Among them, the process recipe library is a database system that stores the production process parameters of various products, including the optimal process parameter combinations verified historically. The reference product parameters are the standard product parameters corresponding to the display unit parameters. The reference process parameters represent the standard production parameter set for a specific product model, including key process parameters such as liquid crystal dispensing volume, dispensing position, and lamination pressure; among them, the liquid crystal dispensing volume refers to the volume of the liquid crystal material injected each time, usually expressed in nanoliters (nL); the dispensing position is used to represent the injection coordinate point of the liquid crystal material, usually expressed in the form of XY coordinates; the lamination pressure is the pressure value applied when assembling the upper and lower substrates, usually expressed in pascals (Pa).

[0044] After the control system obtains the display unit parameters, it needs to execute this step to obtain the corresponding reference process parameters. Specifically, the control system first queries the process recipe library according to the product identification number and locates the process recipe record corresponding to the product. Then, the system extracts the set of reference process parameters in this record, including parameters such as liquid crystal dispensing volume, dispensing position distribution map, bonding pressure curve, etc. At the same time, the system also obtains the effective range and compensation coefficient of these parameters, providing a basis for subsequent parameter optimization.

[0045] S103. Correct the reference process parameters to initial process parameters according to the difference coefficient between the display unit parameters and the reference product parameters.

[0046] Among them, the difference coefficient is a quantitative index that measures the degree of difference between the current product and the reference product in various dimension parameters, including size difference coefficient, spacing difference coefficient, and structure difference coefficient. The initial process parameters represent the optimized process parameter set after difference compensation, including initial dispensing volume, initial dispensing position, and initial bonding pressure.

[0047] After the control system obtains the reference process parameters, it needs to execute this step to achieve intelligent optimization of the parameters. Specifically, the control system first analyzes the difference characteristics between the current product and the reference product through a deep learning model and calculates the difference coefficients of each dimension. Then, the system calculates the liquid crystal usage correction coefficient, dispensing position offset, and pressure compensation value respectively according to the weighted combination value of the size difference coefficient, spacing difference coefficient, and structure difference coefficient; the liquid crystal usage correction coefficient is used to represent the adjustment ratio of the liquid crystal usage; the dispensing position offset is the position compensation value of the dispensing coordinates; the pressure compensation value is used to represent the adjustment amount of the bonding pressure. Finally, the system applies these correction amounts to the reference process parameters to generate the initial process parameters applicable to the current product.

[0048] In some embodiments, the difference coefficient calculation and parameter correction can be achieved in multiple ways: Optionally, it can be achieved through the following steps: Extract the feature vectors of the current product and the reference product; Calculate the difference metric through a deep neural network; Generate multi-dimensional difference coefficients; Calculate the parameter correction amount; Apply the correction amount to obtain the initial parameters. Optionally, it can be achieved through the following steps: Construct a product feature comparison matrix; Use a fuzzy calculation method to evaluate the degree of difference; Establish a difference-correction mapping relationship; Calculate the parameter compensation value; Update the process parameter set. It can be understood that other methods can also be used to achieve difference analysis and parameter correction, such as based on expert experience rules, statistical modeling, etc., which are not limited here.

[0049] S104. During the process of executing the liquid crystal dropping process with the initial process parameters, real-time collect the liquid crystal spreading state data and generate a process stability index according to the liquid crystal spreading state data.

[0050] Among them, the liquid crystal spreading state data represents the dynamic characteristic information during the diffusion process of the liquid crystal material on the substrate, including the spreading area ratio and the bubble distribution information; the spreading area ratio refers to the ratio of the actual coverage area of the liquid crystal material to the theoretical area; the bubble distribution information is used to represent the size, quantity, and position distribution of the bubbles generated during the spreading process. The process stability index refers to a comprehensive index used to evaluate the stability degree of the production process, usually including parameters such as spreading uniformity and bubble residue rate.

[0051] After obtaining the initial process parameters, the control system needs to execute this step to achieve real-time monitoring of the production process. Specifically, the control system first sends the initial process parameters to the production equipment to start the liquid crystal dropping process. At the same time, the system uses the vision detection system to collect liquid crystal spreading images in real time, perform image processing and feature extraction to obtain dynamic data such as the spreading area change rate and the bubble quantity change rate. The system inputs these data into a preset evaluation model and calculates a comprehensive index reflecting the process stability.

[0052] In some embodiments, the spreading state monitoring and stability evaluation can be achieved in multiple ways: Optionally, it can be achieved through the following steps: Configure a high-speed camera to collect a sequence of spreading images; execute image preprocessing and segmentation algorithms; extract the spreading contour and bubble features; calculate the dynamic change index; generate a stability score. Optionally, it can be achieved through the following steps: Deploy a multi-sensor data acquisition system; obtain process data such as pressure and temperature in real time; fuse multi-source data to calculate state features; evaluate the process fluctuation degree; output a stability report. It can be understood that other methods can also be used to achieve production process monitoring and stability evaluation, such as based on optical detection, acoustic detection, etc., which are not limited here.

[0053] S105. When the process stability index deviates from the preset stable range, perform parameter correction based on the process parameter compensation model to obtain the process correction parameters and their corresponding product characteristics.

[0054] Among them, the preset stable range refers to the allowable fluctuation interval of the process stability index, which is used to judge whether the production process needs to be adjusted. The process correction parameters refer to the final set of process parameters after optimization and adjustment. The product characteristics are used to represent the product specifications and process requirement information corresponding to the correction parameters.

[0055] When the control system detects process fluctuations, it needs to execute this step to achieve dynamic optimization of the parameters. Specifically, the control system first determines whether the process stability index exceeds the preset range. When adjustment is required, the system activates the process parameter compensation model, determines the deviation between the current state and the target state based on the process parameter compensation model, and calculates the correction amount of each parameter. The system comprehensively considers the coupling relationship between the parameters to generate the optimal process correction parameters and records the corresponding product characteristic information at the same time.

[0056] It should be noted that when training the process parameter compensation model, a large amount of complete historical production data is first required. The data acquisition end extracts environmental parameter data from the historical database of production equipment, including temperature data in the range of 15-35°C, relative humidity data in the range of 35%-65%, and air pressure data in the range of 980-1020hPa. At the same time, the optimal process parameter data corresponding to these environmental conditions are collected, including the liquid crystal dispensing volume accurate to 0.1nL, the dispensing position coordinates with an accuracy of 0.01mm in the XY direction, the fitting pressure value with an accuracy of 0.1Pa, and the product quality data corresponding to each set of parameters, such as key indicators such as spreading uniformity and bubble residual rate. Then enter the data preprocessing stage, first eliminate the data generated by abnormal working conditions such as equipment failure and material breakage. The retained valid data is normalized, and the numerical ranges of environmental parameters and process parameters are uniformly mapped to the [-1, 1] interval. Then, the environmental parameters and process parameters are aligned according to the timestamp to construct a seven-dimensional input feature vector (including three environmental parameters: temperature, humidity, and air pressure, and four process parameters: baseline dispensing amount, baseline dispensing position X / Y, and baseline bonding pressure) and a four-dimensional output label vector (including optimal dispensing amount, optimal dispensing position X / Y, and optimal bonding pressure).

[0057] The process parameter compensation model is trained using a multi-layer perceptron network structure. The network contains four hidden layers with 128, 256, 128, and 64 nodes, respectively, which reflects the design idea of ​​feature extraction first and dimensionality reduction compression later. Each hidden layer uses the ReLU activation function, and the output layer uses linear activation to obtain continuous parameter prediction values. The mean square error is selected as the loss function, and the parameters are optimized by the Adam optimizer. The learning rate is set to 0.001 and the iteration batch size is 32. The data is divided into a training set and a validation set in a ratio of 8:2. The validation set loss is monitored during training. The early stopping mechanism is triggered when the validation loss no longer decreases for 10 consecutive epochs. The training goal is to make the MSE of the predicted process parameters and the optimal process parameters less than the preset threshold of 0.05. The input layer of the process parameter compensation model is set with 7 nodes, which receive three environmental parameters and four benchmark process parameters respectively. The design purpose of the four hidden layers is to gradually extract features and establish complex relationships between parameters. A batch normalization layer is added after the second hidden layer to accelerate training convergence, and a layer with a dropout rate of 0.3 is set after the third hidden layer to prevent overfitting. In particular, the residual connection is introduced to connect the input layer directly to the third hidden layer, which helps to maintain the original parameter information. The four nodes of the output layer correspond to the compensated dispensing amount, the X / Y coordinates of the dispensing position, and the fitting pressure.

[0058] When using this process parameter compensation model, the control system first collects the current environmental parameters in real time, obtains the reference process parameters corresponding to the current product, normalizes all input data, and then sends it into the model. The process parameter compensation model outputs the compensated process parameters, and through inverse normalization, the actual executable process parameter values are obtained. The model automatically collects environmental parameters and updates process parameters every 10 minutes. When the environmental parameters change by more than the preset threshold, an immediate update will also be triggered. The compensated process parameters are sent to the production equipment for execution in real time, and at the same time, the product quality after the execution of the compensation parameters is continuously monitored, and the compensation effect data is recorded for subsequent model optimization.

[0059] S106. Execute the liquid crystal dropping process with the process correction parameters, and store the process correction parameters and product characteristics in the process recipe library.

[0060] Among them, the storage of product characteristics is used to establish the correspondence between product specifications and process parameters, which is convenient for subsequent rapid call.

[0061] After the control system obtains the optimized process parameters, it needs to execute this step to verify and save the optimization results. Specifically, the control system first sends the corrected process parameters to the production equipment and executes the liquid crystal dropping process again. After confirming that the parameter effect meets the requirements, the system stores this set of parameters and their corresponding product characteristic information in the process recipe library and updates the recipe database. At the same time, the system establishes an association index between the parameters and product characteristics to optimize the recipe retrieval efficiency.

[0062] In some embodiments, parameter verification and storage can be achieved in multiple ways: Optionally, it can be achieved through the following steps: execute the verification production batch; collect quality inspection data; evaluate the parameter optimization effect; update the recipe database; establish a parameter index relationship. Optionally, it can be achieved through the following steps: configure the parameter verification scheme; monitor the production process data; analyze the yield improvement effect; store the effective recipe; optimize the retrieval structure. It can be understood that other methods can also be used to verify and store process parameters, such as through distributed storage, cloud synchronization, etc., which are not limited here.

[0063] The following further describes the more specific process of the method provided in this embodiment. Please refer to Figure 2 , which is another process schematic diagram of the production control method for liquid crystal display screens in the embodiments of the present application.

[0064] S201. Clean and annotate the historical production data to construct a training sample set.

[0065] Among them, the historical production data represents various process data and quality records accumulated during the production process of liquid crystal display panels.

[0066] Before establishing the process parameter optimization model, the control system needs to execute this step to obtain high-quality training data. Specifically, the control system first extracts historical production records from the production management system, including equipment operation data, process parameter records, and quality inspection results. Then, the system classifies and organizes the data, excluding abnormal data caused by factors such as equipment failures, material shortages, and human interventions. Next, the system correlates and matches the dispensing, bonding, etc. process data with the quality inspection results based on timestamps to establish data correlation relationships. Finally, the system classifies and labels the data according to product models and defect types to form a standardized training sample set.

[0067] In some embodiments, the control system determines the dispensing historical data of the nozzle pressure, dispensing speed, and liquid crystal material temperature of the liquid crystal dispensing equipment; determines the bonding historical records of the pressure curve, bonding speed, and vacuum degree of the liquid crystal bonding equipment; determines the quality inspection data including alignment accuracy, dead pixel defects, and foreign object contamination; correlates and matches the dispensing historical data and bonding historical records with the quality inspection data based on timestamps to obtain historical correlation data; classifies and labels the historical correlation data according to product models and defect types to obtain a training sample set including process parameter sequences and quality ratings.

[0068] Among them, the dispensing historical data represents the process parameter information recorded during the operation of the liquid crystal dispensing equipment; the nozzle pressure refers to the pressure value when the dispensing equipment outputs the liquid crystal material; the dispensing speed is used to represent the injection rate of the liquid crystal material; the liquid crystal material temperature represents the working temperature of the liquid crystal during dispensing. The bonding historical record refers to the process parameter record during the liquid crystal bonding process; the pressure curve is used to represent the change characteristics of the pressure over time during bonding; the bonding speed refers to the movement rate during the assembly of the upper and lower substrates; the vacuum degree represents the vacuum state parameter in the bonding cavity. The quality inspection data refers to the inspection results related to product quality; the alignment accuracy is used to represent the accuracy of the alignment of the upper and lower substrates; the dead pixel defect refers to the abnormal display of the display unit; the foreign object contamination represents the impurity particles introduced during the production process.

[0069] Before establishing the process parameter optimization model, the control system needs to execute this step to construct a high-quality training data set. Specifically, the control system first extracts the historical parameter records of the dispensing and bonding processes from the production equipment database, including equipment operation parameters and process condition data. Then, the system obtains product quality data from the quality inspection system, including various defect information. The system matches the process data with the quality data based on the time series to establish a causal relationship. Finally, the system classifies and labels the valid data to form a standardized training sample set. In some embodiments, the control system also identifies and excludes invalid data caused by equipment abnormalities and other reasons.

[0070] In some embodiments, historical data processing and sample set construction can be achieved in various ways: Optionally, it can be achieved through the following steps: Configure a data acquisition interface; perform data format conversion; establish a time series index; perform data cleaning; perform association analysis; generate labeled samples. Optionally, it can be achieved through the following steps: Design a data extraction scheme; perform anomaly detection; construct an association matrix; filter valid data; label sample features; verify data quality. It can be understood that other methods can also be used to process historical data and construct training samples, such as through data mining, feature engineering, etc., which are not limited here.

[0071] S202. Perform model training based on the training sample set to obtain an initial process parameter model.

[0072] After obtaining the training sample set, the control system needs to execute this step to construct an optimized process parameter model. Specifically, the control system first performs feature engineering on the training samples to extract key process parameter features. Then, the system selects an appropriate machine learning algorithm, such as a deep neural network, and sets the model structure and hyperparameters. Next, the system iteratively trains the model using the training data and optimizes the model parameters through a loss function. Finally, the system evaluates the model performance using a validation set to ensure that the model has good generalization ability.

[0073] In some embodiments, model training and optimization can be achieved in various ways: Optionally, it can be achieved through the following steps: Design a model architecture and a loss function; divide the training set and the validation set; perform a batch training process; monitor the model convergence; evaluate the model performance metrics; save the optimal model parameters. Optionally, it can be achieved through the following steps: Construct a feature vector space; select the type of learning algorithm; configure the training hyperparameters; perform cross-validation; analyze the model stability; export the model file. It can be understood that other methods can also be used to train and optimize the model, such as through transfer learning, ensemble learning, etc., which are not limited here.

[0074] S203. Collect environmental parameters of the production equipment, including temperature, humidity, and air pressure, and establish an environmental mapping relationship between the environmental parameters and the process parameters.

[0075] Among them, the environmental parameters represent the measured values of various environmental factors affecting the production process; the temperature parameter refers to the temperature change data of the production environment; the humidity parameter is used to represent the fluctuation of the relative air humidity; the air pressure parameter refers to the change data of the atmospheric pressure.

[0076] After obtaining the initial process parameter model, the control system needs to execute this step to achieve environmental adaptability optimization. Specifically, the control system first collects production environment data in real time through a distributed sensor network, including parameters such as temperature, humidity, and air pressure. Then, the system analyzes the correlation between environmental parameters and process indicators to establish environmental impact factors. Next, the system constructs a mapping function between environmental parameters and process parameters to quantify the impact of environmental changes on the process. Finally, the system verifies the accuracy of the mapping relationship and establishes an environmental compensation mechanism.

[0077] In some embodiments, environmental parameter acquisition and mapping relationship establishment can be achieved in multiple ways: Optionally, it can be achieved through the following steps: Deploy environmental monitoring sensors; collect multi-dimensional environmental data; analyze parameter correlations; establish a mapping function model; verify the compensation effect; optimize the mapping relationship. Optionally, it can be achieved through the following steps: Configure a data acquisition system; perform data preprocessing; calculate impact factors; construct a regression model; evaluate model accuracy; generate a compensation strategy. It can be understood that other methods can also be used to achieve environmental parameter acquisition and mapping relationship establishment, such as through fuzzy control, neural networks, etc., which are not limited here.

[0078] S204. Perform environmental compensation on the initial process parameter model based on the environmental mapping relationship to obtain a process parameter compensation model.

[0079] After establishing the environmental mapping relationship, the control system needs to execute this step to improve the environmental adaptability of the model. Specifically, the control system first analyzes the sensitivity of each process parameter to environmental changes to determine the compensation weight. Then, the system integrates the environmental mapping relationship into the initial model to construct a compensation mechanism considering environmental impacts. Next, the system verifies the model performance under different environmental conditions and optimizes the compensation strategy. Finally, the system generates a process parameter compensation model with environmental adaptability.

[0080] In some embodiments, environmental compensation and model optimization can be achieved in multiple ways: Optionally, it can be achieved through the following steps: Evaluate parameter environmental sensitivity; design a compensation algorithm; update the model structure; perform model verification; optimize the compensation strategy; export the compensation model. Optionally, it can be achieved through the following steps: Construct an environmental impact matrix; calculate compensation coefficients; integrate the compensation mechanism; test model performance; adjust compensation parameters; generate the final model. It can be understood that other methods can also be used to achieve environmental compensation and model optimization, such as through adaptive control, dynamic programming, etc., which are not limited here.

[0081] S205. Obtain the display unit parameters of the current production task.

[0082] Referring to step S101, the control system will determine the display unit parameters.

[0083] S206. Extract the reference product parameters and reference process parameters corresponding to the product identification number from the process recipe library.

[0084] Referring to step S102, the control system determines the reference process parameters.

[0085] S207. Correct the reference process parameters to initial process parameters according to the difference coefficient between the display unit parameters and the reference product parameters.

[0086] Referring to step S103, the control system generates the initial process parameters.

[0087] In some embodiments, the control system determines the size difference coefficient, the spacing difference coefficient, and the structure difference coefficient based on the display unit parameters and the reference product parameters; calculates the liquid crystal dosage correction coefficient, the dispensing position offset, and the pressure compensation value according to the weighted combination value of the size difference coefficient, the spacing difference coefficient, and the structure difference coefficient; multiplies the liquid crystal dispensing amount in the reference process parameters by the liquid crystal dosage correction coefficient to obtain the initial dispensing amount; performs coordinate transformation on the dispensing position in the reference process parameters based on the dispensing position offset to obtain the initial dispensing position; superimposes the bonding pressure in the reference process parameters and the pressure compensation value to obtain the initial bonding pressure; integrates the initial dispensing amount, the initial dispensing position, and the initial bonding pressure into the initial process parameters.

[0088] Among them, the size difference coefficient is the deviation quantization value of the product size parameters; the spacing difference coefficient is used to represent the change degree of the electrode spacing; the structure difference coefficient represents the difference quantization value of the product structure characteristics. The liquid crystal dosage correction coefficient is used to represent the adjustment ratio of the dispensing amount; the dispensing position offset is the compensation distance of the dispensing coordinates; the pressure compensation value represents the adjustment amount of the bonding pressure.

[0089] After obtaining the reference process parameters, the control system needs to execute this step to achieve differential adjustment of the parameters. Specifically, the control system first analyzes the differences between the current product and the reference product in terms of size, spacing, and structure through a deep learning algorithm to generate standardized difference coefficients. Then, the system sets the weights of each difference coefficient based on historical experience and expert knowledge and calculates the comprehensive score. Next, the system calculates the correction values of the liquid crystal dosage, the dispensing position, and the bonding pressure according to the comprehensive score. Finally, the system applies these correction values to the reference parameters to generate the initial process parameters adapted to the current product characteristics.

[0090] In some embodiments, parameter difference analysis and correction can be achieved in various ways: Optionally, it can be achieved through the following steps: Extract parameter feature vectors; calculate the difference matrix; determine the weight coefficients; generate the correction amount; perform parameter adjustment; verify the parameter effectiveness. Optionally, it can be achieved through the following steps: Construct a feature comparison model; analyze the degree of difference; optimize the weight configuration; calculate the compensation value; update the process parameters; evaluate the adjustment effect. It can be understood that other methods can also be used to achieve parameter difference analysis and correction, such as through fuzzy inference, adaptive control, etc., which are not limited here.

[0091] S208. During the process of performing the liquid crystal dropping process with the initial process parameters, the liquid crystal spreading state data is collected in real time, and a process stability index is generated based on the liquid crystal spreading state data.

[0092] Referring to step S104, the control system will determine the process stability index.

[0093] In some embodiments, the control system will collect the image sequence of the liquid crystal spreading process in real time, perform image segmentation and feature extraction on the image sequence, and obtain the dynamic spreading contour of the liquid crystal material; based on the dynamic spreading contour, determine the spreading area change rate and the bubble number change rate per unit time; based on the spreading area change rate and the bubble number change rate, determine the process stability index.

[0094] Among them, the image sequence represents the images of the liquid crystal spreading process collected continuously. The dynamic spreading contour refers to the boundary shape of the liquid crystal material changing with time. The spreading area change rate represents the change speed of the liquid crystal coverage area per unit time; the bubble number change rate is used to represent the dynamic characteristics of bubble generation and disappearance.

[0095] During the process of the control system performing the liquid crystal dropping process, this step needs to be executed to achieve real-time monitoring of the process. Specifically, the control system first collects the liquid crystal spreading images in real time through a high-speed industrial camera to form a continuous image sequence. Then, the system uses image processing algorithms to preprocess and segment each frame of the image, and extract the contour of the liquid crystal region. The system tracks the contour change, calculates the area change rate and the bubble characteristics. Finally, the system inputs these dynamic characteristics into the evaluation model to generate a quantitative index reflecting the process stability.

[0096] It should be noted that the training data of the preset evaluation model is sourced from the real-time monitoring system on the production line, mainly including the high-speed camera image sequence data during the liquid crystal spreading process. Each set of training data contains an image sequence with a continuous duration of 1 minute and a frame rate of 60 fps, and is equipped with verified process stability score labels. Through preprocessing, key dynamic features such as the spreading area change rate (the ratio of the spreading area increment per second to the theoretical area) and the bubble number change rate (the ratio of the change in the number of bubbles per second to the initial number of bubbles) are obtained from the image data. In the data preprocessing stage, computer vision algorithms are used to process the image sequence; first, image enhancement and noise reduction are performed, and then the adaptive threshold segmentation algorithm is used to extract the liquid crystal spreading area and the bubble area. Morphological processing is performed on the segmentation results to obtain accurate spreading contours and bubble distribution information. Then, the feature changes between consecutive frames are converted into time series data to construct the model input feature matrix. The feature matrix contains features in multiple time dimensions such as the spreading area sequence, the bubble number sequence, and the spreading rate sequence.

[0097] The preset evaluation model adopts a long short-term memory network (LSTM) structure to capture the temporal dynamic features of the liquid crystal spreading process. The network includes a bidirectional LSTM layer for feature extraction, an attention mechanism layer for highlighting key temporal features, and a fully connected layer for feature fusion. The model input is the feature sequence within 60 seconds, and the output is the process stability score within the range of 0 - 1. The mean squared error loss function is used for training, and the Adam optimizer is used for parameter optimization, with the learning rate set to 0.0005. The training objective is to make the mean squared error between the predicted score and the expert score less than 0.1.

[0098] When using the preset evaluation model, the control system continuously receives the real-time image stream and performs feature extraction and score prediction once per second. When the predicted score is lower than 0.8 or the scores show a downward trend for three consecutive times, the process parameter adjustment mechanism is triggered. The stability score output by the model is simultaneously used as the input of the process parameter compensation model to form a closed-loop control.

[0099] In some embodiments, process monitoring and evaluation can be achieved in multiple ways: Optionally, it can be achieved through the following steps: configure image acquisition parameters; perform real-time image processing; extract dynamic features; calculate change metrics; evaluate process stability; generate a monitoring report. Optionally, it can be achieved through the following steps: set camera acquisition conditions; construct an image processing flow; analyze contour features; track parameter changes; calculate stability metrics; output evaluation results. It can be understood that other methods can also be used to achieve process monitoring and evaluation, such as through multi-sensor fusion, deep learning, etc., which are not limited herein.

[0100] S209. When the process stability index deviates from the preset stable range, parameter correction is performed based on the process parameter compensation model to obtain the process correction parameters and their corresponding product characteristics.

[0101] Referring to step S105, the control system will perform parameter correction when the process stability index deviates from the preset stable range.

[0102] S210. Execute the liquid crystal dropping process with the process correction parameters, and store the process correction parameters and product characteristics in the process recipe library.

[0103] Referring to step S106, the control system will execute the liquid crystal dropping process and store the relevant parameters.

[0104] S211. Determine the predicted product yield according to the process correction parameters.

[0105] Among them, the predicted product yield refers to the qualified rate of the product estimated based on the current process parameters.

[0106] After obtaining the process correction parameters, the control system needs to execute this step to evaluate the parameter optimization effect. Specifically, the control system first inputs the corrected process parameters into the yield prediction model, taking into account the current environmental conditions and equipment status. Then, the system calculates the expected product yield based on historical data and parameter correlation. The system also analyzes the influence degree of each parameter on the yield, generates a quality risk assessment report. Finally, the system outputs the predicted yield result and the predicted confidence interval.

[0107] It should be noted that the training data of this yield prediction model comes from historical production records, including a complete set of process parameters and the corresponding product yield data. The set of process parameters includes core parameters such as liquid crystal dispensing volume (accurate to 0.1 nL), dispensing position coordinates (accuracy 0.01 mm), bonding pressure (accuracy 0.1 Pa), etc., as well as environmental parameters such as temperature, humidity, and air pressure. The yield data includes the final product yield and specific statistical information of various defects. In the data preprocessing stage, outlier detection and processing are first performed. The 3σ criterion is used to identify abnormal data points, and expert experience is combined to correct or eliminate outliers. All parameters are standardized, and the values are mapped to a unified interval. Special attention is paid to the correlation analysis between parameters, and the feature dimension is reduced by principal component analysis (PCA) to remove redundant information. At the same time, time window features are introduced to capture the trend information of parameter changes over time.

[0108] The yield prediction model adopts an ensemble learning framework, combining Gradient Boosting Decision Tree (GBDT) and Deep Neural Network (DNN). GBDT is good at dealing with non-linear relationships and feature combinations, while DNN is more suitable for extracting deep feature patterns. The prediction results of the two sub-models are fused through weighted averaging to obtain the final predicted yield. The training process uses cross-validation, with the Mean Squared Logarithmic Error (MSLE) as the loss function, which can balance the influence of high-yield and low-yield samples. The training objective is to control the relative error between the predicted yield and the actual yield within 3%.

[0109] When using this yield prediction model, the control system inputs the current combination of process parameters and environmental parameters, and the yield prediction model outputs the predicted yield and a 95% confidence interval. When the predicted yield is more than 2% lower than the target yield, the warning mechanism is triggered. The model is updated online regularly (such as every 4 hours) using the latest production data to maintain prediction accuracy.

[0110] S212. Real-time collect production status data including dispensing uniformity, bubble distribution, spreading rate, and mixed-line switching frequency.

[0111] Among them, dispensing uniformity represents the degree of consistency of the distribution of liquid crystal materials; bubble distribution refers to the spatial position and size characteristics of bubbles during the spreading of liquid crystals; spreading rate is used to represent the time-varying characteristics of the diffusion of liquid crystal materials; and mixed-line switching frequency refers to the time interval for the production line to switch between different product models.

[0112] During the production process, the control system needs to execute this step to achieve real-time monitoring of the process. Specifically, the control system first collects image sequences of the liquid crystal dispensing and spreading processes through multiple vision detection systems. Then, the system performs real-time processing on the collected images to extract key indicators such as dispensing uniformity and bubble characteristics. At the same time, the system records product switching information and calculates the mixed-line production rhythm. The system standardizes these data to form a data stream reflecting the current production status.

[0113] In some embodiments, the collection and processing of production status data can be achieved in various ways: Optionally, it can be achieved through the following steps: Configure a multi-source data acquisition system; perform real-time image processing; extract process feature parameters; calculate status indicators; record switching information; generate status data packets. Optionally, it can be achieved through the following steps: Deploy a detection sensor array; collect multi-dimensional monitoring data; analyze process characteristics; evaluate process stability; track product switching; integrate status information. It can be understood that other methods can also be used to achieve production status monitoring and data collection, such as through online detection, multi-sensor fusion, etc., which are not limited here.

[0114] S213. Determine the probability of process anomaly risk based on the production status data.

[0115] Among them, the process anomaly risk probability refers to the evaluation value of the possibility of quality problems occurring.

[0116] After obtaining the production status data, the control system needs to execute this step to achieve anomaly early warning. Specifically, the control system first inputs the real-time collected status data into the anomaly early warning model and matches it with the historical anomaly patterns. Then, the system calculates the occurrence probabilities of various anomalies and determines the risk levels according to the impact degrees. The system also analyzes the anomaly development trends and predicts potential risks. Finally, the system determines the early warning levels and processing priorities according to the risk assessment results.

[0117] It should be noted that the training data of this anomaly early warning model includes complete records of normal and abnormal production statuses. The real-time collected status data includes multi-dimensional parameters such as dispensing uniformity (uniformity score obtained through image analysis), bubble distribution (statistical distribution of bubble sizes and positions), spreading rate (change in spreading area per unit time), and mixed-line switching frequency (production switching interval time). Each status is labeled with normal / abnormal labels, as well as specific anomaly types and severities. In the data preprocessing stage, the sliding time window method is adopted to convert the continuous status data into sequences of fixed length. Statistical features including mean, standard deviation, trend coefficient, etc. are calculated for the data within each time window. The frequency domain features of the signal are extracted through wavelet transform to capture abnormal fluctuation patterns. At the same time, combined with expert experience, composite features such as the aggregation index of bubble distribution and the stability index of spreading uniformity are constructed.

[0118] This anomaly early warning model adopts a multi-scale convolutional neural network structure, combined with a self-attention mechanism. Convolution kernels of different scales are used to capture anomaly patterns of different time spans, and the self-attention mechanism is used to learn the dynamic correlations between parameters. The model outputs include the anomaly risk probability and anomaly type prediction. The focal loss function is used for training to solve the problem of sample imbalance. At the same time, a temporal consistency constraint is introduced to ensure the temporal continuity of the prediction results. The training goal is to achieve an anomaly detection rate of more than 95%, while controlling the false alarm rate below 5%.

[0119] In some embodiments, anomaly risk assessment and early warning can be achieved in various ways: Optionally, it can be achieved through the following steps: loading the early warning model parameters; executing the anomaly detection algorithm; calculating the risk probability; evaluating the anomaly level; generating early warning information; triggering the response mechanism. Optionally, it can be achieved through the following steps: constructing the status feature vector; matching the anomaly pattern library; analyzing the risk trend; determining the early warning level; formulating the processing strategy; pushing the early warning message. It can be understood that other ways can also be adopted to achieve anomaly early warning and risk assessment, such as through expert systems, rule reasoning, etc., which are not limited here.

[0120] S214. Generate a production warning message when the predicted yield of the product is lower than the preset yield threshold or when the probability of process anomaly risk is greater than the preset probability threshold.

[0121] When the control system detects an abnormal situation, it needs to execute this step to issue a warning in a timely manner. Specifically, the control system first compares the current predicted yield and the probability of abnormal risk with the preset thresholds. When the thresholds are exceeded, the system analyzes the historical change trajectory of the process parameters to identify the deviation trend. Then, the system evaluates the potential impact of the parameter deviation on the product quality, including the affected degree of different quality characteristics. Finally, the system integrates the analysis results to generate a warning report containing detailed information.

[0122] In some embodiments, the generation and release of the warning message can be achieved in various ways: Optionally, it can be achieved through the following steps: Compare the threshold trigger conditions; Analyze the parameter change trend; Evaluate the quality impact; Generate a warning report; Determine the processing suggestions; Push the warning message. Optionally, it can be achieved through the following steps: Detect the abnormal trigger conditions; Extract the trend features; Calculate the impact degree; Integrate the warning content; Set the warning level; Send the warning notice. It can be understood that other ways can also be adopted to achieve the generation and release of the warning message, such as through visual display, multi-channel push, etc., which are not limited here.

[0123] In some embodiments, the control system extracts the parameter deviation trend and the quality impact assessment in the production warning message to generate a process parameter optimization strategy; determines multiple groups of candidate process parameter combinations in the process parameter optimization strategy, and conducts simulation verification on each group of candidate process parameters to obtain the corresponding simulation results; the simulation results include the uniformity of liquid crystal spreading, the bubble residue rate, and the expected yield of the product; conducts a simulation score on the simulation results, and selects the candidate process parameter with the highest simulation score as the optimized process parameter; uses the optimized process parameter for production verification, and real-time collects the process data including the liquid crystal injection volume, spreading speed, bubble size, and distribution position; calculates the utilization rate of liquid crystal material, the product qualification rate, and the equipment operation rate according to the process data, and generates an optimization evaluation report including cost-benefit analysis.

[0124] Among them, the process parameter optimization strategy refers to the parameter adjustment plan formulated according to the abnormal analysis results. The parameter deviation trend is used to represent the direction and amplitude of the change of the process parameters over time. The quality impact assessment refers to the analysis result of the affected degree of the product quality. The candidate process parameter combination refers to multiple groups of optional parameter sets generated according to the optimization strategy. The expected yield of the product is used to represent the predicted product qualification rate. The simulation score refers to the comprehensive evaluation score of the simulation results.

[0125] When the control system detects a process anomaly, it needs to execute this step to achieve optimized adjustment of process parameters. Specifically, the control system first designs a parameter optimization plan based on the parameter deviation trend and quality impact analysis. Then, the system automatically generates multiple sets of parameter combinations according to the optimization strategy and verifies the effects of each set of parameters through a simulation system. The system scores the simulation results from multiple dimensions and selects the optimal parameter combination. Next, the system uses the selected parameters for small-batch production verification and collects detailed process data. Finally, the system comprehensively analyzes the verification results, evaluates the optimization effect, and generates an evaluation report including economic benefit analysis.

[0126] In some embodiments, process parameter optimization and verification can be achieved in various ways: Optionally, it can be achieved through the following steps: formulating an optimization strategy plan; generating a set of parameter combinations; performing simulation verification; evaluating parameter effects; conducting production verification; analyzing optimization benefits; generating an evaluation report; and updating the parameter database. Optionally, it can be achieved through the following steps: constructing an optimization objective function; designing a parameter combination plan; establishing a simulation model; running a simulation test; selecting the optimal parameters; performing verification production; calculating benefit indicators; and forming an optimization report. It can be understood that other methods can also be used to achieve process parameter optimization and verification, such as through intelligent algorithm optimization, expert system assistance, etc., which are not limited here.

[0127] In the embodiments of the present application, due to the adoption of a comprehensive control scheme based on deep learning for product feature difference analysis, intelligent optimization of process parameters, and real-time status monitoring, combined with an environmental impact compensation and anomaly warning mechanism, it is possible to achieve rapid adaptation and dynamic optimization of process parameters, effectively solving the problems of low parameter adjustment efficiency and poor adaptability in the prior art, and thus realizing the intelligent control of mixed-line production of liquid crystal display panels.

[0128] The control system in the embodiments of the present invention application will be described from the perspective of hardware processing. Please refer to Figure 3 , which is a schematic structural diagram of an entity device of the control system in the embodiments of the present application.

[0129] It should be noted that Figure 3 The structure of the control system shown is only an example and should not impose any limitations on the functions and usage scope of the embodiments of the present invention.

[0130] As Figure 3As shown, the control system includes a Central Processing Unit (CPU) 301, which can perform various appropriate actions and processes according to the program stored in the Read-Only Memory (ROM) 302 or the program loaded from the storage section 308 into the Random Access Memory (RAM) 303, such as executing the method described in the above embodiments. In the RAM 303, various programs and data required for system operation are also stored. The CPU 301, ROM 302, and RAM 303 are connected to each other via a bus 304. An Input / Output (I / O) interface 305 is also connected to the bus 304.

[0131] The following components are connected to the I / O interface 305: an input section 306 including an audio input device, a button switch, etc.; an output section 307 including a Liquid Crystal Display (LCD), an audio output device, an indicator light, etc.; a storage section 308 including a hard disk, etc.; and a communication section 309 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. A removable medium 311, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 310 as needed so that a computer program read from it can be installed into the storage section 308 as needed.

[0132] Specifically, according to an embodiment of the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present invention includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network via the communication section 309, and / or installed from the removable medium 311. When the computer program is executed by the Central Processing Unit (CPU) 301, various functions defined in the present invention are executed.

[0133] It should be noted that specific examples of computer-readable storage media may include, but are not limited to: electrical connections with one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fibers, portable compact disc read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above. In the present invention, a computer-readable storage medium can be any tangible medium that contains or stores a program, and this program can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0134] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present invention. Among them, each block in the flowchart or block diagram may represent a module, a program segment, or a part of code, and the above-mentioned module, program segment, or part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than that marked in the accompanying drawings.

[0135] Specifically, the control system of this embodiment includes a processor and a memory. A computer program is stored on the memory. When the computer program is executed by the processor, it implements the production control method of the liquid crystal display screen provided in the above-mentioned embodiment.

[0136] On the other hand, the present invention also provides a computer-readable storage medium, which may be included in the control system described in the above-mentioned embodiment; or it may exist alone without being assembled into the control system. The above storage medium carries one or more computer programs. When the above one or more computer programs are executed by a processor of the control system, the control system implements the production control method of the liquid crystal display screen provided in the above-mentioned embodiment.

[0137] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the various embodiments of the present application.

[0138] As used in the foregoing embodiments, depending on the context, the term "when" may be construed to mean "if" or "after" or "in response to determining" or "in response to detecting". Similarly, depending on the context, the phrase "when determining" or "if (the stated condition or event) is detected" may be construed to mean "if determining" or "in response to determining" or "when (the stated condition or event) is detected" or "in response to detecting (the stated condition or event)".

[0139] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the foregoing embodiments can be implemented. The processes can be completed by relevant hardware instructed by a computer program, and the program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the foregoing method embodiments. The foregoing storage media include: various media such as ROM or random access memory RAM, magnetic disks, or optical discs that can store program codes.

Claims

1. A production control method for a liquid crystal display screen, characterized in that: Applied to a control system, the method comprises: Obtaining display unit parameters of the current production task; the display unit parameters include effective display area size, electrode spacing value and product identification number; Extracting the reference product parameters and reference process parameters corresponding to the product identification number from the process formula library; the reference process parameters include the liquid crystal dispensing amount, dispensing position and lamination pressure; Correcting the reference process parameters to initial process parameters according to the difference coefficient between the display unit parameters and the reference product parameters; In the process of executing the liquid crystal dripping process with the initial process parameters, real-time collection of liquid crystal spreading state data is performed, and a process stability index is generated according to the liquid crystal spreading state data; the liquid crystal spreading state data includes spreading area ratio and bubble distribution information; When the process stability index deviates from the preset stability range, parameter correction is performed based on the process parameter compensation model to obtain process correction parameters and corresponding product characteristics; The liquid crystal dripping process is executed with the process correction parameters, and the process correction parameters and the product characteristics are stored in the process recipe library.

2. The method according to claim 1, characterized in that: The step of correcting the reference process parameters to initial process parameters according to the difference coefficient between the display unit parameters and the reference product parameters specifically includes: Determining a size difference coefficient, a spacing difference coefficient, and a structure difference coefficient based on the display unit parameters and the reference product parameters; Calculating a liquid crystal usage correction coefficient, a dispensing position offset, and a pressure compensation value according to a weighted combination value of the size difference coefficient, the spacing difference coefficient, and the structure difference coefficient; Multiplying the liquid crystal dispensing amount in the reference process parameters by the liquid crystal usage correction coefficient to obtain an initial dispensing amount; Performing coordinate transformation on the dispensing position in the reference process parameters based on the dispensing position offset to obtain an initial dispensing position; Superimposing the bonding pressure in the reference process parameter and the pressure compensation value to obtain an initial bonding pressure; The initial dispensing amount, the initial dispensing position and the initial bonding pressure are integrated as initial process parameters.

3. The method according to claim 1, characterized in that The step of collecting liquid crystal spreading state data in real time during the process of executing the liquid crystal dripping process with the initial process parameters, and generating a process stability index according to the liquid crystal spreading state data specifically includes: Collecting an image sequence of the liquid crystal spreading process in real time, and performing image segmentation and feature extraction on the image sequence to obtain a dynamic spreading profile of the liquid crystal material; Determine the spreading area change rate and the bubble number change rate per unit time according to the dynamic spreading profile; A process stability index is determined according to the spreading area change rate and the bubble number change rate.

4. The method according to claim 1, characterized in that: Before the step of obtaining the display unit parameters of the current production task, the method further includes: Clean and annotate historical production data and build a training sample set; Performing model training based on the training sample set to obtain an initial model of process parameters; Collect environmental parameters of production equipment including temperature, humidity and air pressure, and establish an environmental mapping relationship between the environmental parameters and process parameters; The environmental compensation is performed on the process parameter initial model based on the environmental mapping relationship to obtain a process parameter compensation model.

5. The method according to claim 4, characterized in that The steps of cleaning and labeling the historical production data and constructing a training sample set specifically include: Determine the dispensing history data of the nozzle pressure, dispensing speed and liquid crystal material temperature of the liquid crystal dispensing equipment; Determine the bonding history of the pressure curve, bonding speed and vacuum degree of the LCD bonding equipment; Determine the quality inspection data including alignment accuracy, bad pixel defects and foreign matter contamination; Correlate and match the dispensing history data and the laminating history data with the quality inspection data based on the timestamp to obtain history correlation data; The historical associated data are classified and labeled according to product models and defect types to obtain a training sample set including process parameter sequences and quality ratings.

6. The method according to claim 1, characterized in that After the step of executing the liquid crystal dripping process with the process correction parameters and storing the process correction parameters and the product characteristics in the process recipe library, the method further includes: Determining a predicted product yield rate according to the process correction parameters; Real-time collection of production status data including dispensing uniformity, bubble distribution, spreading rate and mixing line switching frequency; Determining the process abnormality risk probability based on the production status data; When the predicted yield of the product is lower than a preset yield threshold, or when the process abnormality risk probability is greater than a preset probability threshold, production warning information is generated.

7. The method according to claim 6, characterized in that When the predicted product yield is lower than a preset yield threshold, or when the process abnormality risk probability is greater than a preset probability threshold, after the step of generating production warning information, the method further includes: Extracting parameter deviation trends and quality impact assessments from the production warning information to generate process parameter optimization strategies; Determine multiple groups of candidate process parameter combinations in the process parameter optimization strategy, and perform simulation verification on each group of candidate process parameters to obtain corresponding simulation results; the simulation results include liquid crystal spreading uniformity, bubble residual rate and expected product yield; Performing simulation scoring on the simulation results, and selecting the candidate process parameters with the highest simulation scores as the optimized process parameters after optimization; The optimized process parameters are used to perform production verification, and process data including liquid crystal injection amount, spreading speed, bubble size and distribution position are collected in real time; Based on the process data, the utilization rate of liquid crystal materials, the product qualification rate and the equipment utilization rate are calculated, and an optimization evaluation report including a cost-benefit analysis is generated.

8. A control system, characterized in that: The control system includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the control system to execute the method described in any one of claims 1-7.

9. A computer-readable storage medium comprising instructions, characterized in that: When the instructions are executed on a control system, the control system is caused to execute the method according to any one of claims 1 to 7.

10. A computer program product, characterized in that When the computer program product is run on a control system, the control system is caused to execute the method according to any one of claims 1 to 7.

Images