Display panel main spacer quality prediction method, terminal device and storage medium

Abstract

The application relates to the technical field of virtual metrology, and particularly provides a quality prediction method for a display panel main spacer, aiming to solve the problems of detection lag and high scrappage risk existing in the current first-piece inspection process. The method comprises the following steps: receiving a first prediction request carrying process parameters sent by a client; obtaining first metrology data; obtaining a first fusion feature tensor according to the process parameters and the first metrology data; obtaining a main spacer height estimated value and a step difference estimated value according to the first fusion feature tensor; generating a quality prediction result of a display panel main spacer to be predicted according to the height estimated value and the step difference estimated value; and returning the quality prediction result to the client. According to the application, the historical working condition data of the preset process parameters and the metrology equipment are fused, the quality prediction is completed before the physical detection of the first-piece product, the zero-waiting prediction of the first-piece quality is realized, the batch scrappage caused by the detection lag is effectively avoided, and the prediction accuracy and the production efficiency are significantly improved.

Description

Technical Field

[0001] This application relates to the field of virtual measurement technology, specifically providing a method for predicting the quality of the main spacer of a display panel, a terminal device, and a storage medium. Background Technology

[0002] In display panel manufacturing, the height (Main PSHeight, MPSH) and step (MS) of the main photo spacer (Main PS) are the core parameters that determine the optical and mechanical properties of the panel, and its first-piece inspection is a key step in starting the photolithography process.

[0003] However, the current first article inspection process has the following technical defects: First, the existing method requires the glass substrate to be sent to an offline film thickness gauge for measurement after the glass substrate has completed the complete photolithography process. This process takes several hours, during which the entire batch of glass substrates is in a waiting state, and the production line cannot continue to move forward, directly causing a loss of production capacity; Second, once the first MPSH or MS fails, the entire batch of products faces the risk of being scrapped, causing huge economic losses. Summary of the Invention

[0004] This application aims to solve the aforementioned technical problems, namely, to address the issues of delayed testing and high risk of scrapping in the existing first article inspection process.

[0005] In a first aspect, this application provides a method for predicting the quality of a display panel's main spacer, applied on a server side, comprising:

[0006] Receive a first prediction request sent by the client, and extract the process parameters carried in the first prediction request, wherein the process parameters include preset values ​​for photoresist coating thickness, exposure machine dosage, developer temperature, and baking temperature.

[0007] Acquire first measurement data of the measuring device within a preset time period, wherein the first measurement data includes time series data of the rotation speed of the coating unit and acceleration change curve, attenuation curve of the output energy of the exposure lamp, concentration value and temperature value of the developer and temperature distribution data of the baking unit.

[0008] The process parameters and the first measurement data are fused to obtain a first fused feature tensor.

[0009] The first fused feature tensor is sent to the algorithm service module, and the algorithm service module returns the estimated value of the main septum height and the estimated value of the step difference; wherein, the step difference is the height difference between the main septum region and the non-main septum region;

[0010] Based on the estimated height of the main spacer and the estimated step difference, a quality prediction result for the main spacer of the display panel to be predicted is generated, and the quality prediction result is returned to the client.

[0011] Optionally, acquiring the first measurement data of the measuring device within a preset time period includes:

[0012] A data acquisition command is sent to the measuring device through the first communication interface;

[0013] Receive the first measurement data of the measuring device within a preset time period returned by the measuring device through the first communication interface;

[0014] The measuring equipment includes a coating machine, an exposure machine, a developing machine, and a baking machine.

[0015] Optionally, the step of fusing the process parameters and the first measurement data to obtain a first fused feature tensor includes:

[0016] Send a data processing request to the database engine, the data processing request carrying the first measurement data;

[0017] Receive ordered feature data returned by the database engine; wherein, the ordered feature data is a fixed-length ordered array generated by the database engine after extracting and aligning features from the first measurement data according to pre-configured feature configuration data;

[0018] The ordered feature data and the process parameters are concatenated to obtain the first fused feature tensor.

[0019] Optionally, sending the first fused feature tensor to the algorithm service module and receiving the estimated height of the main septum and the estimated step difference returned by the algorithm service module includes:

[0020] The first fused feature tensor is encapsulated into a first inference request, and the first inference request is sent to the algorithm service module; the first inference request includes the first fused feature tensor and product information of the main spacer of the display panel to be predicted;

[0021] Receive the estimated height of the main septum and the estimated step difference returned by the algorithm service module.

[0022] Optionally, the step of generating a quality prediction result for the main spacer of the display panel to be predicted based on the estimated height of the main spacer and the estimated step difference, and sending the quality prediction result to the client, includes:

[0023] Obtain a first preset threshold, a second preset threshold, a third preset threshold, and a fourth preset threshold, wherein the first preset threshold is the upper limit of the height specification of the main spacer of the display panel to be predicted, the second preset threshold is the lower limit of the height specification of the main spacer of the display panel to be predicted, the third preset threshold is the upper limit of the step specification of the main spacer of the display panel to be predicted, and the fourth preset threshold is the lower limit of the step specification of the main spacer of the display panel to be predicted.

[0024] If the estimated height of the main septum is greater than or equal to the second preset threshold and less than or equal to the first preset threshold, and the estimated step difference is greater than or equal to the fourth preset threshold and less than or equal to the third preset threshold, then a first quality prediction result is generated and sent to the client, wherein the first quality prediction result indicates that the septum is of acceptable quality.

[0025] If the estimated height of the main septum is greater than the first preset threshold, or the estimated height of the main septum is less than the second preset threshold, or the estimated step difference is greater than the third preset threshold, or the estimated step difference is less than the fourth preset threshold, then a second quality prediction result is generated, wherein the second quality prediction result indicates that the septum is of substandard quality.

[0026] Optionally, after generating the second quality prediction result, the method further includes:

[0027] The estimated height of the main diaphragm and the estimated step difference are input into the preset analysis module to obtain the process adjustment parameters output by the preset analysis module, and the process adjustment parameters are sent to the client.

[0028] Optionally, the method further includes:

[0029] Receive a second prediction request sent by the client, wherein the second prediction request carries the corrected process parameters;

[0030] The modified process parameters and the first measurement data are fused to obtain a second fused feature tensor.

[0031] The second fused feature tensor is sent to the algorithm service module, and the corrected main septum height estimate and the corrected step difference estimate are received from the algorithm service module.

[0032] Based on the corrected estimated height of the main septum and the corrected estimated step difference, a third quality prediction result is generated and returned to the client.

[0033] In a second aspect, this application provides a method for predicting the quality of the main spacer of a display panel, applied to a client, comprising:

[0034] The process parameters input by the user are obtained, and a first prediction request is generated based on the process parameters, wherein the first prediction request carries the process parameters input by the user; the process parameters include preset values ​​for photoresist coating thickness, exposure machine dosage, developer temperature, and baking temperature.

[0035] Send the first prediction request to the server;

[0036] Receive the quality prediction results returned by the server.

[0037] Optionally, receiving the quality prediction result returned by the server includes:

[0038] Receive a first quality prediction result returned by the server, wherein the first quality prediction result indicates that the spacer material is of acceptable quality; or...

[0039] The system receives a second quality prediction result and process adjustment parameters returned by the server. The second quality prediction result indicates that the septum material is of substandard quality. The process adjustment parameters are obtained by the server after generating the second quality prediction result by inputting the estimated height of the main septum material and the estimated step difference into a preset analysis module.

[0040] Optionally, the method further includes:

[0041] If the process adjustment parameters are received, a second prediction request is sent to the server; wherein the second prediction request carries the corrected process parameters;

[0042] The server receives a third quality prediction result, which is generated by the server based on the corrected main septum height estimate and the corrected step difference estimate. The corrected main septum height estimate and the corrected step difference estimate are obtained by the server performing feature fusion of the corrected process parameters and the first measurement data to obtain a second fused feature tensor, and then sending the second fused feature tensor to the algorithm service module for inference.

[0043] In a third aspect, this application provides a terminal device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method as described in any of the first aspects of this application; or, the processor executes the computer program to implement the method as described in any of the second aspects of this application.

[0044] In a fourth aspect, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method as described in any of the first aspects of this application; or, when executed by a processor, the computer program implements the method as described in any of the second aspects of this application.

[0045] With the above technical solution, the quality prediction method for the main spacer of the display panel provided in this application obtains the process parameters input by the user through the client before the start of the photolithography process, and sends a first prediction request carrying the process parameters input by the user to the server. The server performs feature fusion between the process parameters input by the user and the historical operating condition data of the measurement equipment within a preset time period to obtain a first fused feature tensor, and inputs the first fused feature tensor into the algorithm service module for inference to obtain the quality prediction result of the main spacer of the display panel to be predicted, and displays it on the client. This application completes the prediction of the height and step difference of the main spacer before the start of the photolithography process, without waiting for the first product to complete the complete photolithography process and be sent to the offline measurement equipment for actual measurement. Therefore, this application eliminates the waiting time of several hours in the traditional first article inspection, the entire batch of glass substrates does not need to stop, the production line can continue to advance, and the production capacity of the photolithography process is effectively improved. When the prediction results show that the height or step difference of the main spacer is unqualified, the process parameters can be proactively adjusted before the lithography machine is put into formal production, avoiding the scrapping of the entire batch due to the first piece being unqualified in the traditional way, and significantly reducing manufacturing costs. Attached Figure Description

[0046] The preferred embodiments of this application are described below with reference to the accompanying drawings, in which:

[0047] Figure 1 This is a schematic diagram of the structure of the display panel main spacer quality prediction system provided in the embodiments of this application;

[0048] Figure 2 This is a flowchart of the method for predicting the quality of the main spacer material of the display panel provided in the embodiments of this application. Figure 1 ;

[0049] Figure 3 This is a flowchart of the method for predicting the quality of the main spacer material of the display panel provided in the embodiments of this application. Figure 2 ;

[0050] Figure 4 This is a flowchart of the method for predicting the quality of the main spacer material of the display panel provided in the embodiments of this application. Figure 3 ;

[0051] Figure 5 This is a flowchart of the method for predicting the quality of the main spacer material of the display panel provided in the embodiments of this application. Figure 4 ;

[0052] Figure 6 This is a flowchart of the method for predicting the quality of the main spacer material of the display panel provided in the embodiments of this application. Figure 5 ;

[0053] Figure 7 This is a flowchart of the method for predicting the quality of the main spacer material of the display panel provided in the embodiments of this application. Figure 6 ;

[0054] Figure 8 This is a schematic diagram of the structure of the terminal device provided in the embodiments of this application. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0056] Unless otherwise defined, the technical or scientific terms used in this application shall have the ordinary meaning understood by one of ordinary skill in the art to which this application pertains. The terms "first," "second," and similar terms used in this application do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Similarly, the terms "an," "a," or "the," and similar terms do not indicate a quantity limitation, but rather indicate the presence of at least one. The terms "comprising," "including," or similar terms mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects.

[0057] In this application, "multiple or several" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. The character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0058] In display panel manufacturing, the main spacer (Main PS) is a columnar structure formed by the exposure and development of photosensitive resin coated on a glass substrate (usually a color filter substrate). Its height (MPSH) directly determines the thickness of the liquid crystal layer (Cell Gap), while the step difference, i.e., the height difference between the main spacer area and the non-main spacer area, affects the initial alignment of liquid crystal molecules and the compressive strength of the display panel. Therefore, the height (MPSH) and step difference (MS) have a decisive impact on the performance of the display panel.

[0059] However, the current first-piece inspection process has the following technical defects: First, the existing method requires the glass substrate to be sent to an offline film thickness gauge for measurement after the glass substrate has completed the complete photolithography process. This process takes several hours, during which the entire batch of glass substrates is in a waiting state, and the production line cannot continue to move forward, directly causing a loss of production capacity; Second, if the height or step difference of the first main spacer is not up to standard, the entire batch of products will face the risk of being scrapped, causing huge economic losses.

[0060] Based on the aforementioned technical problems, the quality prediction method for the main spacer of the display panel provided in this application obtains the process parameters input by the user through the client before the start of the photolithography process, and sends a first prediction request carrying the user-input process parameters to the server. The server performs feature fusion between the user-input process parameters and the historical operating condition data of the measurement equipment within a preset time period to obtain a first fused feature tensor. The first fused feature tensor is then input into the algorithm service module for inference to obtain the quality prediction result of the main spacer of the display panel to be predicted, and is visualized on the client. This application completes the prediction of the height and step difference of the main spacer before the start of the photolithography process, without waiting for the first product to complete the entire photolithography process and be sent to the offline measurement equipment for actual measurement. Therefore, this application eliminates the waiting time of several hours in the traditional first article inspection, the entire batch of glass substrates does not need to stop, the production line can continue to advance, and the production capacity of the photolithography process is effectively improved. When the prediction results show that the height or step difference of the main spacer is unqualified, the process parameters can be proactively adjusted before the lithography machine is put into formal production, avoiding the scrapping of the entire batch due to the first piece being unqualified in the traditional way, and significantly reducing manufacturing costs.

[0061] See Figure 1 The diagram shows a quality prediction system for a display panel main spacer provided in an embodiment of this application. The system may include: a client 100, a server 200, and multiple measuring devices 300.

[0062] The client 100 serves as a front-end interactive terminal for engineers, providing a human-machine interface for industrial control computers, laptops, tablets, or lithography machines within the control room of the display panel manufacturing production line. Engineers input the process parameters for the current production batch onto the client 100's visual interface. These parameters include preset values ​​for photoresist coating thickness, exposure machine dosage, developer temperature, and baking temperature. After inputting the process parameters, the engineer clicks the "One-Click Prediction" button on the client 100's visual interface. The client 100 then encapsulates these process parameters into a first prediction request and sends it to the server 200 via the factory's internal Ethernet. For example, communication between the client 100 and the server 200 is based on either HTTP or MQTT protocols.

[0063] The server-side unit 200 can be deployed locally, at the edge, or in the cloud. For example, when deployed locally, the server-side unit 200 is a dedicated virtual measurement server deployed in the production line's local server room; when deployed at the edge, the server-side unit 200 is an industrial edge computing gateway deployed near the lithography machine; when deployed in the cloud, the server-side unit 200 is a virtual machine instance on a private or hybrid cloud platform. These multiple deployment methods allow the system to flexibly adapt to display panel manufacturing production lines of different sizes and architectures. The specific deployment method can be selected based on the production line's real-time requirements or data security policies.

[0064] The server 200 is used to receive the first prediction request sent by the client 100. After receiving the first prediction request, it parses and extracts the process parameters in it, and calls the interface module to send a data acquisition instruction to the vector measurement device 300.

[0065] The interface module is a software functional unit deployed on the server 200. It provides a unified communication entry point, adapts to at least one industrial communication protocol, and enables data interaction between the server 200 and the measurement device 300. Industrial communication protocols include, but are not limited to, SECS / GEM, Modbus / TCP, OPC UA, or MQTT protocols. Specifically, in response to the data acquisition command from the server 200, the interface module selects the corresponding protocol driver based on the protocol type of each measurement device, sends data read requests to each measurement device 300, receives the first measurement data returned by each measurement device 300, parses the first measurement data into a unified format, and returns it to the server 200.

[0066] Optionally, the interface module also includes a data caching submodule for temporarily storing the first measurement data obtained from each measurement device 300. When different measurement devices 300 are distributed in different network segments or have uneven response speeds, the data caching submodule will aggregate the raw data received from multiple measurement devices. After obtaining all the measurement data within a preset time period, it will uniformly perform protocol parsing and format conversion, and send the integrated data to the server 200 to reduce the resource consumption of the server 200 caused by frequent small data packet transmissions.

[0067] The measurement equipment 300 includes a coating machine, an exposure machine, a developing machine, and a baking machine. Each measurement equipment 300 is equipped with a data acquisition and storage module, which is used to collect equipment operating parameters in real time and temporarily store them in a local cache or embedded database.

[0068] The measurement device 300 communicates with the server 200 via an industrial Ethernet network within the factory. In one embodiment of this invention, after receiving a data acquisition instruction from the server 200, the measurement device 300 acquires the first measurement data collected and stored by the measurement device 300 within a preset time period, encapsulates the first measurement data into a response message according to a preset data format, and returns it to the server 200 via the SECS / GEM protocol or the Modbus / TCP protocol.

[0069] In another implementation, the measurement device 300 continuously collects first measurement data and reports it to a time-series database or message queue middleware in real time. Upon receiving the first prediction request from the client 100, the server 200 directly reads the first measurement data within a preset time period from the time-series database or message queue middleware to reduce communication latency and equipment load. The above two implementations can be configured according to the production line network conditions and equipment capabilities; this application embodiment does not limit this configuration.

[0070] After acquiring the process parameters and the first measurement data, the server 200 performs feature fusion on the process parameters and the first measurement data to obtain a first fused feature tensor. The first fused feature tensor is sent to the algorithm service module, and the server receives the estimated height and step difference of the main spacer returned by the algorithm service module. Based on the estimated height and step difference of the main spacer, the server generates a quality prediction result for the main spacer of the display panel to be predicted, and returns the quality prediction result to the client 100.

[0071] The following will combine Figure 2 The method for predicting the quality of the main spacer material in a display panel provided in this application embodiment will be described in detail. For example... Figure 2 As shown, and not as a limitation, the method includes steps S10 to S17:

[0072] Step S10: The client obtains the process parameters input by the user and generates a first prediction request based on the process parameters.

[0073] In this embodiment, the client is a front-end interactive terminal for engineers, including an industrial control computer, portable computer, tablet computer, or the human-machine interface built into the lithography machine in the control room of the display panel manufacturing line. Engineers input the process parameters of the main spacer material of the display panel to be predicted on the client's visual interface. These process parameters include preset values ​​for photoresist coating thickness, exposure machine dosage, developer temperature, and baking temperature.

[0074] For example, before an engineer initiates a new batch of photolithography processes for a display panel, the engineer inputs the process parameters for that display panel on the client interface. For instance, the engineer sets the preset values ​​for the photoresist coating thickness to 3.5 μm, the exposure machine dose to 150 mJ / cm², the developer temperature to 23.0 ℃, and the baking temperature to 110 ℃. After inputting the process parameters, the engineer clicks the "One-Click Prediction" button on the client interface. The client then retrieves the process parameters entered by the engineer and encapsulates these parameters into a first prediction request.

[0075] Step S11: The client sends a first prediction request to the server, and the server receives the first prediction request accordingly.

[0076] In this embodiment, the server-side refers to computing devices deployed within the display panel manufacturing production line. Deployment methods include: local deployment, i.e., a dedicated virtual metrology server deployed in the production line's local server room; edge deployment, i.e., an industrial edge computing gateway with sufficient computing power deployed near the lithography machine; and cloud deployment, i.e., a virtual machine instance deployed on a private or hybrid cloud platform. These three deployment methods can be flexibly selected based on the production line's real-time requirements and data security strategies.

[0077] After generating a first prediction request, the client sends it to the server. Correspondingly, the server receives the first prediction request from the client. For example, the client and server can communicate using either the HTTP protocol or the MQTT protocol.

[0078] Step S12: The server extracts the process parameters carried in the first prediction request.

[0079] In one implementation, after receiving the first prediction request, the server first performs a validity check on the request. For example, the validity check includes verifying whether each process parameter is within a preset reasonable value range. For instance, verifying whether the preset value for photoresist coating thickness is between 2.0 μm and 5.0 μm, the preset value for exposure machine dosage is between 100 mJ / cm² and 200 mJ / cm², the preset value for developer temperature is between 21.0 ℃ and 25.0 ℃, and the preset value for baking temperature is between 100 ℃ and 130 ℃. If the check passes, the server parses the first prediction request, extracts the process parameters, product identifier, and request timestamp, and temporarily stores these in the server's cache. If the check fails, the server returns a parameter error message to the client to guide the engineer to re-enter process parameters that meet the validity check.

[0080] Step S13: The server obtains the first measurement data of the measurement device within a preset time period.

[0081] The measuring equipment includes a coating machine, an exposure machine, a developing machine, and a baking machine. The first set of measuring data includes time-series data of the rotation speed of the coating unit and acceleration variation curves, attenuation curves of the output energy of the exposure lamp, concentration and temperature values ​​of the developing solution, and temperature distribution data of the baking unit.

[0082] The preset time period is a continuous time window preceding the current moment, and its duration is preset according to the process characteristics, such as 30 minutes preceding the current moment. It should be noted that the preset time period can also be set independently for different types of measurement data. For example, a longer time window can be used for developer concentration data to reflect the concentration drift trend. This application embodiment does not limit this.

[0083] In this embodiment, the server acquires the first measurement data, including several optional implementation methods. In one implementation, in response to the first prediction request, the server sends a data acquisition instruction to the data acquisition system of each measurement device through an interface module. Each measurement device reads historical measurement data within a corresponding preset time period from its local cache or embedded database according to the data acquisition instruction and returns it to the server. In another implementation, each measurement device continuously reports real-time acquired sensor data to a time-series database or message queue middleware at a fixed frequency. The server reads the measurement data within the most recent preset time period from the time-series database or message queue middleware through a subscription method.

[0084] The above two methods are merely illustrative examples, and one can be selected for configuration according to the actual situation. This application embodiment does not limit the specific implementation method of the server side obtaining the first measurement data.

[0085] For example, after receiving the first prediction request, the server adopts any of the above implementation methods to obtain the first measurement data collected and stored by each measurement device within 30 minutes before the current time from the coating machine, exposure machine, developing machine and baking machine.

[0086] Step S14: The server performs feature fusion on the process parameters and the first measurement data to obtain the first fused feature tensor.

[0087] Specifically, after the server extracts the process parameters and obtains the first measurement data, it performs time alignment and spatial alignment on the first measurement data, and then concatenates the time-space aligned first measurement data and the process parameters along the feature dimension to obtain the first fused feature tensor.

[0088] Step S15: The server sends the first fused feature tensor to the algorithm service module and receives the estimated value of the main septum height and the estimated value of the step difference returned by the algorithm service module.

[0089] The step difference is the height difference between the main septum area and the non-main septum area.

[0090] In this embodiment, the algorithm service module can be deployed on the server side (e.g., Figure 2 (As shown), it can also be deployed independently on another computing device or cloud server (such as...). Figure 3 (As shown). The algorithm service module internally deploys a pre-trained virtual measurement model, which is a regression model that takes the first fusion feature tensor as input and the estimated height of the main septum and the estimated step difference as output. For example, the virtual measurement model is a regression model constructed based on algorithms such as deep neural networks, random forests, gradient boosting decision trees, or support vector regression.

[0091] After receiving the first fused feature tensor, the algorithm service module calls the virtual measurement model, inputs the first fused feature tensor into the virtual measurement model for inference calculation, and the virtual measurement model outputs the estimated value of the main septum height and the estimated value of the step difference, and returns the estimated value of the main septum height and the estimated value of the step difference to the server through the algorithm service module.

[0092] In this embodiment, to avoid blocking the server's backend main thread during long-duration deep learning inference, the algorithm service module uses an asynchronous callback mechanism to return the inference results. Specifically, after receiving the first fused feature tensor and triggering the inference calculation of the virtual measurement model, the algorithm service module immediately releases the server's backend main thread resources without synchronously waiting. After the inference calculation is completed, the algorithm service module asynchronously pushes the inference result message, which includes the estimated value of the main septum height, the estimated value of the segment difference, and the inference timestamp, to the server through a pre-registered callback interface. For example, the callback interface can be implemented using a message queue (e.g., Kafka) or an HTTP Webhook. Through the above asynchronous callback mechanism, the server's backend main thread does not need to block and wait for the inference calculation to complete before continuing to process other requests, thereby significantly improving the system throughput and operational stability in high-concurrency prediction request scenarios.

[0093] In a preferred embodiment, the virtual measurement model is trained offline based on historical batch data. The training data includes historical process parameters, historical equipment measurement data, and corresponding measured main septum height and measured step difference values. The historical process parameters and historical equipment measurement data originate from process parameters input by the client and operating condition data collected by the measurement equipment in historical production batches. The measured main septum height and measured step difference values ​​originate from offline physical measurements of the first piece of the historical batch. The training objective is to minimize the mean square error between the predicted and measured values. After model training, it is deployed to the algorithm service module.

[0094] In a specific example, such as Figure 3 As shown, when the algorithm service module is independently deployed on server 2, after server 1 obtains the first fusion feature tensor, it sends the first fusion feature tensor to server 2. The algorithm service module on server 2 receives the first fusion feature tensor and calls the virtual measurement model to perform inference on the virtual measurement model. The virtual measurement model outputs the estimated value of the main septum height and the estimated value of the step difference, and returns the estimated value of the main septum height and the estimated value of the step difference to server 1 through server 2.

[0095] Step S16: The server generates a quality prediction result for the main spacer of the display panel based on the estimated height and step difference of the main spacer.

[0096] In this embodiment, the server generates a quality prediction result based on the estimated height of the main septum and the estimated step difference returned by the algorithm service module, combined with preset quality judgment rules.

[0097] For example, the preset quality judgment rule is as follows: when the estimated height of the main septum is within the first preset qualified range (e.g., 2.5μm to 3.0μm) and the estimated step difference is within the second preset qualified range (e.g., 0.3 to 0.6μm), the quality prediction result of the main septum is judged to be qualified; otherwise, the quality prediction result of the main septum is judged to be unqualified.

[0098] It should be noted that the first and second preset qualified ranges mentioned above are only illustrative examples, and the specific values ​​can be adjusted according to the process specifications of different products.

[0099] Step S17: The server returns the quality prediction results to the client.

[0100] The server encapsulates the quality prediction results into a response message and returns it to the client via the factory's internal Ethernet. Upon receiving the quality prediction results, the client displays them on a visual interface for engineers to use in decision-making.

[0101] In one alternative implementation, such as Figure 4 As shown, step S13 further includes steps S131 and S132:

[0102] Step S131: The server sends a data acquisition command to the vector measurement device through the first communication interface.

[0103] The first communication interface serves as a unified communication entry point for the interface module deployed on the server side, enabling data interaction between the server and the measurement equipment. The data acquisition command carries a preset time period parameter, which instructs the measurement equipment to return the first measurement data collected and stored within that time period.

[0104] For example, data acquisition commands are sent to various measurement devices via the SECS / GEM protocol or the Modbus / TCP protocol. Specifically, the data acquisition commands include: sending a command to the coating machine to acquire the time series data of the coating unit's rotation speed and acceleration change curve; sending a command to the exposure machine to acquire the output energy decay curve of the exposure lamp; sending a command to the developing machine to acquire the developer concentration and temperature values; and sending a command to the baking machine to acquire the temperature distribution data of the baking unit.

[0105] In one implementation, the data acquisition instruction also carries device identification information to specify the target measurement device for which data needs to be acquired; when data from multiple measurement devices needs to be acquired, the server sends the corresponding data acquisition instruction to each measurement device through the first communication interface.

[0106] Step S132: The server receives the first measurement data of the measurement device within a preset time period, returned by the measurement device through the first communication interface.

[0107] Upon receiving a data acquisition instruction, each measurement device reads the first measurement data stored within the corresponding preset time period from its local cache or embedded database according to the time parameter carried in the instruction, encapsulates it into a response message according to a preset data format, and returns it to the server through the first communication interface.

[0108] For example, the first set of measurement data includes: time-series data of coating unit rotation speed and acceleration variation curve returned by the coating machine; data of exposure lamp output energy decay curve returned by the exposure machine; developer concentration and temperature values ​​returned by the developing machine; and temperature distribution data of the baking unit returned by the baking machine. All of the above measurement data are marked with precise timestamps.

[0109] Optionally, if a measurement device fails to return the first measurement data within a preset time, the first communication interface will retry according to a preset retry strategy; if the data still cannot be obtained after the retry, an exception message indicating that the data acquisition failed will be returned to the server, and the server will return the exception message to the client, prompting the engineer to check the communication status of the corresponding measurement device.

[0110] In another implementation, when the measurement equipment is equipped with a real-time data reporting function, each measurement equipment reports the first measurement data it collects to a time-series database or message queue middleware in real time. After receiving the first prediction request from the client, the server directly reads the first measurement data within a preset time period from the time-series database or message queue middleware to reduce communication latency and equipment load.

[0111] In one alternative implementation, such as Figure 5 As shown, step S14 further includes steps S141 to S143:

[0112] Step S141: The server sends a data processing request to the database engine, and the data processing request carries the first measurement data.

[0113] In this embodiment, the database engine is a distributed database that supports columnar storage and vectorized execution engines. Preferably, the database engine is the ClickHouse database engine. The database engine has a pre-built feature configuration storage area for performing configuration-driven feature extraction, alignment, and ordered array generation operations.

[0114] When the server sends a data processing request to the database engine, it also attaches a timestamp and equipment code as the first measurement data, storing it in the database engine's equipment condition fact table via batch insert or streaming write. The equipment condition fact table is partitioned by product identifier (e.g., Lot ID) and time primary key to support subsequent efficient time window queries. The data processing request also carries the product identifier and time window conditions, instructing the database engine to filter the original operating condition records corresponding to the current process from the equipment condition fact table.

[0115] Step S142: The database engine extracts and aligns the features of the first measurement data according to the pre-configured feature configuration data, generates ordered feature data, and returns the ordered feature data to the server.

[0116] Correspondingly, the server receives ordered feature data returned by the database engine.

[0117] In one possible implementation, step S142 further includes steps S1421 to S1424:

[0118] Step S1421: In response to the data processing request sent by the server, the database engine reads the feature configuration array from the feature configuration store.

[0119] Specifically, the feature configuration array is a list of strings, where each element is a complete parameter identifier to be included in the algorithm service. The format is process code##parameter name, for example, 'F300-00##AVG_TEMP' represents the average temperature of process F300-00, and 'F800-00##1_00_Shot01ExposurePower' represents the exposure power of the first exposure field in process F800-00. The feature configuration array can be dynamically added to or deleted from based on changes in product model or process version, without requiring modification to the server-side application code or recompilation or deployment of the application.

[0120] The database engine calls the first array expansion function and the array enumeration function to expand the feature configuration array into a feature configuration table. Specifically, the first array expansion function expands the array-type columns into multiple rows, and the array enumeration function assigns each expanded element its position index in the original array. After this processing, the feature configuration table contains at least a feature index field and a feature name field, where the feature index is a continuously increasing integer used to uniquely determine the order of each feature dimension in the subsequent output tensor.

[0121] This step expands the feature configuration array into a feature configuration table with integer indices, establishing a mapping relationship from business parameter identifiers to numerical indices. This provides an indexing basis for the sorting and aggregation of feature values ​​and the output of fixed-length arrays in subsequent steps, while also achieving complete decoupling of the feature selection logic from the application source code.

[0122] Step S1422: The database engine performs dynamic matching of the first measurement data with the feature configuration array.

[0123] The database engine filters out the original operating condition records corresponding to the current equipment process from the equipment operating condition fact table, based on the product identifier and time window conditions passed in from the server.

[0124] Since a single original record may contain multi-valued fields (such as multi-site measurement values ​​from a sensor matrix), the database engine calls the second array expansion function to break down the multi-valued fields in the same original record into a flat structure. Multi-valued fields include, but are not limited to, multiple parameter name fields, multiple site name fields, and corresponding multiple site value fields under the same process code (such as multi-site measurement values ​​from a sensor matrix). The second array expansion function breaks down multiple site value fields in the same row of records into multiple independent rows.

[0125] Subsequently, the database engine uses a string concatenation function to concatenate the process code and parameter name using the same delimiter to generate a candidate feature name string. The database engine then performs an inner join operation between the candidate feature name string and the feature name field in the feature configuration table, retaining only records whose feature names exist in the feature configuration array.

[0126] This step uses a feature configuration table as a whitelist filtering condition, and completes feature filtering and dimension reduction at the database storage layer. It extracts only the parameters specified in the configuration array, avoiding the server-side full transmission of massive amounts of data and cyclic matching, thereby effectively reducing network I / O overhead between the server and the database engine.

[0127] Step S1423: The database engine extracts the deduplicated sample identifiers of the matched records from step S1422, and performs a Cartesian product operation on the deduplicated sample identifiers and the feature configuration table to generate a complete sample-feature combination template. The database engine performs a left join operation on the sample-feature combination template and the feature values ​​actually extracted in step S1422. For feature dimensions that do not have corresponding values ​​in the equipment operating condition fact table, empty strings or preset default values ​​are automatically filled using a null value handling function.

[0128] This step uses a combination of Cartesian product and left connection to ensure that different samples have consistent feature vector dimensions, guaranteeing that the output feature vector has a fixed geometric dimension and preventing the algorithm from crashing due to input dimension mismatch.

[0129] Step S1424: The database engine performs feature aggregation and ordered vector generation.

[0130] The database engine groups and aggregates the output of step S1423 according to the sample identifier. Within each group, the database engine calls an array sorting function to sort the feature key-value pairs in ascending order based on the feature index, ensuring that the order of the output feature values ​​is consistent with the order of the elements in the feature configuration array.

[0131] Subsequently, the database engine calls the array aggregation function to encapsulate the sorted feature values ​​into an ordered feature value array, and then concatenates the device identifier with the ordered feature value array to output a fixed-length, completely ordered numerical array as ordered feature data, which is returned to the server through the database connection protocol.

[0132] In this embodiment, steps S1421 to S1424 are executed entirely within the database engine. The server only needs to send a data processing request to the database engine. Operations such as feature extraction, feature filtering, dimension alignment, missing value imputation, sorting aggregation, and ordered array generation are all automatically completed by the database engine based on pre-configured feature configuration data, without requiring any feature engineering operations in the application layer code on the server side.

[0133] Step S143: The server performs feature concatenation on the ordered feature data and process parameters to obtain the first fused feature tensor.

[0134] In this embodiment, firstly, the server expands the ordered feature data into a first two-dimensional feature matrix according to the sample dimension. The shape of this matrix is ​​[number of samples, number of features], where the number of columns corresponds to the total number of configured features. Then, the server copies and expands the process parameter vector obtained in step S101, making its number of rows the same as the number of rows in the first two-dimensional feature matrix, forming a second two-dimensional feature matrix. Next, the server uses the start and end timestamps in the timestamp information as time window labels and appends them to the corresponding dimensions of the first and second two-dimensional feature matrices, respectively. Finally, the server calls the tensor concatenation operator to concatenate the first and second two-dimensional feature matrices along the feature dimensions, generating a three-dimensional first fused feature tensor. Its data structure is represented as [number of samples, total number of fused features, time step], where the total number of fused features is the sum of the dimension of the ordered feature data and the dimension of the process parameters, and the time step is defined by the start and end timestamps of the timestamp information.

[0135] In a preferred embodiment, after completing feature fusion, the server also associates the first fused feature tensor with the corresponding batch identifier information and stores it in the database to support subsequent root cause analysis.

[0136] In one alternative implementation, such as Figure 6 As shown, step S15 further includes steps S151 and S152.

[0137] Step S151: The server encapsulates the first fused feature tensor into a first inference request and sends the first inference request to the algorithm service module.

[0138] In a specific example, the server packages the first fused feature tensor, which contains batch identification information derived from the first prediction request sent by the client, into a first inference request. This first inference request is then sent to the algorithm service module. This step aims to utilize the batch identification information as a callback credential to ensure that the estimated height of the primary septum and the estimated step difference returned by the algorithm service module can be accurately traced back to the corresponding display panel to be predicted.

[0139] For example, the first inference request is sent to the algorithm service module using the HTTP or gRPC protocol. After receiving the first inference request, the algorithm service module parses and extracts the first fusion feature tensor and batch identifier information, and inputs the first fusion feature tensor into the virtual measurement model for inference calculation.

[0140] Step S152: The server receives the estimated height of the main septum and the estimated step difference returned by the algorithm service module.

[0141] After the algorithm service module completes the model inference, it encapsulates the estimated height of the main septum and the estimated step difference into an inference response and returns it to the server via HTTP or gRPC.

[0142] Among them, the main spacer height prediction value is used to characterize the predicted height of the main spacer of the display panel to be predicted, and the step difference prediction value is used to characterize the predicted height difference between the main spacer area and the non-main spacer area.

[0143] In an optional implementation, step S16 further includes steps S161 to S163:

[0144] Step S161: The server obtains the first preset threshold, the second preset threshold, the third preset threshold, and the fourth preset threshold.

[0145] The first preset threshold is the upper limit of the height specification of the main spacer of the display panel to be predicted, the second preset threshold is the lower limit of the height specification of the main spacer of the display panel to be predicted, the third preset threshold is the upper limit of the step specification of the main spacer of the display panel to be predicted, and the fourth preset threshold is the lower limit of the step specification of the main spacer of the display panel to be predicted.

[0146] In this embodiment, the first preset threshold, the second preset threshold, the third preset threshold, and the fourth preset threshold are derived from any of the following methods: specification requirement information carried in the first prediction request sent by the client; querying and obtaining from the manufacturing execution system or product specification database by the server based on the product model; or obtaining from the process formula by the server based on the batch identification information.

[0147] For example, for a certain model of display panel, the height of the main spacer is specified to be between 2.4μm and 2.6μm, that is, the first preset threshold (upper limit) is 2.6μm and the second preset threshold (lower limit) is 2.4μm; the step difference is specified to be between 0.08μm and 0.15μm, that is, the third preset threshold (upper limit) is 0.15μm and the fourth preset threshold (lower limit) is 0.08μm.

[0148] Step S162: If the server determines that the estimated height of the main septum is greater than or equal to the second preset threshold and less than or equal to the first preset threshold, and the estimated step difference is greater than or equal to the fourth preset threshold and less than or equal to the third preset threshold, then the server generates the first quality prediction result and sends the first quality prediction result to the client.

[0149] The first quality prediction result indicates that the septum material is of acceptable quality. This means that the height and step difference of the main septum material in the first piece of the current batch meet the design specifications. The server returns the first quality prediction result to the client, displaying the estimated height and step difference of the main septum material in a dashboard format on the client's visual interface. The first quality prediction result is marked with a green indicator or a "Acceptable" label, visually indicating to the engineer that the quality status of the first piece in the current batch meets the design specifications. Based on this acceptable feedback and the input process parameters, the engineer then executes the photolithography process operation for the current display panel.

[0150] For example, when the algorithm service module returns a predicted height of 2.51 μm, a predicted step difference of 0.12 μm, and a specification range of MPSH: 2.4 μm~2.6 μm, MS: 0.08 μm~0.15 μm, since 2.4 μm ≤ 2.51 μm ≤ 2.6 μm and 0.08 μm ≤ 0.12 μm ≤ 0.15 μm, the quality is deemed acceptable, and a first quality prediction result is generated.

[0151] Step S163: If the server determines that the estimated height of the main septum is greater than the first preset threshold, or the estimated height of the main septum is less than the second preset threshold, or the estimated step difference is greater than the third preset threshold, or the estimated step difference is less than the fourth preset threshold, then the server generates the second quality prediction result.

[0152] The second quality prediction result indicates that the septum material is substandard. The server returns the second quality prediction result to the client, where it is displayed in red or labeled "substandard" on the client's visualization interface.

[0153] In an optional implementation, after step S163, the method further includes step S164:

[0154] Step S164: The server inputs the estimated height of the main septum and the estimated step difference into the preset analysis module, obtains the process adjustment parameters output by the preset analysis module, and sends the process adjustment parameters to the client.

[0155] The preset analysis module is deployed on the server side and has pre-configured mapping relationships between process parameters and the height of the main diaphragm, as well as between process parameters and step differences. Based on the deviations between the estimated height of the main diaphragm and the estimated step difference and their respective design specifications, the preset analysis module uses the mapping relationships to inversely calculate the adjustment direction and magnitude of each process parameter, thus generating process adjustment parameters.

[0156] The process adjustment parameters include the name of the process parameter to be adjusted, the adjustment direction, and the suggested adjustment amount. The preset analysis module generates corresponding attribution analysis results and / or process adjustment parameters based on the attribution contribution. For example, when the estimated height of the main septum is lower than the design specification, the preset analysis module outputs specific attribution analysis results, such as a 63% contribution from the developer chamber pressure fluctuation and a 28% contribution from the exposure energy deviation, to directly guide engineers in locating the root cause of the problem; or the preset analysis module directly outputs the process adjustment parameter "exposure dose: increase by 5 mJ / cm²"; when the estimated step difference exceeds the design specification, the preset analysis module outputs corresponding process adjustment parameters, such as "developer temperature: decrease by 0.3 ℃".

[0157] The server encapsulates the process adjustment parameters into a response message and returns it to the client via the factory's internal Ethernet. After receiving the process adjustment parameters, the client displays them on a visual interface to assist engineers in adjusting the process parameter settings for subsequent batches.

[0158] For example, the client displays the attribution contribution of each device parameter in the virtual measurement model to the prediction results in the form of a pie chart or bar chart, and / or displays adjustment measures suggestions corresponding to the attribution contribution in the suggestion column, providing engineers with actionable directions for process adjustments.

[0159] In this embodiment, the preset analysis module, based on the real-time measurement deviation of the height and step difference of the main septum, reversely decouples the independent influence weights of each process parameter through the built-in mapping relationship, thereby generating quantifiable and traceable process adjustment parameters. This solves the technical problem in the prior art where process parameter adjustment relies too much on human experience and is difficult to decouple from real-time equipment conditions due to the lag in first-piece inspection.

[0160] In one alternative implementation, such as Figure 7 As shown, the method further includes steps S18 to S21:

[0161] Step S18: The server receives the second prediction request sent by the client.

[0162] The second prediction request carries the corrected process parameters. These corrected process parameters are the updated values ​​obtained by the engineer after modifying the process parameters for the current batch on the client's visual interface based on the process adjustment parameters returned in step S164. The corrected process parameters include at least one of the following: corrected photoresist coating thickness preset value, corrected exposure machine dosage preset value, corrected developer temperature preset value, and corrected baking temperature preset value.

[0163] For example, when the process adjustment parameter output in step S164 is "increase exposure dose by 2mJ / cm²", the engineer changes the preset value of the exposure machine dose from 150mJ / cm² to 152mJ / cm² on the client interface and clicks the "predict again" button. The client encapsulates the corrected process parameters into a second prediction request and sends it to the server.

[0164] Step S19: The server performs feature fusion on the corrected process parameters and the first measurement data to obtain the second fused feature tensor.

[0165] In this embodiment, the server uses the corrected process parameters as updated static scalar features and fuses them with the first measurement data in the same manner as described above to obtain a second fused feature tensor. The first measurement data remains unchanged because it reflects the actual operating state of the equipment at the moment of the first piece processing and does not change with the corrected process parameters input by the user.

[0166] The specific feature fusion method is the same as step S13, including: scalar feature encoding of the corrected process parameters; temporal feature extraction and spatial feature encoding of the first measurement data; and dimensional concatenation of the above features to generate a second fused feature tensor.

[0167] Step S20: The server sends the second fused feature tensor to the algorithm service module and receives the corrected main septum height estimate and the corrected step difference estimate returned by the algorithm service module.

[0168] The server encapsulates the second fused feature tensor into a second inference request and sends it to the algorithm service module. The algorithm service module takes the second fused feature tensor as input, performs inference calculations through the virtual measurement model, and outputs the corrected estimated height of the main septum and the corrected estimated step difference.

[0169] Step S21: The server generates a third mass prediction result based on the corrected estimated height of the main septum and the corrected estimated step difference, and returns the third mass prediction result to the client.

[0170] The server generates a third quality prediction result based on the corrected estimated height of the main septum material and the corrected estimated step difference, combined with a first preset threshold, a second preset threshold, a third preset threshold, and a fourth preset threshold. If the server determines that the corrected estimated height of the main septum material is greater than or equal to the second preset threshold and less than or equal to the first preset threshold, and the corrected estimated step difference is greater than or equal to the fourth preset threshold and less than or equal to the third preset threshold, then the third quality prediction result indicates that the quality of the main septum material is qualified. If the server determines that the corrected estimated height of the main septum material is greater than the first preset threshold, or less than the second preset threshold, or greater than the third preset threshold, or less than the fourth preset threshold, then the third quality prediction result indicates that the quality of the main septum material is unqualified.

[0171] The server returns the third-party quality prediction results to the client, where they are displayed to the engineer on the client's visualization interface.

[0172] When the third quality prediction result is qualified, the engineer can confirm that the process adjustment is effective and release the current batch; when the third quality prediction result is still unqualified, the engineer can perform a new round of adjustment operations on the corrected process parameters.

[0173] Optionally, when the client receives a third quality prediction result that is unqualified, the client also receives updated process adjustment parameters returned by the server.

[0174] In one possible implementation, after the first product completes actual measurement, the measurement equipment or Manufacturing Execution System (MES) transmits the actual height value of the main spacer, the step value, and the corresponding glass substrate identifier back to the server via a preset data interface. The server combines the corresponding first fusion feature tensor with the actual value to form a new training sample and stores it in a preset training database. The first fusion feature tensor and the actual value are associated through the glass substrate identifier. The algorithm service module uses the new training samples in the training database to fine-tune the virtual measurement model or perform online learning, enabling the virtual measurement model to continuously optimize the prediction accuracy of the first product.

[0175] Figure 8 This is a schematic diagram of the structure of the terminal device provided in the embodiments of this application. For example... Figure 8 As shown, the terminal device 8 of this embodiment includes: at least one processor 80 ( Figure 8 The diagram shows only one processor, a memory 81, and a computer program 82 stored in the memory 81 and executable on the at least one processor 80, which, when executing the computer program 82, implements the steps in any of the above embodiments of the method for predicting the quality of the main spacer of the display panel.

[0176] The terminal device may be a desktop computer, laptop, handheld computer, or cloud server, etc. This terminal device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that... Figure 8 This is merely an example of terminal device 8 and does not constitute a limitation on terminal device 8. It may include more or fewer components than shown in the figure, or combine certain components, or different components, such as input / output devices, network access devices, etc.

[0177] The processor 80 may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0178] In some embodiments, the memory 81 may be an internal storage unit of the terminal device 8, such as a hard disk or memory of the terminal device 8. In other embodiments, the memory 81 may be an external storage device of the terminal device 8, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the terminal device 8. Furthermore, the memory 81 may include both internal and external storage units of the terminal device 8. The memory 81 is used to store the operating system, applications, boot loader, data, and other programs, such as the program code of the computer program. The memory 81 can also be used to temporarily store data that has been output or will be output.

[0179] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the steps in the above-described method embodiments.

[0180] This application provides a computer program product that, when run on a terminal device, enables the terminal device to implement the steps described in the various method embodiments above.

[0181] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of this application can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying computer program code to a device / terminal equipment, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electrical carrier signals or telecommunication signals.

[0182] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0183] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0184] In the embodiments provided in this application, it should be understood that the disclosed devices / terminal equipment and methods can be implemented in other ways. For example, the device / terminal equipment embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling or direct coupling or communication connection may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0185] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0186] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

[0187] The technical solutions of this application have been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of this application is obviously not limited to these specific embodiments. Without departing from the principles of this application, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of this application.

Claims

1. A method for predicting the quality of the main spacer material in a display panel, characterized in that, Applied to the server side, including: Receive a first prediction request sent by the client, and extract the process parameters carried in the first prediction request, wherein the process parameters include preset values ​​for photoresist coating thickness, exposure machine dosage, developer temperature, and baking temperature. Acquire first measurement data of the measuring device within a preset time period, wherein the first measurement data includes time series data of the rotation speed of the coating unit and acceleration change curve, attenuation curve of the output energy of the exposure lamp, concentration value and temperature value of the developer and temperature distribution data of the baking unit. The process parameters and the first measurement data are fused to obtain a first fused feature tensor. The first fused feature tensor is sent to the algorithm service module, and the algorithm service module returns the estimated value of the main septum height and the estimated value of the step difference; wherein, the step difference is the height difference between the main septum region and the non-main septum region; Based on the estimated height of the main spacer and the estimated step difference, a quality prediction result for the main spacer of the display panel to be predicted is generated, and the quality prediction result is returned to the client.

2. The method for predicting the quality of the main spacer material in a display panel according to claim 1, characterized in that, The acquisition of the first measurement data of the measurement device within a preset time period includes: A data acquisition command is sent to the measuring device through the first communication interface; Receive the first measurement data of the measuring device within a preset time period returned by the measuring device through the first communication interface; The measuring equipment includes a coating machine, an exposure machine, a developing machine, and a baking machine.

3. The method for predicting the quality of the main spacer material in a display panel according to claim 2, characterized in that, The step of fusing the process parameters and the first measurement data to obtain a first fused feature tensor includes: Send a data processing request to the database engine, the data processing request carrying the first measurement data; Receive ordered feature data returned by the database engine; wherein, the ordered feature data is a fixed-length ordered array generated by the database engine after extracting and aligning features from the first measurement data according to pre-configured feature configuration data; The ordered feature data and the process parameters are concatenated to obtain the first fused feature tensor.

4. The method for predicting the quality of the main spacer material in a display panel according to claim 1, characterized in that, The step of sending the first fused feature tensor to the algorithm service module and receiving the estimated height of the main septum and the estimated step difference returned by the algorithm service module includes: The first fused feature tensor is encapsulated into a first inference request, and the first inference request is sent to the algorithm service module; the first inference request includes the first fused feature tensor and product information of the main spacer of the display panel to be predicted; Receive the estimated height of the main septum and the estimated step difference returned by the algorithm service module.

5. The method for predicting the quality of the main spacer material in a display panel according to claim 1, characterized in that, The step of generating a quality prediction result for the main spacer of the display panel based on the estimated height of the main spacer and the estimated step difference, and sending the quality prediction result to the client, includes: Obtain a first preset threshold, a second preset threshold, a third preset threshold, and a fourth preset threshold, wherein the first preset threshold is the upper limit of the height specification of the main spacer of the display panel to be predicted, the second preset threshold is the lower limit of the height specification of the main spacer of the display panel to be predicted, the third preset threshold is the upper limit of the step specification of the main spacer of the display panel to be predicted, and the fourth preset threshold is the lower limit of the step specification of the main spacer of the display panel to be predicted. If the estimated height of the main septum is greater than or equal to the second preset threshold and less than or equal to the first preset threshold, and the estimated step difference is greater than or equal to the fourth preset threshold and less than or equal to the third preset threshold, then a first quality prediction result is generated and sent to the client, wherein the first quality prediction result indicates that the septum is of acceptable quality. If the estimated height of the main septum is greater than the first preset threshold, or the estimated height of the main septum is less than the second preset threshold, or the estimated step difference is greater than the third preset threshold, or the estimated step difference is less than the fourth preset threshold, then a second quality prediction result is generated, wherein the second quality prediction result indicates that the septum is of substandard quality.

6. The method for predicting the quality of the main spacer in a display panel according to claim 5, characterized in that, After generating the second quality prediction result, the method further includes: The estimated height of the main diaphragm and the estimated step difference are input into the preset analysis module to obtain the process adjustment parameters output by the preset analysis module, and the process adjustment parameters are sent to the client.

7. The method for predicting the quality of the main spacer in a display panel according to claim 6, characterized in that, The method further includes: Receive a second prediction request sent by the client, wherein the second prediction request carries the corrected process parameters; The corrected process parameters and the first measurement data are fused to obtain a second fused feature tensor. The second fused feature tensor is sent to the algorithm service module, and the corrected main septum height estimate and the corrected step difference estimate are received from the algorithm service module. Based on the corrected estimated height of the primary septum and the corrected estimated step difference, a third quality prediction result is generated and returned to the client.

8. A method for predicting the quality of a main spacer in a display panel, characterized in that, Applied to the client side, including: The process parameters input by the user are obtained, and a first prediction request is generated based on the process parameters, wherein the first prediction request carries the process parameters input by the user; the process parameters include preset values ​​for photoresist coating thickness, exposure machine dosage, developer temperature, and baking temperature. Send the first prediction request to the server; Receive the quality prediction results returned by the server.

9. The method for predicting the quality of the main spacer in a display panel according to claim 8, characterized in that, Receiving the quality prediction result returned by the server includes: Receive a first quality prediction result returned by the server, wherein the first quality prediction result indicates that the spacer material is of acceptable quality; or... The system receives a second quality prediction result and process adjustment parameters returned by the server. The second quality prediction result indicates that the septum material is of substandard quality. The process adjustment parameters are obtained by the server after generating the second quality prediction result by inputting the estimated height of the main septum material and the estimated step difference into a preset analysis module.

10. The method for predicting the quality of the main spacer in a display panel according to claim 9, characterized in that, The method further includes: If the process adjustment parameters are received, a second prediction request is sent to the server; wherein the second prediction request carries the corrected process parameters; The server receives a third quality prediction result, which is generated by the server based on the corrected main septum height estimate and the corrected step difference estimate. The corrected main septum height estimate and the corrected step difference estimate are obtained by the server performing feature fusion of the corrected process parameters and the first measurement data to obtain a second fused feature tensor, and then sending the second fused feature tensor to the algorithm service module for inference.

11. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1 to 7; or, when the processor executes the computer program, it implements the method as described in any one of claims 8 to 10.

12. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 7; or, when the computer program is executed by a processor, it implements the method as described in any one of claims 8 to 10.

Images