Glassware production whole-process tracing method and system based on MES

Abstract

The application discloses a glass device production whole-process tracing method and system based on MES, and belongs to the field of glass device production, wherein the method comprises the following steps: deploying edge computing nodes at each production station and each measurement station of the glass device; reading the two-dimensional code on the glass device and the two-dimensional code on the carrier through a code reading device to obtain product identification information and carrier identification information, and uploading the edge computing nodes; obtaining measurement data of the glass device, including size data, optical data and appearance data; the edge computing nodes compare the obtained measurement data with the stored standard threshold value of the corresponding process in real time to determine whether there is an abnormal trend; if it is determined that there is an abnormal trend, the edge computing nodes automatically trigger a tracing process based on the product identification information and the carrier identification information. The application solves the technical problems of single tracing range, high production loss and low quality control efficiency of the existing glass device tracing technology.

Description

Technical Field

[0001] This invention relates to the field of glass device manufacturing, and specifically to a method and system for tracing the entire glass device manufacturing process based on MES. Background Technology

[0002] The production of glass components, especially high-end optical glass and display glass, is characterized by long process chains, stringent precision requirements, large production batches, and irreversible high-value processes such as coating and bonding. Traditional MES-based production traceability systems generally adopt a centralized cloud-based data processing architecture. Measurement data from all workstations must be completely uploaded to the cloud server before anomaly analysis and traceability operations can be performed. This not only suffers from high data transmission latency and network fluctuations that can easily lead to data loss or delayed anomaly response, but also allows for passive traceability only after non-conforming products are found during final product inspection. It cannot identify early anomaly trends such as process drift in real time during production. At the same time, the scope of existing traceability methods is limited to directly related individual products or carriers, lacking a precise expansion mechanism based on historical anomaly propagation patterns. This can easily lead to insufficient traceability scope resulting in non-conforming products being missed and flowing into subsequent processes, or excessive traceability scope causing unnecessary production stoppages and wasted re-inspection costs. It also makes it difficult to quickly locate the root cause of anomalies, which seriously restricts the improvement of yield and cost control in glass component production. Summary of the Invention

[0003] This application provides a method and system for tracing the entire production process of glass components based on MES. It aims to solve the technical problems of existing glass component traceability technologies, such as limited traceability scope, resulting in high production losses and low quality control efficiency. The goal is to improve the efficiency of abnormal handling and the accuracy of traceability, while reducing production losses and quality control costs.

[0004] In view of the above problems, this application provides a method and system for tracing the entire production process of glass devices based on MES.

[0005] The first aspect disclosed in this application provides a method for tracing the entire glass device manufacturing process based on MES, the method comprising: Edge computing nodes are deployed at each production station and measurement station for glass components, and the barcode readers and measurement devices at each station are connected to the edge computing nodes. The barcode readers read the QR codes on the glass components and the carriers to obtain product identification information and carrier identification information, which are then uploaded to the corresponding edge computing nodes. The measurement devices obtain measurement data of the glass components at the corresponding stations, including dimensional data, optical data, and appearance data, and upload them to the corresponding edge computing nodes. Each edge computing node compares the obtained measurement data with the standard thresholds for the corresponding processes stored in the MES system in real time to determine if there are any abnormal trends. If an abnormal trend is determined, the edge computing nodes automatically trigger a traceability process based on the product identification information and the carrier identification information.

[0006] Another aspect disclosed in this application provides a full-process traceability system for glass device manufacturing based on MES, the system comprising: The system includes a connection module for deploying edge computing nodes at each production and measurement station of the glass components, and connecting the corresponding barcode readers and measurement devices at each station to the edge computing nodes; a data acquisition module for reading QR codes on the glass components and carriers using the barcode readers to obtain product identification information and carrier identification information, and uploading this information to the corresponding edge computing nodes; a measurement module for acquiring measurement data of the glass components at the corresponding stations, including dimensional data, optical data, and appearance data, using the measurement devices, and uploading this data to the corresponding edge computing nodes; a judgment module for each edge computing node to compare the acquired measurement data with the standard thresholds for the corresponding processes stored in the MES system in real time to determine if there are any abnormal trends; and a traceability module for automatically triggering a traceability process based on the product identification information and carrier identification information if an abnormal trend is detected.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages: By deploying edge computing nodes at each production and measurement station, local real-time data processing and statistical process control analysis are achieved. Combined with the integrated technology of dynamic binding of product and carrier dual codes, dynamic determination of traceability extension K value based on historical anomaly inspection dataset, and automatic locking and isolation of abnormal carriers by the material transfer controller of the entire station, standardized traceability data packages and re-inspection tasks are automatically generated. This realizes the transformation from passive post-event traceability to pre-event anomaly warning and automatic in-event interception, improving the efficiency of anomaly handling and the accuracy of traceability, and reducing production losses and quality control costs.

[0008] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0009] Figure 1 This application provides a flowchart illustrating a method for tracing the entire glass device manufacturing process based on MES. Figure 2 A schematic diagram of the structure of a MES-based glass device manufacturing process traceability system is provided for embodiments of this application.

[0010] Explanation of reference numerals in the attached diagram: Connection module 11, Acquisition module 12, Measurement module 13, Judgment module 14, Traceability module 15. Detailed Implementation

[0011] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0012] The overall concept of the technical solution provided in this application is as follows: This application provides a method and system for tracing the entire production process of glass components based on MES. By deploying edge computing nodes at each production and measurement station, it enables local real-time data processing and anomaly trend prediction. Combined with integrated technologies such as product and carrier dual-code binding to establish a full-process association, dynamically determining the traceability extension K value based on historical anomaly inspection datasets, and automatic locking and isolation of abnormal carriers by the material transfer controllers at all stations, the anomaly response delay is compressed from the traditional second level to the millisecond level. It also automatically generates standardized traceability data packages and re-inspection tasks, realizing the transformation from passive post-event traceability to pre-event warning and in-event interception, improving the efficiency of anomaly handling and the accuracy of full-process traceability, and reducing production losses and quality control costs.

[0013] After introducing the basic principles of this application, various non-limiting embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0014] Example 1, as Figure 1 As shown in the embodiment of this application, a method for tracing the entire glass device manufacturing process based on MES is provided. The method includes: S100: Edge computing nodes are deployed in each production station and measurement station of glass devices, and the corresponding code reading devices and measurement devices of the stations are connected to the edge computing nodes.

[0015] Specifically, the first step is to conduct a thorough investigation of all production stations on the glass device production line, including process stations that directly change the physical form of the product, such as cutting, edging, coating, cleaning, and bonding, as well as measurement stations, including quality inspection process stations such as dimensional measurement, optical performance testing, and appearance defect detection. An edge computing node should be independently deployed next to the control cabinet of each station. This node is a small industrial-grade computing device deployed on the production site close to the data source, which has the ability to process local data in real time, respond with low latency, and store data locally when the network is down.

[0016] Subsequently, the dedicated equipment for each workstation was integrated. Regarding the selection of barcode readers, for the QR codes on the glass components themselves, due to the transparent and reflective nature of glass, an industrial-grade high-resolution vision barcode reader was chosen and deployed above the workstation's feed inlet. This automatically triggers scanning to obtain product identification information, i.e., the unique identification code for each glass component. This is typically achieved through laser etching on the edge of the component, recording basic information such as batch number, production date, and production order. For the trays / boxes carrying the glass components, metal-etched QR codes are used, paired with an industrial barcode scanner to read the carrier's identification information, i.e., the unique code identifying each carrier container, including... Information such as carrier number, capacity, and applicable processes can be linked to product identification to enable rapid traceability of batch products. All barcode readers are directly connected to the edge computing nodes of the corresponding workstations via USB or industrial Ethernet interfaces. Measurement equipment includes dimensional measuring instruments, optical inspection instruments, and appearance inspection machines. These devices generally have standard industrial communication interfaces and establish physical connections with edge computing nodes via shielded twisted-pair cables or industrial switches. At the same time, the corresponding device drivers and data parsing plugins are pre-installed in the edge computing nodes to convert unstructured measurement data output by different devices into structured data in a unified format.

[0017] Finally, the system initialization configuration is completed. Workstation identity registration is completed in each edge computing node, and the node ID is uniquely bound to the corresponding workstation number and process type. A two-way encrypted communication link is established with the MES system, and data upload rules are configured: code reading data is uploaded in real time, measurement data is synchronized in real time at the single product granularity, and data is automatically temporarily stored in the local solid-state drive of the edge computing node when the network is interrupted. After the network is restored, the data is automatically resumed to the MES system. At the same time, basic data such as the process standard threshold and historical statistical process control parameters corresponding to the workstation are pre-downloaded from the MES system to prepare for subsequent real-time anomaly judgment.

[0018] S200: The QR code on the glass device and the QR code on the carrier are read by the QR code reading device to obtain product identification information and carrier identification information, and then uploaded to the corresponding edge computing node.

[0019] Specifically, when the carrier carrying the glass component arrives at the feed inlet of the corresponding workstation at a constant speed along the production line conveyor belt, the photoelectric sensors deployed on both sides of the feed inlet—that is, non-contact object detection devices—determine whether the object is in place by emitting and receiving modulated infrared light. They immediately send a millisecond-level trigger signal to the corresponding barcode reader, simultaneously initiating the scanning process for the QR codes on both the glass component and the carrier. For the laser-etched QR codes on the glass component itself, the barcode reader automatically switches to a polarized light source mode adapted to the glass material, a high frame rate scanning mode, and a reflection suppression algorithm. After reading the original QR code string, it parses it to obtain the product identification information. Each glass component has a unique, lifetime identification code, typically etched with a high-precision laser onto the edge of the non-functional area. The code includes the production work order ID, batch number, globally unique serial number, production date, production line number, and shift information. Simultaneously, a barcode reader scans the metal-etched QR codes on the high-temperature resistant, anti-static trays or containers carrying the glass components, extracting the carrier identification information—the unique code for each reusable carrier container. This code includes the carrier number, rated capacity, last calibration date, applicable process range, and historical maintenance records. This information, linked to the product identification, enables rapid traceability of batch products.

[0020] After the barcode reader obtains the original QR code string, it first executes the built-in cyclic redundancy check algorithm to verify the data integrity. If the verification fails, it automatically triggers a rescanning mechanism, rescanning up to 3 times consecutively, adjusting the scanning angle and light source brightness each time. If it still cannot be read successfully, it immediately sends a barcode reading anomaly signal to the corresponding edge computing node. The edge node generates a rescan reminder containing the anomaly location and anomaly type and pushes it to the operator's handheld terminal. For double-code data that passes verification, the edge computing node will pre-calibrate according to the production line conveyor belt speed and workstation spacing within a preset time window, usually 500ms, to ensure that the barcode reading data of the same carrier and all the glass devices it carries are classified into the same batch. It establishes a dynamic binding relationship between all product identification information in the batch and the corresponding carrier identification information, that is, it generates a temporary carrier-product mapping table, records the product identification of all glass devices currently carried by each carrier, and synchronously adds metadata such as binding timestamp, workstation number, barcode reader number, and operator ID.

[0021] Finally, the edge computing node will upload the parsed and bound structured data to its local time-series database in real time via the MQTT 3.1.1 protocol, a lightweight, low-bandwidth, and highly reliable publish / subscribe IoT communication protocol. At the same time, it is configured with a network outage caching mechanism. When the communication link with the MES system is interrupted, the data will be automatically temporarily stored in the industrial-grade local solid-state drive of the edge node and will be automatically resumed after the network is restored, ensuring the continuity and integrity of production data.

[0022] S300: Obtain measurement data of the glass device under the corresponding workstation through the measurement device, including size data, optical data, and appearance data, and upload it to the corresponding edge computing node.

[0023] Specifically, through the automated collaborative work and standardized data processing of multiple types of measurement equipment: when the carrier carrying the glass component to be tested is transported to the designated position of the measurement station via a conveyor belt, after the photoelectric sensor detects the carrier's arrival and completes the reading and binding of the product identification and carrier identification, the edge computing node immediately sends a start measurement command to the corresponding measurement equipment, triggering a fully automated measurement process; among which, the dimensional data is collected by high-precision contact or non-contact dimensional measuring instruments, referring to the key geometric parameters of the glass component, including length, width, thickness, chamfer angle, hole position accuracy, etc., which directly affect the assembly compatibility and structural strength of the product. Non-contact measuring instruments usually use laser triangulation or machine vision technology, which can avoid scratches on the glass surface caused by contact measurement, and are particularly suitable for For the inspection of ultra-thin glass devices below 0.1mm, optical data is collected by specialized optical equipment such as spectrometers, transmittance meters, and refractive index meters. These data are parameters that determine the optical performance of glass devices, including visible light transmittance, reflectance, haze, refractive index uniformity, and full-spectrum transmittance curves from 380 to 780 nm. These are the most important quality indicators for high-end glass devices. Appearance data is collected by a high-speed linear array camera paired with an AI vision inspection system. This data refers to various defects on the surface of the glass devices, including micron-level scratches, bubbles, chipping, black spots, dirt, and printing defects. The AI ​​vision inspection system can automatically identify, classify, and measure the size of defects through pre-trained deep learning algorithms. The inspection speed can reach 10 products per second, which is far more efficient and accurate than manual visual inspection.

[0024] After various measurement devices complete their testing, they transmit the raw measurement data to the corresponding edge computing nodes via standard industrial communication interfaces such as RS485, Modbus RTU / TCP, and TCP / IP. The device drivers and data parsing plugins pre-installed in the edge computing nodes automatically convert the unstructured raw data output by different devices, including binary streams, CSV format text, and high-definition defect images, into structured data in a unified format. At the same time, they extract the device operating status information, including device temperature, runtime, last calibration date, sensor voltage, and correlate it with the measurement data to perform data validity verification. If uncalibrated devices, abnormal sensors, or data exceeding the physical reasonable range are detected, the data is immediately marked as invalid, and a device anomaly warning is generated and pushed to the operator terminal to avoid erroneous data interfering with subsequent anomaly trend judgment.

[0025] Subsequently, the verified measurement data will be precisely associated with the product identification information already read by the workstation. Each measurement data will be given unique metadata such as product ID, vehicle ID, workstation number, measurement timestamp, equipment number, and operator ID, forming a complete quality data record at the single-product granularity. This record will be stored in the local time-series database of the edge computing node, which is a database optimized for time-series data and can handle hundreds of high-concurrency data writes and fast queries per second.

[0026] Finally, the edge computing nodes will upload the structured measurement data to the MES system in real time via the MQTT 3.1.1 lightweight IoT protocol according to the preset transmission rules. At the same time, industrial-grade local caching and breakpoint resume mechanism are enabled. When the communication link with the MES system is interrupted, all measurement data will be automatically temporarily stored in the high-capacity solid-state drive of the edge node. The cache capacity can support 72 hours of continuous production data storage. After the network is restored, the system will automatically resume the data transmission from the breakpoint to ensure the continuity and integrity of production data and prevent data loss due to short-term network fluctuations.

[0027] S400: Each edge computing node will compare the acquired measurement data with the standard thresholds of the corresponding process stored in the MES system in real time to determine whether there is an abnormal trend.

[0028] Specifically, when a workstation starts up or a process is switched, each edge computing node automatically pre-caches the standard thresholds for the corresponding process from the MES system. These thresholds are the upper and lower limits that each quality parameter must meet, as determined by the enterprise based on product design drawings, industry quality standards, and customer requirements. They serve as the basis for determining whether a single product is qualified. Additionally, the system also includes the historical statistical process control upper limit, or SPC control upper limit. This limit is calculated using statistical process control methods based on massive amounts of measurement data from the workstation's stable production status over the past 30 days. Its range is stricter than the standard thresholds and is used to capture minor abnormal fluctuations in the production process that have not yet led to product non-compliance. Simultaneously, the system also synchronously presets the deviation coefficient, sliding window size, and trend judgment rules. All calculation logic is executed locally on the edge node.

[0029] Subsequently, after the edge node obtains the structured measurement data that is precisely bound to the product identifier, it first performs data preprocessing to remove previously marked invalid abnormal equipment data, duplicate data, and dirty data that exceeds the physical reasonable range. Then, it uses a moving average filtering algorithm to eliminate data jitter caused by sensor noise and accidental environmental factors, ensuring the reliability of subsequent comparison results.

[0030] Subsequently, for each glass device, dozens of key measurement parameters in three categories—size, optics, and appearance—we calculate the deviation from the corresponding target value. The deviation is defined as (actual measured value - target value) / (control upper limit - target value), which is used to standardize and quantify the degree of deviation of the measured value from the ideal state. The larger the value, the more severe the process fluctuation. When the deviation of any parameter exceeds the preset warning coefficient, which is usually set to 0.8, that is, when the measured value reaches 80% of the SPC control upper limit, it is immediately marked as a potential anomaly.

[0031] To avoid misjudgments caused by a single random anomaly, the edge node maintains an adjustable sliding window, typically taking 5-10 consecutive products. This window is dynamically adjusted based on the production line cycle time and process characteristics. It continuously stores the measurement data and deviation results of the most recent N products and applies classic trend judgment rules from statistical process control for comprehensive analysis. These rules include: 7 consecutive points continuously rising or falling, indicating unidirectional drift in the process; 9 out of 10 points being on the same side of the target value, indicating a shift in the center of gravity of the process; 2 out of 3 consecutive points falling in the area above 2 / 3 of the control upper limit, indicating a sharp increase in fluctuations in the process. This effectively identifies gradually deteriorating systematic abnormal trends in the production process, rather than isolated random fluctuations.

[0032] When the data within the sliding window meets any trend judgment rule, or the deviation of a single product exceeds 1.0, i.e., reaches or exceeds the SPC control limit, the edge computing node will immediately determine that there is an abnormal trend. At the same time, it will record the precise timestamp of the abnormality, the workstation number, the abnormal parameter type, the abnormal value, the deviation value, and all historical data within the current sliding window, providing a complete abnormal context for subsequent traceability processes. If no abnormal trend is detected, the measurement data and normal production status markers will be synchronously uploaded to the MES system, updating the full-process production file of the corresponding product. The edge node will also automatically synchronize the latest standard thresholds and SPC control limits from the MES every hour to ensure that the judgment standards of all workstations are consistent. When the MES system updates process parameters or quality requirements, the edge node will automatically load the new configuration without manual on-site intervention.

[0033] S500: If an abnormal trend is detected, the edge computing node will automatically trigger the traceability process based on the product identification information and the vehicle identification information.

[0034] Specifically, when an edge computing node confirms a systematic abnormal trend through statistical process control analysis, it will immediately extract the abnormal product identification information corresponding to the glass device that triggered the abnormality from the local time-series database. This information includes the unique identification code of the single or multiple glass devices that led to the abnormal trend determination, containing their complete production trajectory index and abnormal carrier identification information, i.e., the unique code of the carrier that carries these abnormal glass devices. This information serves as the core anchor point for expanding the traceability scope. These identifiers are marked as traceable in the local database, i.e., data tags with the highest priority. This will prohibit the corresponding products and carriers from entering irreversible processes such as coating and bonding. At the same time, they will be highlighted in red in the MES system, blocking the normal flow of abnormal products from the data level.

[0035] Subsequently, the edge computing node will automatically retrieve complete process data of all N glass components passing through the workstation within the preset anomaly time window, which is a time interval set based on the production line cycle time and process response time, usually 5-15 minutes before and after the anomaly occurs, to cover the entire process from the occurrence of the anomaly to its detection. This includes measurement data of each product at all upstream workstations, equipment operating parameters, environmental temperature and humidity records, operator information, and carrier flow trajectory. It will generate an anomaly warning message containing the anomaly type, anomaly parameter name, actual measured value, deviation value, and preliminary assessment of the impact range. This message will be pushed in real time to the operator's handheld terminal, team leader monitoring dashboard, and equipment maintenance terminal of the MES system via the MQTT protocol, while triggering on-site audible and visual alarm devices to remind relevant personnel to handle the situation promptly.

[0036] Subsequently, the edge computing node synchronously sends a request to the MES system to obtain the historical binding records of the abnormal vehicle over the past 72 hours. This records the binding and unbinding timestamp logs of the vehicle with different batches of glass components at various workstations, fully documenting the vehicle's circulation path and carrying history. The traceability extension range is then calculated according to the K-value dynamic determination method preset in this solution: First, the historical abnormal inspection dataset of the workstation over the past 6 months is retrieved from the MES system. The maximum abnormal vehicle span before and after marking the abnormality is identified among all similar abnormal events. This represents the maximum number of consecutive vehicles actually affected by a single abnormal event in history, reflecting the diffusion characteristics of this type of abnormality. Then, the tolerance thresholds of 1-2 vehicles are superimposed as the final K-value. The system automatically identifies the K associated vehicles before and after the abnormality occurs and marks all glass components within these vehicles as requiring traceability, thus achieving precise expansion of the traceability range.

[0037] Next, the edge computing node broadcasts a lock signal containing all abnormal carrier codes to the material transfer controllers at all workstations on the production line. These industrial programmable logic controllers, responsible for controlling the conveyor belt speed, diversion direction, and pneumatic devices, are the core execution units for production line logistics scheduling. This signal has the highest priority and overrides normal production scheduling instructions. When any workstation's material transfer controller detects an abnormal carrier from the lock signal arriving at its diversion point, it immediately suspends the normal flow logic at that diversion point and automatically activates the pneumatic baffle, a fast-lifting physical blocking device driven by compressed air with a response time of less than 100ms. This allows for precise diversion without damaging carriers and glass components, pushing the abnormal carrier into a specially designed isolation buffer zone. This buffer zone is a separate temporary storage area next to the production line, completely isolated from normal production logistics, equipped with independent code readers, status indicator lights, and video surveillance, used to store abnormal products and carriers awaiting re-inspection.

[0038] As the abnormal carrier is pushed into the isolation buffer, the edge computing node automatically records the isolation time, isolation station number, and buffer location number, generating a standardized isolation event. This event is then fully reported to the MES system, triggering a full re-inspection of all glass components within the isolation buffer. The MES system automatically updates the product status based on the re-inspection results: for products that pass the re-inspection, the traceability status mark is cleared and a release instruction is generated, allowing the material transfer controller to return them to the normal production process; for products that fail the re-inspection, they are marked as scrap and pushed to the scrap handling process. Simultaneously, the re-inspection data is synchronized to the quality analysis module of the MES system for subsequent anomaly cause analysis and process improvement.

[0039] Furthermore, in the method provided in the application embodiment, the edge computing node automatically triggers the traceability process based on the product identification information and the carrier identification information, including: obtaining the abnormal product identification information and abnormal carrier identification information corresponding to the glass device with an abnormal trend, marking it as a traceability required state, and uploading it to the MES; recording the complete process data of N glass devices before and after the time window of the abnormal trend occurrence, generating abnormal early warning information, and pushing it to the operator terminal of the MES for re-inspection reminder.

[0040] Specifically, when an edge computing node confirms a systematic abnormal trend through deviation calculation and SPC trend analysis, it will immediately retrieve the abnormal product identification information corresponding to the glass device that triggered the abnormality judgment from the local time-series database, which is an industrial-grade database optimized for time-series data. This database can query product and carrier association data at any point in time. This information includes the unique lifetime identification code of the single or multiple glass devices that triggered the SPC control rules or exceeded the deviation limit. The code contains core information such as production work order ID, batch number, global serial number, production line number, and shift. This information is the sole basis for locating the entire production trajectory of a specific product. The abnormal carrier identification information is the unique code of the reusable tray / box that carries these abnormal glass devices. This code records the carrier's flow path and carrying history. These identifications are uniformly marked as traceable in the local database. This is the highest priority data label preset by the system, which has exclusive control authority and will automatically prohibit the corresponding product from entering irreversible high-value processes such as coating and bonding. At the same time, a red abnormality mark is added to the product's entire process file. When any subsequent workstation's barcode reader scans the label, it will automatically pause the production process and pop up a warning message, thus blocking the normal flow of abnormal products from the data level.

[0041] After marking, the edge computing node will upload the marked anomaly identification information and associated anomaly context data, including the anomaly trigger time, anomaly workstation number, anomaly parameter type, actual measurement value, deviation value, and specific SPC rule triggered, to the MES system in real time via the MQTT 3.1.1 lightweight IoT protocol. The transmission process adopts an acknowledgment and retransmission mechanism. If the MES does not return a receipt acknowledgment, it will automatically retry up to 3 times. When the network is disconnected, the data is temporarily stored on the local industrial-grade solid-state drive of the edge node. After the network is restored, the data will automatically resume from the breakpoint, ensuring that the MES global product status database and the edge node remain synchronized in real time.

[0042] Next, the edge computing node will automatically retrieve all glass components that have passed through the current abnormal workstation within the time interval, based on a preset abnormal time window. This dynamic time interval is typically set to 5-10 minutes before and after the abnormality occurs. The N value, representing the number of products within the time window, will be automatically calculated based on the real-time production line cycle time. The node will then extract complete process data from the raw material input to the current workstation, covering the dimensions, optics, and appearance measurement data of each upstream workstation, as well as equipment operating parameters, including cutting blade speed, coating machine vacuum, cleaning machine water temperature, and workshop environmental data, including temperature and humidity, cleanliness, static electricity, operator ID, process parameter change records, and carrier binding history. All data is associated with the corresponding product identifier, forming a structured traceability data package.

[0043] Finally, the edge computing node generates standardized anomaly warning information based on the aforementioned traceability data packet. This information includes the name and number of the abnormal workstation, the precise time of the anomaly, the name of the abnormal quality parameter, a comparison between the standard threshold and the actual measured value, the deviation, the triggered SPC trend rule, the initial impact range (i.e., N products and their corresponding carrier numbers), and clear re-inspection requirements, including the sampling ratio, re-inspection items, and a 15-minute completion time limit. Through the MES system's hierarchical push mechanism, the warning information and re-inspection reminders are accurately pushed to the operator terminals of the corresponding workstations, including on-site industrial touchscreens and on-duty operator handheld PDAs. In case of serious anomalies, the warning is also pushed to the team leader's terminal, triggering on-site audible and visual alarm devices. This ensures that relevant personnel receive the warning within 30 seconds and initiate the re-inspection process. After completing the re-inspection, the operator must submit the re-inspection results on the terminal, and the system will automatically update the product status based on the results.

[0044] Furthermore, in the method provided in the application embodiment, each edge computing node compares the acquired measurement data with the standard thresholds of the corresponding process stored in the MES system in real time to determine whether there is an abnormal trend. This includes: each edge computing node pre-caching the standard threshold range and historical statistical process control upper limit corresponding to each workstation from the MES; then calculating the deviation of the measurement data of each glass device from the standard threshold range and historical statistical process control upper limit. If any deviation exceeds a preset coefficient, it is determined that there is an abnormal trend.

[0045] Specifically, each edge computing node will automatically initiate a data synchronization request to the MES system when the workstation is powered on, the production line process is switched, or the MES system quality standards are updated. It will pre-cache the standard threshold range of the corresponding process of the current workstation, which is the upper and lower limit range of each quality parameter that the enterprise must meet according to product design drawings, mandatory industry standards and customer customization requirements, as well as the historical statistical process control upper limit, i.e. the control upper limit. This upper limit is calculated by statistical process control methods based on the measurement data of the workstation under stable production conditions in the past 30-90 days. Its range is significantly narrower than the standard threshold range, which can capture early minor process drifts that have not yet caused product defects.

[0046] Meanwhile, the pre-cached content also includes preset deviation coefficients, parameter priority weights, and data filtering rules. All calculation logic and configuration parameters are stored in the local non-volatile memory of the edge node, ensuring that even if the communication link with the MES system is interrupted, the edge node can still independently complete the anomaly judgment. After obtaining the structured measurement data that has been accurately bound to the product identifier, the edge computing node first performs a data validity verification and preprocessing process, automatically removing invalid data previously marked as equipment anomalies, redundant data uploaded repeatedly, and dirty data that exceeds the physical reasonable range, including negative glass thickness and transmittance exceeding 100%. Then, a 3-point moving average filtering algorithm is used to eliminate data jitter caused by accidental factors such as sensor electronic noise and slight vibration of the production line, ensuring the reliability of the subsequent deviation calculation results.

[0047] Subsequently, the edge computing nodes calculate the deviation of each glass device from its target value for dozens of key quality parameters across three categories: size, optics, and appearance. A standardized formula for calculating relative deviation is used. Deviation = |Actual measured value - Parameter target value| / (Historical statistical process control upper limit - Parameter target value); By converting parameters of different dimensions and magnitudes into dimensionless values ​​between 0 and 1, horizontal comparisons of the degree of anomalies in different quality parameters are achieved. The closer the value is to 1, the closer the process fluctuation is to the boundary of being out of control. At the same time, the deviation of the parameter from the standard threshold range is calculated as an auxiliary reference. When the deviation of the SPC control upper limit of any quality parameter exceeds the preset coefficient, the coefficient is dynamically adjusted according to the product accuracy requirements, and the edge computing node will immediately determine that there is an abnormal trend. After the anomaly is determined, the edge node will immediately record the precise timestamp of the anomaly, the name of the abnormal parameter, the actual measured value, the target value, the control upper limit, the deviation value, and the equipment operating status at the time of the trigger, providing a complete anomaly context for the traceability process. If no anomaly is detected, the measurement data and the normal status mark are synchronously uploaded to the MES system to update the full-process quality file of the corresponding product. The edge node will also automatically synchronize the latest standard threshold and SPC control upper limit with the MES system every 2 hours to ensure that the judgment standards of all workstations on the entire production line are consistent. When the MES system issues a process parameter change notification, the edge node will automatically update its local configuration within 10 seconds without manual on-site operation.

[0048] Furthermore, the method provided in the application embodiment, which obtains abnormal product identification information and abnormal carrier identification information corresponding to glass devices with abnormal trends and marks them as traceable states, also includes: requesting historical binding records of abnormal carriers from the MES; identifying K carriers before and after the occurrence of the abnormality based on the historical binding records, and marking all glass devices in the K carriers as traceable states, where K is an integer greater than 1.

[0049] Specifically, when an edge computing node completes the traceability status of a glass device and its directly carrying carrier that triggers an abnormal trend (i.e., the highest priority control tag preset by the system), it will automatically prohibit the corresponding product from entering irreversible high-value processes such as coating and bonding. At the same time, when all workstation barcode readers scan, a red warning will pop up and the production process will be suspended. After marking, it will immediately send a historical binding record request to the MES system via the encrypted MQTT 3.1.1 protocol. The historical binding record is a data asset managed throughout the entire lifecycle of the MES system. It completely stores all binding and unbinding event logs of each reusable carrier since it has been put into use. Each record includes information such as the carrier's unique identifier, a list of product identifiers for all glass devices in the corresponding batch, the binding workstation number, the precise binding timestamp, the unbinding timestamp, the binding operator ID, and the current status of the carrier.

[0050] Upon receiving the request, the MES system will return all binding records of the abnormal vehicle over the past 72 hours and the corresponding vehicle flow timeline of the workstation within 100 milliseconds. The edge computing node then dynamically calculates the K value: First, it retrieves the historical abnormal inspection dataset of the workstation over the past 6 months from the MES system's quality database. This dataset includes the trigger time, abnormality type, list of affected vehicles, final re-inspection results, and root cause analysis reports of all closed abnormal events. From this dataset, it filters out all historical events with the same abnormality type as the current one, including dimensional deviations, optical transmittance drift, and batch appearance of appearance scratches. It then identifies the maximum abnormal vehicle span confirmed by the final re-inspection among these events, which is the maximum number of consecutive vehicles actually affected by a single abnormal event of the same type in history. This objectively reflects the diffusion characteristics and impact range of this type of abnormality. Finally, it adds a tolerance threshold of 1-2 vehicles to the maximum abnormal vehicle span to cover statistical sample errors, unrecorded edge vehicles, and minor deviations in the time of abnormality occurrence. Finally, it determines the K value for this traceability, where K is an integer greater than 1.

[0051] Subsequently, the edge computing node uses the arrival timestamp of the abnormal vehicle at the current workstation as a benchmark, and identifies the K consecutive vehicles before and after the abnormal vehicle's arrival in the vehicle flow sequence table returned by MES in chronological order, forming a complete list of associated vehicles. Then, it retrieves the product identification information of all glass devices bound to each of these 2K vehicles from the MES system through a batch query interface.

[0052] Finally, the edge computing node will add traceable status tags to all glass components in these associated vehicles in batches in the local time-series database and the global product status database of the MES system. At the same time, it will add associated abnormal event IDs to the full process file of each tagged product to ensure that the production data and abnormal context of all associated products can be viewed with one click during subsequent traceability. After the tagging is completed, the edge node will send a batch tagging confirmation receipt to the MES system and update the total number of affected products and the number of vehicles in this traceability simultaneously.

[0053] Furthermore, in the method provided in the application embodiment, the method for determining the K value includes: collecting a historical anomaly inspection dataset; identifying the maximum abnormal vehicle span before and after marking the anomaly based on the historical anomaly inspection dataset; and superimposing a tolerance threshold on the maximum abnormal vehicle span to obtain the K value.

[0054] Specifically, the edge computing node first automatically triggers a K-value calculation task after each anomaly trend determination. This task then sends a data collection request to the MES system's quality data warehouse—an enterprise-level structured database that centrally stores all production quality-related data, including all anomaly event records, re-inspection results, root cause analysis reports, and process parameter change logs since the system went live. The request collects a historical anomaly inspection dataset, which must meet strict screening criteria: the time range is limited to the past 6-12 months; the anomaly and type are completely consistent with the currently triggered anomaly, including dimensional deviations, optical transmittance drift, and batch scratches; the workstations and equipment models are the same, and a complete root cause analysis and case closure process has been completed. Simultaneously, incomplete data, unclear root causes, or isolated anomalies caused by accidental factors, such as sudden power outages or human error, are automatically removed to ensure the homogeneity and validity of the dataset.

[0055] After data collection, the edge computing nodes standardize the dataset and extract key features for each historical anomaly event, namely the precise timestamp of the anomaly trigger, the number of the initially marked anomaly vehicle, the list of the actual affected vehicles confirmed by the final full-item re-inspection (i.e., all vehicles containing non-conforming products, not just the initially marked vehicles), and the flow sequence of the vehicles on the production line. Then, the anomaly vehicle span for each event is calculated, which is the number of consecutive vehicles from the first vehicle containing non-conforming products to the last vehicle containing non-conforming products. This objectively reflects the range of vehicles that the anomaly spreads from its generation to its detection.

[0056] Subsequently, the maximum abnormal vehicle span is extracted from all calculated abnormal vehicle spans. This is the number of vehicles with the widest actual impact in the history of similar abnormal events. This value represents the spread boundary of this type of abnormality in the worst case.

[0057] Finally, a tolerance threshold is superimposed on the maximum abnormal vehicle span. This threshold is a positive integer set based on product accuracy requirements and abnormal risk levels, typically between 1 and 2. It is used to cover statistical sample errors, minor deviations in the time of abnormal occurrence, timing misalignments during vehicle flow, and extreme cases not recorded in historical data. For example, when the maximum abnormal vehicle span is 4, adding a tolerance threshold for one vehicle results in K=5, ensuring that the traceability scope can fully cover all potentially affected vehicles. At the same time, the K value is forcibly constrained to be an integer greater than 1. The calculated K value is automatically synchronized to the parameter configuration library of the MES system. Meanwhile, the edge computing nodes will periodically re-execute the K value calculation process once a month. When a new major abnormal event is closed, an update will also be triggered immediately to ensure that the K value always reflects the latest production process characteristics.

[0058] Furthermore, in the method provided in the application embodiment, obtaining abnormal product identification information and abnormal carrier identification information corresponding to glass devices with abnormal trends and marking them as traceable states, further includes: sending a locking signal to the material transfer controllers of all workstations for the abnormal carrier where the glass device marked as traceable is located, the signal containing the abnormal carrier code of the abnormal carrier; after receiving the locking signal, the transfer controller automatically activates the pneumatic baffle when the abnormal carrier arrives at the diversion port, pushing the abnormal carrier into the isolation buffer.

[0059] Specifically, once the edge computing node completes the traceability status marking of all associated carriers and their internal glass components and confirms data synchronization with the MES system, it will immediately broadcast a highest-priority locking signal to the material transfer controllers of all workstations on the production line via the real-time industrial Ethernet protocol. These controllers are industrial programmable logic controllers deployed in the control cabinets of each workstation, responsible for controlling the conveyor belt speed, carrier diversion direction, and various pneumatic actuators. They are the execution units for production line logistics scheduling. The signal content includes the unique carrier code of all abnormal carriers, the signal validity period (usually set to 24 hours), the target isolation buffer number, and the emergency handling level. This signal is exclusive and will cover all normal production scheduling instructions, ensuring that the interception of abnormal carriers takes precedence over any other logistics operation.

[0060] Subsequently, upon receiving a locking signal, each material handling controller immediately stores all abnormal carrier codes into the carrier arrival status in the locked carrier list in its local non-volatile memory, and continuously monitors the branching point of its workstation. The carrier list refers to the key branching points in the production line used to guide carriers to normal processes, rework areas, or isolation buffer zones, and each workstation is equipped with at least one. When an abnormal carrier carrying traceable products is conveyed to the branching point of any workstation by the conveyor belt, the diffuse reflection photoelectric sensor deployed at the front end of the branching point will detect the carrier's arrival and trigger the industrial barcode reader to scan the metal-etched QR code on the carrier's surface. The barcode reader then transmits the read carrier code to the material handling controller in real time. The conveyor controller performs a millisecond-level comparison with the locally locked carrier list. If the comparison is successful, the material conveyor controller immediately suspends the normal flow logic of the diversion port, cuts off the power to the conveyor belt leading to the subsequent process, and activates the pneumatic baffle, a fast-lifting physical blocking device driven by clean compressed air with a soft silicone contact head at the front end, to smoothly push the abnormal carrier into a specially set isolation buffer zone, which is an independently defined physical isolation area next to the production line, completely isolated from normal production logistics. It is equipped with independent barcode readers, warehouse indicator lights, video surveillance, and anti-static shelves. If the comparison fails, the carrier proceeds to the next process according to the normal flow.

[0061] If an abnormal vehicle has already passed through the current workstation's diversion point, the material transfer controller will immediately forward an emergency interception signal to the controllers of adjacent upstream workstations and all downstream workstations. Simultaneously, it will send a vehicle location warning to the edge computing node, which will then push the warning to the on-duty operator's handheld terminal for manual interception. After the abnormal vehicle is successfully pushed into the isolation buffer, the material transfer controller will automatically record the precise isolation timestamp, isolation workstation number, buffer zone location number, and execution status. This information will be fed back to the edge computing node in real time, and the edge node will then synchronously report it to the MES system to update the physical location and status of the corresponding vehicle. This provides a basis for the accurate scheduling of subsequent re-inspection tasks and the centralized management of abnormal products.

[0062] Furthermore, in the method provided in the application embodiment, while the abnormal vehicle is pushed into the isolation buffer, the isolation time, isolation station number and buffer location number are recorded to generate an isolation event; the isolation event is reported to the MES to trigger the re-inspection task of all glass devices in the isolation buffer, and the normal process flow is entered after receiving the re-inspection qualified feedback signal.

[0063] Specifically, after the material conveying controller controls the pneumatic baffle to smoothly push the abnormal vehicle into the designated compartment of the isolation buffer, the arrival sensor deployed at the bottom of each compartment in the buffer immediately sends a vehicle arrival confirmation signal to the corresponding edge computing node. The arrival sensor is a non-contact photoelectric sensor used to accurately detect whether the vehicle has completely entered the compartment and is stably placed, thereby triggering the isolation event generation process. The edge computing node automatically collects and records three core pieces of information: isolation time, isolation station number (i.e., the unique identification code of the station that performed this isolation operation), and buffer compartment number (i.e., the unique code of each independent storage location in the isolation buffer, in the format of buffer number-row number-column number). At the same time, it automatically associates the previously generated abnormal event ID, abnormal vehicle code, abnormal type, number of affected products, parameter values ​​that triggered the abnormality, and other complete context information to generate a standardized isolation event, which is the data unit for the MES system to perform quality traceability, task scheduling, and responsibility identification. It is stored in structured JSON format and includes 12 mandatory fields such as unique event ID, event type, occurrence time, associated abnormal ID, vehicle information, location information, and execution device information to ensure the integrity and traceability of event information.

[0064] After an isolation event is generated, the edge computing node reports it to the MES system in real time via encrypted MQTT 3.1.1 protocol. The transmission process uses a three-way handshake confirmation mechanism. If the MES does not return a receipt confirmation within 500ms, the edge node will automatically retransmit up to 3 times. When the network is down, the event data is temporarily stored on a local industrial-grade solid-state drive. After the network is restored, the transmission will automatically resume to ensure that the isolation event is not lost. After receiving the isolation event and completing the data verification, the MES system will immediately update the global status of the corresponding carrier and all glass components inside it, changing it from traceability and locking to pending re-inspection. At the same time, based on the anomaly type and product process, the system will also update the global status of the corresponding carrier and all glass components inside it. The system is required to automatically generate re-inspection tasks, which are standardized quality inspection work orders pre-configured in the MES system. The work orders include a list of carriers to be re-inspected, precise location of the compartments, and targeted re-inspection items. For example, if the optical transmittance is abnormal, the focus should be on re-inspecting the full-spectrum transmittance curve of 380-780nm, haze, and refractive index uniformity. If the dimensions are abnormal, the focus should be on re-inspecting the length, thickness, chamfer angle, and hole position accuracy. The system also includes the sampling ratio, acceptance criteria, completion deadline, and designated responsible person. The task is then accurately pushed to the handheld PDA and on-site industrial touch screen of the corresponding quality inspector, and the sound and light prompts of the corresponding compartment in the buffer zone are triggered to remind the quality inspector to handle the issue in a timely manner.

[0065] After receiving the task, the quality inspector scans the QR code on the buffer zone hopper with a handheld PDA to confirm the task receipt, retrieves the abnormal carrier, and sends it to the corresponding measurement station for re-inspection. The measurement equipment automatically binds the re-inspection data with the corresponding product identifier and uploads it to the MES system in real time. The MES system automatically compares the re-inspection data with the pre-stored standard thresholds to generate a preliminary judgment result. After the quality inspector confirms the judgment result, they submit a re-inspection feedback signal on the PDA. For carriers that pass all re-inspections, the MES system automatically clears its traceability status label and the lock signal of the entire station, and sends a release command to the material transfer controller in the isolation buffer zone. The controller activates the pneumatic pusher of the corresponding hopper, pushing the carrier back to the main conveyor belt so that it can continue to enter the subsequent normal process flow. For carriers with non-conforming products in the re-inspection, the MES system will mark the non-conforming products separately as pending scrapping or rework and generate corresponding scrapping or rework tasks. Conforming products will have their lock marks cleared and be released. All re-inspection results will be synchronously written into the full-process quality file of the corresponding product and permanently stored for future reference.

[0066] In summary, the MES-based glass device manufacturing process traceability method provided in this application has the following technical effects: By deploying edge computing nodes at each production and measurement station, local real-time data processing and statistical process control analysis are achieved. Combined with the integrated technology of dynamic binding of product and carrier dual codes, dynamic determination of traceability extension K value based on historical anomaly inspection dataset, and automatic locking and isolation of abnormal carriers by the material transfer controller of the entire station, standardized traceability data packages and re-inspection tasks are automatically generated. This realizes the transformation from passive post-event traceability to pre-event anomaly warning and automatic in-event interception, improving the efficiency of anomaly handling and the accuracy of traceability, and reducing production losses and quality control costs.

[0067] Example 2 is based on the same inventive concept as the MES-based glass device manufacturing process traceability method in the previous examples, such as... Figure 2 As shown in the embodiment of this application, a glass device manufacturing process traceability system based on MES is provided. The system includes: The connection module 11 is used to deploy edge computing nodes at each production station and measurement station of the glass device, and connect the barcode reading device and measurement device of the corresponding station to the edge computing node; the acquisition module 12 is used to read the QR code on the glass device and the QR code on the carrier through the barcode reading device, obtain product identification information and carrier identification information, and upload them to the corresponding edge computing node; the measurement module 13 is used to obtain the measurement data of the glass device at the corresponding station through the measurement device, including size data, optical data, and appearance data, and upload them to the corresponding edge computing node; the judgment module 14 is used for each edge computing node to compare the acquired measurement data with the standard threshold of the corresponding process stored in the MES system in real time to determine whether there is an abnormal trend; the traceability module 15 is used for the edge computing node to automatically trigger the traceability process based on the product identification information and the carrier identification information if an abnormal trend is determined to exist.

[0068] Furthermore, the traceability module 15 is also used to perform the following steps: obtain abnormal product identification information and abnormal carrier identification information corresponding to glass devices with abnormal trends, mark them as traceable, and upload them to MES; record the complete process data of N glass devices before and after the time window of the abnormal trend, generate abnormal warning information, and push it to the operator terminal of MES for re-inspection reminder.

[0069] Furthermore, the judgment module 14 is also used to perform the following steps: each edge computing section pre-caches the standard threshold range and historical statistical process control upper limit corresponding to each workstation from the MES; then, it calculates the deviation of the measurement data of each glass device from the standard threshold range and the historical statistical process control upper limit; if any deviation exceeds the preset coefficient, it is judged that there is an abnormal trend.

[0070] Furthermore, the traceability module 15 is also used to perform the following steps: requesting historical binding records of abnormal vehicles from the MES; identifying K vehicles before and after the occurrence of the abnormality based on the historical binding records, and marking all glass devices in the K vehicles as traceable, where K is an integer greater than 1.

[0071] Furthermore, the traceability module 15 is also used to perform the following steps: collecting a historical anomaly inspection dataset; identifying the maximum abnormal vehicle span before and after marking the anomaly based on the historical anomaly inspection dataset; and superimposing a tolerance threshold on the maximum abnormal vehicle span to obtain the K value.

[0072] Furthermore, the traceability module 15 is also used to perform the following steps: for the abnormal carrier where the glass device marked as requiring traceability is located, a locking signal is sent to the material transfer controller of all workstations, the signal containing the abnormal carrier code of the abnormal carrier; after receiving the locking signal, the transfer controller automatically activates the pneumatic baffle when the abnormal carrier arrives at the diversion port, pushing the abnormal carrier into the isolation buffer.

[0073] Furthermore, the traceability module 15 is also used to perform the following steps: while the abnormal vehicle is pushed into the isolation buffer, record the isolation time, isolation station number and buffer location number, and generate an isolation event; report the isolation event to MES, trigger the re-inspection task of all glass components in the isolation buffer, and enter the normal process flow after receiving the re-inspection qualified feedback signal.

[0074] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for tracing the entire glass device manufacturing process based on MES, characterized in that, include: Edge computing nodes are deployed at each production station and measurement station of glass components, and the corresponding code reading devices and measurement devices at each station are connected to the edge computing nodes. The QR code on the glass device and the QR code on the carrier are read by the QR code reading device to obtain product identification information and carrier identification information, and then uploaded to the corresponding edge computing node; The measurement equipment acquires measurement data of the glass device at the corresponding workstation, including dimensional data, optical data, and appearance data, and uploads it to the corresponding edge computing node. Each edge computing node will compare the acquired measurement data with the standard thresholds of the corresponding process stored in the MES system in real time to determine whether there are any abnormal trends. If an abnormal trend is detected, the edge computing node will automatically trigger a traceability process based on the product identification information and the vehicle identification information.

2. The MES-based glass device manufacturing process traceability method as described in claim 1, characterized in that, The edge computing node automatically triggers a traceability process based on the product identification information and the vehicle identification information, including: Obtain the abnormal product identification information and abnormal carrier identification information corresponding to glass devices with abnormal trends, mark them as requiring traceability, and upload them to MES; Record the complete process data of N glass devices before and after the time window in which the abnormal trend occurs, generate abnormal early warning information, and push it to the operator terminal of MES for re-inspection reminder.

3. The MES-based full-process traceability method for glass device production as described in claim 1, characterized in that, Each edge computing node compares the acquired measurement data with the standard thresholds for the corresponding processes stored in the MES system in real time to determine whether there are any abnormal trends, including: Each edge computing node pre-caches the standard threshold range and historical statistical process control upper limit corresponding to each workstation from the MES; Then, the deviation of the measurement data of each glass device from the standard threshold range and the upper limit of the historical statistical process control is calculated. If any deviation exceeds the preset coefficient, it is determined that there is an abnormal trend.

4. The MES-based glass device manufacturing process traceability method as described in claim 2, characterized in that, Obtain the abnormal product identification information and abnormal carrier identification information corresponding to glass devices exhibiting abnormal trends, mark them as requiring traceability, and also include: Request historical binding records of abnormal vehicles from MES; Based on the historical binding records, identify K carriers before and after the anomaly occurred, and mark all glass devices in the K carriers as being in a traceable state, where K is an integer greater than 1.

5. The MES-based glass device manufacturing process traceability method as described in claim 4, characterized in that, Methods for determining the K value include: Collect historical anomaly detection datasets; Based on the historical anomaly detection dataset, identify the maximum anomalous vehicle span before and after marking the anomaly; The maximum abnormal vehicle span is superimposed with a tolerance threshold and then used as the K value.

6. The MES-based method for full-process traceability of glass device production as described in claim 2, characterized in that, Obtain the abnormal product identification information and abnormal carrier identification information corresponding to glass devices exhibiting abnormal trends, mark them as requiring traceability, and also include: For the abnormal carrier containing the glass device marked as requiring traceability, a locking signal is sent to the material transfer controller of all workstations. The signal contains the abnormal carrier code of the abnormal carrier. After receiving the locking signal, the transmission controller automatically activates the pneumatic baffle when the abnormal vehicle arrives at the diversion port, pushing the abnormal vehicle into the isolation buffer zone.

7. The MES-based glass device manufacturing process traceability method as described in claim 6, characterized in that, include: As the abnormal vehicle is pushed into the isolation buffer, the isolation time, isolation station number, and buffer zone location number are recorded, and an isolation event is generated. The isolation event is reported to MES, triggering a re-inspection task for all glass components within the isolation buffer. After receiving a re-inspection pass feedback signal, the normal process flow is resumed.

8. A MES-based end-to-end traceability system for glass device manufacturing, characterized in that: The system is used to implement the MES-based full-process traceability method for glass device production as described in any one of claims 1 to 7, the system comprising: The connection module is used to deploy edge computing nodes in each production station and each measurement station of glass devices, and to connect the code reading equipment and measurement equipment of the corresponding stations to the edge computing nodes. The data acquisition module is used to read the QR codes on the glass device and the carrier through the code reading device, obtain product identification information and carrier identification information, and upload them to the corresponding edge computing node; The measurement module is used to acquire measurement data of the glass device under the corresponding workstation through the measurement equipment, including size data, optical data, and appearance data, and upload it to the corresponding edge computing node; The judgment module is used to compare the measurement data acquired by each edge computing node with the standard threshold of the corresponding process stored in the MES system in real time to determine whether there is an abnormal trend. The traceability module is used to automatically trigger the traceability process based on the product identification information and the vehicle identification information if an abnormal trend is detected.

Images