Process path optimization system for flexible OLED manufacturing process

Abstract

The invention relates to the field of industrial data processing and management, and discloses a process path optimization system for a flexible OLED (Organic Light Emitting Diode) manufacturing process, which comprises a module for acquiring commercial economic attributes, physical states and available path options of work orders of work-in-process, a fusion module for generating economic attribute dynamic portraits of the work orders of work-in-process for the path options, and a processing module for processing the dynamic portraits of the work orders of work-in-process. The decision module is used for caching portraits and subscribing key physical resource state change events, and when it is diagnosed that the economic attribute of the original optimal process path is inferior to other available process path options in the cache due to the occurrence of the events, the decision module is used for selecting the optimal path according to a preset optimization strategy. According to the method, the dynamic guarding mechanism with decision timeliness is established, the decision can be automatically cancelled and re-optimization can be triggered when the original optimal decision fails due to sudden change of the physical environment, and the management risk of high-value work orders is avoided.

Description

TECHNICAL FIELD

[0001] The present application relates to a process path optimization system for flexible OLED manufacturing process, belonging to the technical field of industrial data processing and management. BACKGROUND

[0002] In such complex industrial activities as flexible OLED manufacturing, the manufacturing execution system (MES) is the core tool for production process management and supervision, which tracks the physical state of work-in-process (WIP) and schedules physical resources according to the preset process path to ensure the orderly execution of the production process at the physical level. However, these industrial data processing systems usually only include physical execution elements in their architecture definition, i.e. their definition of process path is mainly based on physical processes, and their decision logic (such as first-in-first-out) is separated from the business economic attributes of the work order, and the production line is just a "machine" that rigidly executes the process program. In the application scenario of flexible OLED manufacturing with high value and high timeliness, when high-profit orders and low-profit orders conflict on key bottleneck equipment, this decision mechanism based on pure physical state may lead to the automatic execution of a physically reasonable but economically costly path selection by the system, and management needs to rely on manual intervention outside the system to try to correct the inconsistency between physical execution and management goals.

[0003] To alleviate this problem, the industry has tried to build more complex optimization algorithms on top of existing systems, but these attempts have difficulty solving a fundamental problem, i.e. the lack of economic dimension in the data foundation of industrial data processing systems. As long as business economic data and physical path data are still separated at the system architecture level, the upper-layer algorithm cannot obtain and integrate all the economic elements required for decision-making in real time at the decision point, resulting in a lack of real-time economic rationality in decision-making. Even in some solutions that try to build production topology maps for scheduling optimization, their logic for responding to disturbance events is often limited to passive rescheduling rather than actively guarding the real-time economic rationality of the decision, and in the scenario of flexible OLED manufacturing with high-value work order intensive flow, this rescheduling logic response lag cannot avoid the economic risks caused by the failure of the original decision.

[0004] Therefore, how to provide a data processing system that can dynamically fuse the business economic attributes of the work order with the physical process path options from the data architecture level, and automatically execute management strategies based on the fused economic rationality data rather than pure physical rules to optimize the process path, has become a technical problem to be solved by the present application. SUMMARY

[0005] The application provides a process path optimization system for a flexible OLED manufacturing process, which is mainly used to solve the problem that the existing industrial data processing system cannot directly correlate the business economic attributes of work orders with their physical paths at the data level due to the lack of economic dimensions in the data architecture, so that the path decision is mainly based on physical rules rather than the automatic execution of management strategies combined with economic attributes.

[0006] To achieve the above object, the application provides a process path optimization system for a flexible OLED manufacturing process, which comprises:

[0007] a business attribute extraction module configured to obtain the business economic attributes of work orders in process from an enterprise resource planning system;

[0008] a physical state and path extraction module configured to obtain the current physical state of work orders in process and one or more available process path options from a manufacturing execution system;

[0009] an economic attribute path fusion and portrait generation module connected with the business attribute extraction module and the physical state and path extraction module, and configured to generate an economic attribute dynamic portrait of work orders in process for one or more available process path options;

[0010] a path optimization decision module configured to sort one or more economic attribute dynamic portraits of work orders in process according to a preset optimization strategy, and select an optimal process path;

[0011] a work order in process economic attribute dynamic portrait cache and subscription module configured to cache the economic attribute dynamic portraits of work orders in process generated by the economic attribute path fusion and portrait generation module, and subscribe to the state change events of key physical resources relied on by the optimal process path to the physical state and path extraction module;

[0012] a path economic mutation diagnosis module configured to send a decision invalidation instruction to the manufacturing execution system and trigger the path optimization decision module to re-execute the optimization decision based on the updated economic attribute dynamic portrait of work orders in process when receiving the state change event and diagnosing that the economic attributes of the optimal process path have deteriorated than other available process path options in the cache.

[0013] Preferably, the business economic attributes comprise at least one of order value, order time sensitivity, and penalty risk; the economic attribute path fusion and portrait generation module is further configured to access a process path attribute library to obtain physical cost attributes corresponding to each available process path option and fuse the physical cost attributes into the work-in-process order economic attribute dynamic portrait; the physical cost attributes comprise at least one of process estimated duration, process standard cost, or process standard yield; and the process path attribute library is further configured to store a standard unit processing duration.

[0014] Preferably, the system further comprises a dynamic queue monitoring module configured to respond to the physical state and path extraction module and obtain real-time queued work-in-process order quantities of key physical resources relied on by the one or more available process path options from the manufacturing execution system; wherein the economic attribute path fusion and portrait generation module is further configured to: based on the static physical cost corresponding to the physical cost attributes obtained from the process path attribute library, the real-time queued work-in-process order quantities, and the standard unit processing duration obtained from the process path attribute library, calculate a dynamic physical cost attribute; and fuse the dynamic physical cost attribute as the final physical cost attribute into the work-in-process order economic attribute dynamic portrait.

[0015] Preferably, the static physical cost is a static estimated duration and the dynamic physical cost attribute is a dynamic estimated duration; the economic attribute path fusion and portrait generation module is configured to perform the following calculation: wherein, is the dynamic estimated duration, is the static estimated duration, is the real-time queued work-in-process order quantity, is the standard unit processing duration.

[0016] Preferably, the system further comprises a factory macro state diagnosis module configured to diagnose a current macro operation state of the factory based on aggregated data obtained from the manufacturing execution system; wherein the path optimization decision module is further configured to: store a strategy library configured to store a correspondence between the macro operation state and a plurality of optimization strategies; and automatically select and activate a corresponding optimization strategy from the strategy library according to the diagnosed macro operation state to sort the work-in-process order economic attribute dynamic portrait.

[0017] Preferably, the aggregated data is a total work-in-process order quantity of the factory or an average queue length of key bottleneck equipment; the macro operation state comprises a normal state and a congestion state; and the strategy library is configured to map the normal state to a profit priority optimization strategy based on order value sorting and map the congestion state to a throughput priority optimization strategy based on process estimated duration sorting.

[0018] Preferably, the physical state and path extraction module is further configured to obtain one or more physical recipes respectively associated with the one or more available process path options; the economic attribute path fusion and portrait generation module is further configured to, for each physical recipe, respectively perform data-level fusion with the business economic attributes and generate a corresponding work-in-process order economic attribute dynamic portrait for each physical recipe; and the plurality of physical recipe based work-in-process order economic attribute dynamic portraits are provided to the path optimization decision module for sorting based on the optimization strategy and determining an optimal process path and an optimal physical recipe corresponding thereto.

[0019] Preferably, the application further comprises a management visualization module configured to: obtain the work-in-process order economic attribute dynamic portraits generated by the economic attribute path fusion and portrait generation module for the plurality of work-in-process orders; access a visualization rule library mapping the business economic attributes in the work-in-process order economic attribute dynamic portraits to visual elements; and based on the visualization rule library, generate an economic heat map reflecting real-time economic attribute distribution of the plurality of work-in-process orders on a display interface facing the management personnel.

[0020] Preferably, the visualization rule library is configured to: map the work-in-process orders with high order value and high time sensitivity in the business economic attributes to a first visual style; and map the work-in-process orders with low order value in the business economic attributes to a second visual style; the first visual style and the second visual style are distinguishable in color or dynamic effect.

[0021] Preferably, the application further comprises a decision analysis module configured to: asynchronously obtain and analyze the decision logs generated by the path optimization decision module during operation; based on the decision logs, perform statistical aggregation analysis of economic decision influence on the available process path options and the physical resources relied thereon to identify strategic bottlenecks causing systematic economic path compromise; and a management insight report module configured to generate an insight report for strategic decision-making by the management personnel based on the results of the statistical aggregation analysis.

[0022] Compared with the prior art, the application has the following beneficial effects:

[0023] 1. In the industrial production of flexible OLED manufacturing, from the data architecture level, the business economic attributes of the work order are dynamically data fused with the physical process path options, and based on the fused economic rationality data, rather than pure physical rules, the management strategy is automatically executed to optimize the process path, the driving force of production decision is improved from "physically feasible" to "economically optimal", thereby realizing the maximization of global value; The production line is no longer a "machine" that rigidly executes the process program, but an "economic rationality" that can perceive cost, understand value, and actively pursue profit maximization, which builds a sustainable competitive advantage in the current global flexible OLED high-end manufacturing industry.

[0024] 2. Through a fusion and portrait generation module, the business economic attributes (such as order value or time sensitivity) obtained from the enterprise resource planning system are data bound with the available process path options obtained from the manufacturing execution system. The key to this process is that it generates a corresponding work-in-process economic attribute dynamic portrait for each available path option, and even for each specific physical formula associated with the path. This allows the subsequent path optimization decision module to automatically compare these portraits based on a unified preset optimization strategy, such as the delivery date priority or profit priority rule set by management, and automatically penetrate and sink the decision granularity from macro path selection to the specific execution level. This approach enables the automatic consistent implementation of high-level management intentions at the smallest unit of physical execution, avoiding the loss of economic optimization potential caused by the mismatch between management strategies and physical execution granularity.

[0025] 3. When generating the dynamic economic attribute portrait, the fusion logic is not based on a static process attribute library. It also obtains the real-time queuing number of key physical resources relied on by the available process path through a dynamic queue monitoring module, and calculates a dynamic physical cost attribute (such as estimated time length including waiting time) into the portrait, which ensures that the physical cost relied on by the path optimization decision can reflect the current congestion situation. On this basis, the system also introduces a decision timeliness supervision mechanism: a WEA portrait cache and subscription module will continuously subscribe to the state change events of the key resources of the selected path after the decision is made; Once a sudden event such as failure is received, the path economic mutation diagnosis module will immediately compare the economic of the alternative paths in the cache, and send a decision invalidation instruction and trigger a re-decision when it is determined that the original decision is no longer optimal. The combination of this dynamic cost accounting and post-decision guardianship mechanism enables the management system not only to make economic rationality judgments based on the current working conditions, but also to actively avoid the management risks of high-value work orders caused by physical environment mutations.

[0026] 4、On the basis of the path optimization decision module, a factory macro state diagnosis module and a strategy library are also configured, the diagnosis module is configured to obtain and diagnose the overall macro running state of the factory at a low frequency (for example, judging as a normal or congestion state based on the total work-in-process quantity), and the strategy library stores the optimization strategies corresponding to different macro states preset by the manager (for example, the profit priority strategy corresponds to the normal state, and the flux priority or the shortest time strategy corresponds to the congestion state), when the macro state changes, the diagnosis module triggers the decision module to automatically switch the currently activated optimization strategy, and the hierarchical decoupling architecture enables the micro work order path selection (tactical execution) at the bottom layer to automatically align with the overall factory operation target (strategic switching) at the high level, so that the system can automatically switch from guaranteeing individual profit to guaranteeing global flux when facing different working conditions (such as global congestion); at the same time, the application also uses the work-in-process work order economic attribute dynamic portrait (WEA dynamic portrait) generated by the fusion and portrait generation module in real time, through a management visualization module, the system obtains these portrait data in real time, and according to a preset visualization rule library, for example, defining high-value and time-sensitive work orders as a specific visual style, the commercial economic attributes of the work orders are displayed in real time on the display interface for the management personnel, which superimposes a direct economic heat map on the physical layout of the factory, which translates the internal economic data originally used for machine automatic decision-making into management information directly perceived by human managers, so that the decision basis of the management personnel when performing manual intervention or supervision can be consistent with the automatic decision logic of the system, and is based on the same economic rationality, thereby facilitating the cooperation between automatic management and manual supervision. BRIEF DESCRIPTION OF DRAWINGS

[0027] Fig. 1 It is a functional architecture and data flow diagram of the process path optimization system of the application.

[0028] Fig. 2 It is a system simulation performance comparison chart under normal state of the application.

[0029] Fig. 3 It is a dynamic estimated time calculation interactive timing diagram. DETAILED DESCRIPTION

[0030] In order to facilitate those skilled in the art to understand the application, the application will be further described in detail in combination with the drawings and specific embodiments; it should be noted that the embodiments and technical features in the embodiments can be combined with each other without conflict, and the specific embodiments described herein are only used to explain the application, and not to limit the protection scope of the application.

[0031] The present application provides a process path optimization system for flexible OLED manufacturing process, which is a data processing architecture in the field of industrial data processing and management, and runs on a computing facility such as a standard server; the system logically includes a business attribute extraction module, a physical state and path extraction module, an economic attribute path fusion and portrait generation module, and a path optimization decision module; the core data flow is that the system obtains data from the external enterprise resource planning (ERP) system and manufacturing execution system (MES) in parallel through the business attribute extraction module and the physical state and path extraction module, respectively, performs data-level fusion processing by the economic attribute path fusion and portrait generation module to form an intermediate data object carrying economic and physical dual attributes, and finally analyzes the data object according to the preset management rules (i.e. the preset optimization strategy) by the path optimization decision module and outputs the decision instruction to realize the automatic alignment of industrial data processing and business management goals; in the complex industrial data processing activities such as flexible OLED manufacturing, the physical state of a work order in process and the business value it carries are often separated, in order to solve the information fragmentation problem in the management level, the business attribute extraction module is configured as a standardized data interface service, such as a RESTful API gateway, which is configured to actively connect to the enterprise resource planning (ERP) system; when a work order in process is created or reaches a key decision point, the module is triggered to obtain and analyze the static business economic attributes corresponding to the work order in process from the ERP system through a query request carrying the unique identifier of the work order in process; in a preferred embodiment, these business economic attributes specifically include, but are not limited to: order value, such as a profit rate level quantified as high, medium and low; order time sensitivity, such as the number of remaining days calculated by (promised delivery date-current date) or an emergency degree score; and penalty risk, such as a Boolean value or a risk level score indicating whether there is a high penalty clause; the module provides the subsequent fusion module with the obtained economic attribute data formatted as a structured data object.

[0032] The physical status and path extraction module is also configured as a standardized data interface for real-time connection to the Manufacturing Execution System (MES). When a work-in-process order reaches a process bifurcation or decision point, this module is triggered by an event in the MES. It actively queries the MES to obtain the current physical status of the work-in-process order, such as its specific process or equipment number. More importantly, this module is configured to obtain one or more available process path options, i.e., a list of physical paths that the work-in-process order can be dispatched to next. To achieve more refined management decisions, in a preferred embodiment, the physical status and path extraction module is further configured to obtain one or more physical statuses associated with each available process path option. Formulas; for example, path A might be associated with two options: [Formula R1: High speed, 98% yield] and [Formula R2: Low speed, 99.5% yield]. This module provides this physical state, the list of available paths, and the corresponding formula list to the fusion module. The economic attribute path fusion and profile generation module is the core module for binding management data and industrial data. As a data fusion engine, it is configured to connect the business attribute extraction module and the physical state and path extraction module. To obtain the inherent physical cost of a path, this module is also configured to access a process path attribute library, which is a standard database that stores the static physical cost attributes corresponding to each process path option (or physical formula), such as the estimated process time (i.e., ...). ), process standard cost or process standard yield, and standard unit processing time used for dynamic cost calculation ( During system operation, the fusion module receives commercial economic attributes (e.g., {order value: high, penalty risk: high}) from the commercial attribute extraction module and a list of available paths / recipes (e.g., [{path A, recipe R1}, {path A, recipe R2}]) from the physical state and path extraction module. Its core action is to perform data fusion once for each available option (i.e., each physical recipe) provided by the physical state and path extraction module, generating a corresponding dynamic profile of the economic attributes of the work-in-process order for that option. This dynamic profile of the economic attributes of the work-in-process order is a structured data object that contains at least commercial economic attributes from the commercial attribute extraction module and physical cost attributes from the process path attribute library.

[0033] In a preferred embodiment, to ensure that the physical cost of the decision-making basis reflects the real-time operating conditions, the system also includes a dynamic queue monitoring module. This module is triggered when the fusion module starts and is used to obtain the real-time queued work orders of the key physical resources on which the path options depend from the MES. At this point, the economic attribute path fusion and profile generation module is further configured to calculate a dynamic physical cost attribute, such as dynamically estimated duration. Its calculation procedure is determined as follows: Here's a numerical example: If path A, R1 in the process path attribute library... For 1.0 hour, its key resources The time is 0.5 hours, and the dynamic queue monitoring module can detect it. If there are 3, then the module calculates... for Hours; the module then assigns this dynamic physical cost attribute ( The system integrates the work-in-process (WIP) economic attributes (e.g., path ID: A-R1, order value: high, dynamic estimated duration: 2.5h, yield: 98%) with the business economic attributes ({order value: high}) to generate a dynamic profile of the WIP economic attributes for that option, such as: {path ID: A-R1, order value: high, dynamic estimated duration: 2.5h, yield: 98%}. This module repeats this process for all available options (including {path A, R2}, etc.) and finally outputs a list containing multiple dynamic profiles of the WIP economic attributes based on physical recipes to the path optimization decision module. The path optimization decision module is the final execution module for management intent. The system is configured to receive a dynamic profile list of economic attributes of work-in-process orders generated by the fusion module. To enable micro-level work order decisions to automatically align with the factory's operational goals, the system preferably includes a factory macro-level status diagnosis module. This module is configured to retrieve aggregated data from the MES (Manufacturing Execution System) at low frequencies (e.g., every 10 minutes), such as the total number of work-in-process orders or the average queue length of key bottleneck equipment, and diagnose the current macro-level status of the factory based on a preset threshold (e.g., a total number of work-in-process orders > 5000 indicates a congestion state). The operational status is monitored. Correspondingly, the path optimization decision module is further configured to store a strategy library, pre-configured by administrators, that stores the correspondence between macro-level operational status and multiple optimization strategies. For example, a specific strategy library mapping relationship can be labeled as: {Normal State -> Profit-First Optimization Strategy}, {Congestion State -> Throughput-First Optimization Strategy}. When the factory macro-level status diagnosis module detects a change to congestion, it triggers the path optimization decision module to automatically activate the throughput-first optimization strategy. Subsequently, the core action of the path optimization decision module is to sort the received list of dynamic profiles of work-in-process economic attributes based on the currently activated optimization strategy (i.e., the throughput-first optimization strategy, whose rule can be determined as ascending sorting by the dynamically estimated duration in the dynamic profile of the work-in-process economic attributes), and select the option corresponding to the first profile after sorting (i.e., the profile with the shortest duration), determining it as an optimal process path and its corresponding optimal physical formula. Finally, the module issues an execution command to the MES system via API.

[0034] To address the management risk that the optimal process path output by the path optimization decision module might fail due to sudden equipment failure before its physical execution, the system of this invention also includes a dynamic monitoring mechanism for decision timeliness. This mechanism specifically includes a dynamic profile caching and subscription module for the economic attributes of work-in-process orders. This module is configured to cache all relevant dynamic profiles of the economic attributes of the work-in-process order generated by the fusion module (i.e., including the selected optimal profile and all alternative profiles) after the path optimization decision module makes a decision. Simultaneously, this dynamic profile caching and subscription module is configured to subscribe to the physical state and path extraction module (i.e., the MES interface) for state change events of key physical resources (e.g., the Evaporator-03 equipment) upon which the optimal process path depends. The system also includes a path economic mutation diagnosis module, which is a passively triggered rule engine. When the subscribed state change event is received (before the actual arrival of the work-in-process order), the system... When (for example, when Evaporator-03 becomes faulty), the path economy mutation diagnosis module immediately starts. It updates the dynamic profile of the economic attributes of the work-in-process order for the optimal path, for example, modifying its dynamic estimated duration to a maximum value or invalidating it. Subsequently, the core procedure of the path economy mutation diagnosis module is to compare the economic attributes of this invalidated optimal profile with those of other available process path options in the cache (such as the profile of path B). When it is diagnosed that the economic attributes of the optimal process path are worse than those of other available process path options in the cache, the path economy mutation diagnosis module immediately performs two operations: First, it sends a decision invalidation instruction to the Manufacturing Execution System (MES) to stop the physical movement of the work-in-process order; Second, it triggers the path optimization decision module, forcing it to re-execute the optimization decision process based on the updated list of dynamic profiles of the economic attributes of the work-in-process order (where path A is invalid), thereby automatically selecting the currently most economically optimal path for the work-in-process order.

[0035] To further enhance the use of backend decision data for managerial supervision and intervention, a preferred embodiment of this system also includes a management visualization module. This module is configured to acquire in real-time dynamic profiles of the economic attributes of multiple work-in-process (WIP) work orders generated by the economic attribute path fusion and profile generation module. This module is also configured to access a preset visualization rule base, defined by managers, to map the commercial economic attributes in the dynamic profiles of WIP work orders to specific visual elements. For example, a specific rule might be defined as mapping WIP work orders with high order value and high time sensitivity to a red, flashing first visual style, and mapping WIP work orders with low order value to a green, static second visual style. Based on this visualization rule base, the management visualization module replaces physical location icons with these dynamic visual styles on a manager-facing display interface (e.g., a factory layout dashboard), thereby generating an economic heatmap that intuitively reflects the real-time distribution of economic attributes of multiple WIP work orders, assisting managers in quickly identifying value clusters and risk areas. Finally... To achieve long-term insights and closed-loop management, this system, in a preferred embodiment, also includes a decision analysis module. This module is configured as an asynchronous, non-real-time background data processing service to automatically acquire and analyze decision logs generated by the path optimization decision module during long-term operation during off-peak periods. These decision logs record the work-in-process (WIP) orders, selected paths, abandoned paths, and the dynamic profile of the economic attributes of the WIP orders upon which each decision was based. The core data processing logic of this decision analysis module is to perform statistical aggregation analysis on the economic impact of these decision logs. For example, it is configured to statistically analyze which physical resource (e.g., Evaporator-02) appears most frequently as the low-cost path to be abandoned in all high-value WIP order decisions, thereby identifying strategic bottlenecks leading to systemic economic path compromises. The system also includes a management insight report module, configured to generate a management insight report based on the results of the statistical aggregation analysis. This report reveals the economic bottlenecks, rather than physical bottlenecks, of the factory to managers, providing objective data support for long-term management decisions such as capital investment and maintenance priority ranking.

[0036] Example 1: Application of the system of the present invention in the core bottleneck process of flexible OLED manufacturing, namely the organic evaporation process. In flexible OLED manufacturing, different models of products, such as screens corresponding to different mobile phone models, require the use of different fine metal masks (FMM) on the evaporation equipment to deposit organic light-emitting materials. The evaporation equipment is the most expensive and capacity-limited bottleneck resource in the production line. Changing the mask to produce different models is a physically extremely time-consuming process, which may require 2-3 hours of cleaning, calibration, and vacuuming time. This constitutes a unique and complex process path decision point in OLED manufacturing: when a high-value work order (such as WIP-101) arrives, the system may face a process fork. Path A (pointing to EVAP-03): may be a currently idle ( =0), but for devices with mismatched mask models, selecting this path will result in a statically estimated duration ( This will primarily involve high changeover times; Path B (pointing to EVAP-04): may be a mask model match, but is currently processing other work orders. For devices with a value of 1, selecting this path will result in a dynamically estimated duration ( ). The main component will be queuing time. Traditional MES systems usually only look at whether the equipment is idle (physical status), which may lead to the wrong selection of path A. This results in high-value work orders (WIP-101) incurring huge time costs for model replacement, while low-value work orders (WIP-202) occupy the model-matching path B. The system of this invention is designed to solve this specific decision dilemma that is physically reasonable but economically costly.

[0037] Before the vapor deposition process in flexible OLED manufacturing, a high-value, time-sensitive work-in-process (WIP) order WIP-101 with a high penalty risk and a low-value, flexible-delivery-time WIP-202 arrive at a process bifurcation almost simultaneously. This bifurcation offers two available process path options: path A (pointing to critical physical resource EVAP-03) and path B (pointing to critical physical resource EVAP-04). At this point, the process path optimization system starts running. Its business attribute extraction module obtains the respective business economic attributes of WIP-101 and WIP-202 from the ERP system. Simultaneously, the physical status and path extraction module obtains their current physical status and the two available process path options, path A and path B, from the MES system. At this time, the factory macro-state diagnosis module has diagnosed the current state as normal based on the total number of work-in-process orders in the factory and has triggered the path optimization decision module to activate the profit-first optimization strategy. Immediately afterwards, the economic attribute path fusion and profile generation module starts. It first calls the dynamic queue monitoring module to instantly query and find that EVAP-03 is currently not in the queue ( ), while EVAP-04 has 1 queue ( The fusion module retrieves path A from the process path attribute library. It takes 2.0 hours for path B. It takes 2.5 hours for both routes. All times are 1.0 hour, after which the module executes. The calculation procedure yields the dynamic estimated duration of path A. The dynamic estimated duration for path B is 2.0 hours. The estimated time is 3.5 hours. The fusion module then generates dynamic profiles of the economic attributes of work-in-process orders pointing to path A and path B for WIP-101 and WIP-202 respectively, and submits the profile list to the path optimization decision module. Based on the currently activated profit-first optimization strategy, the path optimization decision module first filters out the profile of WIP-101 with the high order value attribute. Then, among the two options of WIP-101 (path A: 2.0 hours; path B: 3.5 hours), it selects path A with the shortest dynamically estimated time as the optimal process path and issues an instruction to the MES system for WIP-101 to execute path A. At the same time as this decision is issued, the dynamic profile caching and subscription module for the economic attributes of work-in-process orders automatically caches all dynamic profiles of the economic attributes of work-in-process orders of WIP-101 (including the selected path A profile and the rejected path B profile), and immediately subscribes to the physical state and path extraction module for the status change event of the key physical resource EVAP-03 on which path A depends.

[0038] While WIP-101 was being physically transferred to EVAP-03, EVAP-03 suddenly experienced a physical failure, changing its status from idle to faulty. This status change event was immediately received by the work-in-process work order economic attribute dynamic profile caching and subscription module. This event immediately triggered the path economic mutation diagnosis module, which retrieved the cached WIP-101 profile data and dynamically estimated the duration of path A profile. The update is either invalid or a maximum value. In the subsequent real-time comparison, the diagnostic module determines that the economic attribute (invalid) of path A is worse than that of path B in the cache (dynamically estimated duration of 3.5 hours). Accordingly, the path economic change diagnostic module immediately sends a decision invalidation instruction to the MES system, stopping the physical movement of WIP-101 to EVAP-03, and simultaneously triggers the path optimization decision module to re-execute the optimization decision based on the updated dynamic profile list of economic attributes of work-in-process orders (where path A is invalid). After receiving the re-trigger instruction, the path optimization decision module applies the profit-first optimization strategy again, selects from the remaining available options of WIP-101 (i.e., path B), and finally determines path B as the new optimal process path, and issues a new instruction to MES for WIP-101 to execute path B. WIP-101 is then automatically reassigned to the functional EVAP-04, thereby avoiding the management risk of high-value work-in-process orders being assigned to invalid resources due to physical environment changes, while low-value WIP-202 continues to remain in the queue at the decision point during this period.

[0039] Example 2: A test platform based on discrete event simulation was configured to objectively verify the process path optimization system of the present invention, particularly the interaction and decision-making effectiveness between its economic attribute path fusion and profile generation module, factory macro-state diagnosis module, and strategy library. The platform was configured to simulate a key process in a flexible OLED manufacturing process, which includes three parallel-processable physical resources (equipment A, B, and C). A standardized work-in-process (WIP) order input stream was set, containing 1000 WIP orders whose commercial economic attributes were pre-set as a mixture of high-value (20%) and low-value (80%), and each WIP order was assigned a committed delivery date. The experiment included two test groups: a control group configured to use a commercially insensitive control system employed by conventional manufacturing execution systems in the art. The invention employs a First-In-First-Out (FIFO) decision logic based on economic attributes; and a prototype of the invention configured to deploy a process path optimization system. The experiment runs the decision logic of the control group and the prototype under two different factory load conditions, recording key operational performance indicators. In the first experiment, the input rate of work-in-process orders on the simulation platform is set to maintain the average load rate of key physical resources at 75% to simulate a normal state. The control group system strictly follows the physical order (FIFO) of work-in-process orders arriving at the decision point for path allocation. The prototype system's factory macro-state diagnosis module diagnoses the current state as normal based on a 75% load rate and automatically triggers the path optimization decision module to activate a profit-first optimization strategy from the strategy library. Performance data from both systems after processing all 1000 work-in-process orders are recorded in Table 1.

[0040] Table 1: Performance comparison data under normal conditions.

[0041]

[0042] Referring to the data in Table 1, under normal conditions, the FIFO strategy of the control group resulted in frequent delays for high-value work-in-process (WIP) orders because they were prioritized after low-value WIP orders, with an on-time delivery rate of only 68.5%. In contrast, the sample group of this invention, through its economic attribute path fusion and profile generation module, generated a dynamic profile of the economic attributes of each WIP order for its available paths. The path optimization decision module then sorted these profiles according to a profit-first optimization strategy, giving high-value WIP orders decision priority. This increased the on-time delivery rate to 93.2%, and reduced the average cycle time of high-value WIP orders from 8.4 hours to 5.2 hours. The economic value of output increased by approximately 22.3%; in the second group of experiments, the input rate of work-in-process orders on the simulation platform was increased, raising the average load rate of key physical resources to 95% to simulate congestion; the control group system still implemented the FIFO strategy; the sample system of this invention, its factory macro-state diagnosis module, based on aggregated data (e.g., the average queue length of key bottleneck equipment exceeds a preset threshold of 10), diagnosed that it had entered a congestion state and automatically triggered the path optimization decision module to switch from the strategy library and activate the throughput priority optimization strategy, which was configured to prioritize the path with the shortest dynamically estimated time; the performance data of the two groups of systems under this condition are recorded in Table 2;

[0043] Table 2: Performance comparison data under congestion conditions.

[0044]

[0045] Referring to the data in Table 2, under congested conditions, the FIFO strategy of the control group resulted in a large backlog of work-in-process orders at the bottleneck point, with an average global work-in-process order cycle as high as 18.2 hours. In contrast, the factory macro-state diagnosis module of this invention, in conjunction with the strategy library, automatically switches its decision logic from a profit-first optimization strategy to a throughput-first optimization strategy, prioritizing the physically fastest path option (based on calculations by the dynamic queue monitoring module). This strategy switch increased the total system throughput from 55 to 71 (an increase of 29.1%) and reduced the average cycle time of global work-in-process orders from 18.2 hours to 14.1 hours.

[0046] Example 3: This example combines Figs. 1 to 3 A process path optimization system for flexible OLED manufacturing is described, such as... Fig. 1As shown, the system obtains the business and economic attributes of work-in-process orders, such as order value and time sensitivity, from an external Enterprise Resource Planning (ERP) system through a business attribute extraction module. Simultaneously, it obtains the current physical status and available path options of the work order from an external Manufacturing Execution System (MES) through a physical status and path extraction module. The economic attribute path fusion and profile generation module receives the aforementioned business and economic attributes and physical paths, and combines this with the real-time queuing quantity of key resources provided by a dynamic queue monitoring module to calculate the dynamic physical cost. This generates a dynamic profile of the economic attributes of the work-in-process order—a multi-path economic profile—for each available path option. The path optimization decision module then uses a preset optimization... The strategy sorts these profiles to select an optimal process path and sends it to the MES for execution. The optimization strategy based on this decision-making process is automatically activated by a factory macro-state diagnosis module that diagnoses the factory's macro-operating status, such as congestion, based on aggregated data obtained from the MES. To ensure the timeliness of the decision, a dynamic profile caching and subscription module for the economic attributes of work-in-process orders caches all profiles and subscribes to the status change events of key resources from the physical state and path extraction module. Once a status change event is received, the path economic mutation diagnosis module will perform a diagnosis. If the original optimal path fails, it will send a decision invalidation instruction to the MES and trigger the path optimization decision module to re-determine the decision.

[0047] like Fig. 2 As shown, the system of this invention has a higher value than the control group in terms of the on-time delivery rate (%) of high-value work orders; a lower value in terms of the average cycle time (hours) of high-value work orders; and a higher value in terms of the relative value of total output economic value. Fig. 3 As shown, the fusion and profile generation module first triggers the dynamic queue monitoring module, which then queries the MES system for the number of key resources queued for path A and path B, finding them to be 0 and 3 respectively. Simultaneously, the dynamic queue monitoring module retrieves the static processing time of path A (2.0 hours), path B (2.5 hours), and standard unit processing time (0.5 hours) from the process path attribute library. Based on this data, the fusion and profile generation module calculates the dynamic duration, resulting in a dynamic duration of 2.0 + 0 * 0.5 = 2.0 hours for path A and 2.5 + 3 * 0.5 = 4.0 hours for path B. After generating a profile containing this dynamic cost, the fusion and profile generation module submits the updated profile data to the path optimization decision module for final path selection based on the dynamic cost.

[0048] Example 4: When the process path optimization system is first deployed or periodically maintained and calibrated, a calibration procedure needs to be performed to ensure that its internal parameters and decision logic reflect the operating characteristics and management objectives of a specific flexible OLED factory. The thresholds used to distinguish between normal and congested states in the factory macro-state diagnosis module, the quantitative ranking rules for optimization strategies in the path optimization decision module, and the decision invalidation trigger conditions in the path economic change diagnosis module are determined according to the following steps: For the threshold calibration of the factory macro-state diagnosis module, this process utilizes historical manufacturing execution system data; specifically, the total number of work-in-process orders recorded every ten minutes over the past six months, and the corresponding average cycle time of global work-in-process orders are extracted; through statistical analysis, a total... A scatter plot of the number of work-in-process (WIP) orders versus average cycle time was used, and nonlinear regression fitting was performed to identify the inflection point where the average cycle time began to grow rapidly as the total number of WIP orders increased. In this embodiment, the analysis found that when the number of WIP orders exceeded 5000, the growth rate of the average cycle time accelerated. Therefore, a total number of WIP orders >= 5000 was set as the quantitative judgment criterion for triggering a switch from normal to congested state. This threshold was configured in the factory macro-state diagnosis module. For the optimization strategy quantification rules in the path optimization decision module, taking the profit-first optimization strategy as an example, the configuration process is to convert management priorities into calculation rules. This strategy is configured to calculate a comprehensive score for the dynamic profile of the economic attributes of each received WIP order, and its calculation formula is determined as follows: ,in, For the overall score, The values ​​are mapped based on the order value (high, medium, low) (high=10, medium=5, low=1). This is a value obtained by mapping based on the order time sensitivity score. Estimated duration of dynamic events in the portrait (in hours); weighting coefficient This reflects the trade-offs between profit, delivery time, and efficiency, with specific values ​​that can be determined through offline simulation or by management based on business priorities. In a calibration scenario where profit is the primary consideration, the weights can be set as follows: When the path optimization decision module receives a profile of two available options for a work-in-process order, profile 1: {Order value: High ( Time sensitivity score: 8 ), h}, Portrait 2: {Order Value: High ( Time sensitivity score: 8 ), h, yield rate: medium}, then the module calculates the score: , ;because The module will select the path corresponding to portrait 2 as the optimal process path; through this quantitative scoring and ranking procedure, the management intention of prioritizing profits is concretized into executable algorithmic logic.

[0049] The trigger condition in the path economy mutation diagnosis module, which determines that the economic attributes of the original optimal path are now inferior to those of the alternative paths, is also configured as a comparison procedure. This procedure is set as follows: upon receiving a physical resource status change event, if the updated dynamic estimated duration of the original optimal path is... If any of the following conditions are met: (a) (a) Become an invalid value; or (b) The dynamic estimated time exceeds the optimal alternative path in the cache. Reaching a preset time threshold ,Right now The original decision is deemed invalid; where the time threshold is... The calibration is based on the statistical distribution of the physical transfer time of work orders between different processes within the factory, taking its 95th percentile value. If this value is 15 minutes, then... Set to 0.25 hours; this procedure provides an objective and operational quantitative basis for decision invalidation; regarding the standard unit processing time stored in the process path attribute library. The determination process is based on historical data analysis; for each physical resource, records of all work-in-process orders processed over a past period are extracted from the MES database, excluding waiting time and only counting physical processing time; the average of these processing times is calculated, and this average is used as the value of the physical resource. The value is stored in the process path attribute library; if the analysis of historical data of EVAP-03 shows that its average physical time for processing one standard work-in-process order is 1.0 hour, then its corresponding This is calibrated to 1.0 hour; this process ensures that the basic parameters used in the dynamic estimation of duration are derived from actual production data; through the above calibration and configuration procedures, the parameters and decision-making logic within the process path optimization system are given source and operation definitions, enabling the system's automated decision-making behavior to serve the preset management objectives based on objective data and rules.

[0050] Example 5: When the process path optimization system is applied to a new flexible OLED production line or a new product model is introduced, in order to ensure that the physical cost data on which the system makes decisions reflects the actual process capability, a process path attribute library initialization and verification procedure needs to be executed. This procedure first extracts all defined process paths and their associated physical recipe lists from the Manufacturing Execution System and the Recipe Management System through a data interface. For each physical recipe, historical production data is used for statistical analysis to calculate its average physical processing time, average yield, and standard material consumption under stable production conditions, which are then used as its static physical cost attributes. The initial values ​​of standard yield and standard cost are entered into the process path attribute library. For new formulas or paths lacking historical data, the initial physical cost attributes are obtained through small-batch trial production or process simulation. After initialization, the system enters a continuous data verification and update cycle. The system automatically compares the physical cost attributes stored in the process path attribute library with the actual operating data statistics of the path or formula in the past 24 hours every day. When the deviation between the statistical value and the value in the library exceeds a preset threshold, such as a time deviation greater than 10% or a yield deviation greater than 2%, the system automatically marks the attribute and prompts the process engineer to review and confirm it. After confirmation, the system uses a sliding time window, such as the average operating data of the most recent 30 days, to update the corresponding static physical cost attributes in the library.

[0051] To ensure that the economic heatmap generated by the management visualization module provides effective decision support information for managers in specific factories, a visualization rule base calibration procedure needs to be executed after system deployment or when interface rules are adjusted. This procedure begins with the joint participation of system engineers and management representatives. Based on a simulation environment or historical data playback, a predefined set of rules is tested, mapping different combinations of economic attributes of work-in-process orders (e.g., {high value, high time sensitivity}, {medium value, general time sensitivity}, {low value, non-urgent}) to different visual elements such as color, shape, and dynamic effects. During the test, management representatives perform a series of typical identification and judgment tasks on the simulated factory dashboard interface, locating all work-in-process orders with high penalty risk, determining the current value backlog in the vapor deposition area, and recording the time and accuracy of task completion. Based on test feedback and performance data, the mapping rules are adjusted. If red and yellow flashing are difficult to distinguish in high-density areas, one of them may be adjusted to a specific shape marker or its flashing frequency may be adjusted. The testing and adjustment process is repeated until the preset performance target, such as critical work-in-process, is achieved. The work order recognition time is less than 5 seconds, the status misjudgment rate is less than 1%, and the final confirmed mapping rules are solidified into the visualization rule base of the factory instance. In order to enable the economic heat map generated by the management visualization module to objectively and procedurally reflect the economic risk level of work-in-process work orders, the calibration process of the visualization rule base also includes a risk quantification procedure. This procedure is configured to access the process path attribute library to obtain a preset penalty risk coefficient matrix for defining the risk level corresponding to different combinations of business economic attributes. Combined with the order value and order time sensitivity obtained from the business attribute extraction module, a comprehensive economic risk score is calculated for a dynamic profile of the economic attributes of a work-in-process work order. This score is then used to look up in a multi-level risk threshold table to automatically determine a corresponding combination of visual elements. For example, work-in-process work orders with a score between 90 and 100 are fixedly assigned to the first visual style, i.e., red high-frequency flashing, while work-in-process work orders with a score below 30 are fixedly assigned to the second visual style, i.e., green static. In this way, the subjective perception calibration process of managers is transformed into an engineering calibration process of objective risk score thresholds.

[0052] Example 6: After the process path optimization system has been running stably for three months, its decision analysis module is configured to automatically start during the non-production period at the beginning of each month to perform asynchronous statistical aggregation analysis on the decision logs generated by the path optimization decision module of the previous month. These decision logs record each decision event, including the identifier of the work-in-process order involved, its commercial economic attributes (order value, time sensitivity, penalty risk), all available process path options at that time and their corresponding dynamic profiles of the economic attributes of the work-in-process order (including dynamic estimated duration, standard yield, etc.), the optimal process path and physical formula that were finally selected, and the alternative path options that were abandoned. The core processing procedure of the decision analysis module is set as follows: First, all decision events involving high-value work-in-process orders are screened out. Then, among these events, those decision instances where the ultimately chosen path does not have the lowest static physical cost (e.g., the shortest static estimated time or the highest standard yield) are further identified, i.e., economic path compromises have occurred. Next, the module aggregates and statistically analyzes these compromise instances, calculating the frequency with which each physical resource (or a specific process path / formulation combination) appears as a low-cost option to be abandoned, and correlates the work-in-process order value corresponding to these compromise decisions with the estimated delay time (the difference between the dynamic estimated time of the chosen path and the static estimated time of the abandoned path), thereby quantifying the potential economic impact caused by specific physical resource constraints.

[0053] In the specific operational cycle analysis of this embodiment, the decision analysis module processed approximately 1.5 million decision logs recorded last month, identifying about 50,000 instances involving economic path compromises in high-value work-in-process orders. Through aggregated analysis of these compromise instances, it was found that the physical resource EVAP-02 was the key resource upon which the theoretically better path, but which was not selected due to queuing, failures, or other limitations, depended in over 60% of the scenarios. The module further calculated that path compromises related to EVAP-02 estimated to have resulted in an average additional 1.5 hours of processing time for high-value work-in-process orders. Subsequently, the management insight report module automatically generated a management insight report based on this analysis. This report clearly stated that although the physical operational indicators of EVAP-02 (such as utilization rate and failure rate) may be within the normal range, from the perspective of economic decision impact, it was the primary factor leading to path selection compromises for high-value orders last month, constituting a key economic bottleneck in factory operations. This report was... The data is pushed to the production management and equipment maintenance departments, providing them with objective data support based on actual economic consequences when making decisions on equipment maintenance prioritization, capacity planning, or potential technological transformation investments in the next stage, rather than relying solely on traditional physical performance indicators. To utilize the systemic economic path compromise bottlenecks identified by the decision analysis module, the system also includes a closed-loop update procedure for decision parameters. This procedure is configured to automatically extract the physical resource identifiers identified as strategic bottlenecks from the report generated by the management insight report module, access the process path attribute library, and perform a programmed adjustment to the physical cost attribute of the process path option corresponding to the physical resource. For example, the static estimated duration of the path is multiplied by a preset risk adjustment coefficient greater than 1. This adjusted physical cost attribute is then used in the next dynamic physical cost attribute calculation by the economic attribute path fusion and profile generation module, thereby enabling the system to automatically reduce its dependence on the identified bottleneck resource in subsequent path optimization decisions.

[0054] To further illustrate the beneficial effects of the dynamic protection mechanism for decision-making timeliness of the present invention, the following comparative examples are provided.

[0055] Comparative Example 1: Using the same work-in-process (WIP) orders, physical resources, process path attributes, and initial event settings as in Example 1 (see Example 2), specifically, high-value WIP-101 and low-value WIP-202 arrive at the decision point almost simultaneously, with available paths A (EVAP-03, dynamically estimated duration 2.0 hours) and B (EVAP-04, dynamically estimated duration 3.5 hours). The difference is that this comparative example uses a conventional Manufacturing Execution System (MES) representing the prior art. Although this conventional MES can acquire physical status, its architecture lacks a module that integrates commercial economic attributes with physical paths (i.e., it lacks an integration and profile generation module), and it also lacks the path economic mutation diagnosis module of this invention and its triggered decision invalidation mechanism. The decision logic of this conventional MES is configured to execute... The conventional first-in, first-out (FIFO) rule is used in this field. In the operation of this comparative example, the conventional MES processes the queue of decision points according to the FIFO rule. Assuming that WIP-101 arrives at the queue physically before WIP-202, the system processes WIP-101 first according to the FIFO rule. Since the physical duration of path A (2.0 hours) is better than that of path B (3.5 hours), the system makes a physically reasonable decision and dispatches WIP-101 to path A (EVAP-03). While WIP-101 is being physically transferred to EVAP-03, EVAP-03 suddenly experiences a physical failure, and its status changes to failure. The physical layer re-dispatch logic of the conventional MES is triggered. The system withdraws the instruction of WIP-101 and puts it back at the end of the original decision point waiting queue (i.e., after WIP-202).

[0056] At this point, the conventional MES processes the queue again according to its FIFO decision logic. The system detects WIP-202 at the head of the queue and assigns it the only currently available path B (EVAP-04). According to the process path attribute library data in Example 1, path B (EVAP-04)... It takes 2.5 hours. The processing time is 1.0 hour. At this time, EVAP-04 has no queue (Nwip=0), therefore the processing time for low-value work order WIP-202 is 1.0 hour. =2.5 + (0 × 1.0) = 2.5 hours. WIP-101, being high-value, time-sensitive, and carrying a penalty risk, is forced to queue at the decision point until WIP-202 completes processing after 2.5 hours. Only then is WIP-101 dispatched to EVAP-04, whose processing time is also 2.5 hours. Therefore, under this standard MES processing, the total time for the high-value work order WIP-101 from the occurrence of the EVAP-03 failure event to its completion on EVAP-04 is: 2.5 hours (waiting time for WIP-202 to process) + 2.5 hours (WIP-101's own processing time) = 5.0 hours; The test results show that, compared with Embodiment 1 of the present invention (total time of 3.5 hours), the conventional MES lacks a dynamic protection mechanism that can automatically evaluate economic attributes, invalidate the original decision and trigger re-optimization when the physical environment changes suddenly. As a result, after the failure occurs, it automatically executes a physically reasonable (FIFO) but economically costly (making high-value work orders wait) path selection, which causes the high-value work order WIP-101 to be unnecessarily delayed by an additional 1.5 hours, increasing its delivery risk and potential penalty losses.

[0057] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0058] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A process path optimization system for a flexible OLED manufacturing process, characterized by, The system comprises: a business attribute extraction module configured to obtain business economic attributes of the work-in-process order from an enterprise resource planning system; a physical state and path extraction module configured to obtain a current physical state of the work-in-process order and one or more available process path options from a manufacturing execution system; an economic attribute path fusion and portrait generation module connected with the business attribute extraction module and the physical state and path extraction module, and configured to generate an economic attribute dynamic portrait of the work-in-process order for each of the one or more available process path options; a path optimization decision module configured to sort the one or more economic attribute dynamic portraits of the work-in-process order according to a preset optimization strategy, and select an optimal process path; a work-in-process order economic attribute dynamic portrait caching and subscription module configured to cache the economic attribute dynamic portraits of the work-in-process order generated by the economic attribute path fusion and portrait generation module, and subscribe to state change events of key physical resources relied on by the optimal process path from the physical state and path extraction module; a path economic mutation diagnosis module configured to send a decision invalidation instruction to the manufacturing execution system and trigger the path optimization decision module to re-execute optimization decision based on an updated economic attribute dynamic portrait of the work-in-process order when a state change event is received and it is diagnosed that the economic attribute of the optimal process path has deteriorated compared to other available process path options in the cache.

2. The process path optimization system for a flexible OLED manufacturing process of claim 1, wherein, The business economic attributes include at least one of order value, order time sensitivity, and penalty risk; the economic attribute path fusion and portrait generation module is further configured to access a process path attribute library to obtain physical cost attributes corresponding to each available process path option, and also fuse the physical cost attributes into the economic attribute dynamic portrait of the work-in-process order; the physical cost attributes include at least one of process estimated duration, process standard cost, or process standard yield; and the process path attribute library is further configured to store a standard unit processing duration.

3. The process path optimization system for a flexible OLED manufacturing process of claim 2, wherein, The system further comprises a dynamic queue monitoring module configured to respond to the physical state and path extraction module and obtain real-time queuing work-in-process order quantities of key physical resources relied on by the one or more available process path options from the manufacturing execution system in real time; wherein the economic attribute path fusion and portrait generation module is further configured to calculate a dynamic physical cost attribute based on a static physical cost corresponding to the physical cost attribute obtained from the process path attribute library, the real-time queuing work-in-process order quantity, and the standard unit processing duration obtained from the process path attribute library; and fuse the dynamic physical cost attribute as a final physical cost attribute into the economic attribute dynamic portrait of the work-in-process order.

4. The process path optimization system for a flexible OLED manufacturing process of claim 3, wherein, The static physical cost is a static estimated duration, and the dynamic physical cost attribute is a dynamic estimated duration; the economic attribute path fusion and portrait generation module is configured to perform the following calculation: wherein, is a dynamic estimated duration, is a static estimated duration, is a real-time queued work-in-process order quantity, is a standard unit processing duration.

5. The process path optimization system for a flexible OLED manufacturing process of claim 1, wherein, The factory macro state diagnosis module is configured to diagnose a current macro operation state of the factory based on aggregated data obtained from a manufacturing execution system.

6. The process path optimization system for a flexible OLED manufacturing process of claim 5, wherein, The aggregated data is a total work-in-process order quantity or an average queue length of a key bottleneck device in the factory.

7. The process path optimization system for a flexible OLED manufacturing process of claim 1, wherein, The macro operation state includes a normal state and a congestion state.

8. The process path optimization system for a flexible OLED manufacturing process of claim 1, wherein, The strategy library is configured to map the normal state to a profit priority optimization strategy based on order value ranking, and map the congestion state to a throughput priority optimization strategy based on process estimated duration ranking. The physical state and path extraction module is further configured to obtain one or more physical recipes respectively associated with one or more available process path options. The economic attribute path fusion and portrait generation module is further configured to, for each physical recipe, respectively fuse with commercial economic attributes at a data level, and generate a corresponding work-in-process order economic attribute dynamic portrait for each physical recipe. The plurality of work-in-process order economic attribute dynamic portraits based on physical recipes are provided to the path optimization decision module for sorting based on the optimization strategy and determining an optimal process path and an optimal physical recipe corresponding thereto.

9. The process path optimization system for a flexible OLED manufacturing process of claim 8, wherein, The management visualization module is configured to:

10. The process path optimization system for a flexible OLED manufacturing process of claim 1, wherein, obtain work-in-process order economic attribute dynamic portraits generated by the economic attribute path fusion and portrait generation module for a plurality of work-in-process orders in real time; access a visualization rule library that maps commercial economic attributes in the work-in-process order economic attribute dynamic portraits to visual elements; and based on the visualization rule library, generate an economic heat map reflecting real-time economic attribute distribution of the plurality of work-in-process orders on a display interface facing management personnel. The visualization rule library is configured to map work-in-process orders with high order value and high time sensitivity in the commercial economic attributes to a first visual style, and map work-in-process orders with low order value in the commercial economic attributes to a second visual style. The first visual style and the second visual style have distinguishability in color or dynamic effect. The decision analysis module is configured to: asynchronously obtain and analyze decision logs generated by the path optimization decision module during operation; based on the decision logs, perform statistical aggregation analysis of economic decision influence on available process path options and physical resources relied on thereby, to identify strategic bottlenecks causing systematic economic path compromise; and the management insight report module is configured to generate a report for management personnel to use for decision-making based on results of the statistical aggregation analysis.

Images