Display panel production scheduling method and system based on agent cooperation

Abstract

The embodiment of the invention discloses a display panel production scheduling method and system based on agent collaboration, and the method comprises the steps: carrying out the real-time state collaborative updating operation of a plurality of agents in a display panel production system, and obtaining a collaborative state set containing the identification information of the agents and the description of a corresponding production related state; generating an inter-agent cooperation strategy set containing information interaction rules and task allocation priority description based on the set; calling a pre-configured negotiation mechanism to perform conflict detection and negotiation processing on the cooperation strategy set to obtain an optimized cooperation strategy after conflict resolution; and finally, dynamic scheduling operation of display panel production is executed according to the optimized cooperation strategy, and a production scheduling instruction set containing a process execution sequence and an equipment allocation scheme is generated. According to the embodiment of the invention, through intelligent agent cooperation and dynamic scheduling, optimal scheduling of display panel production is realized, and the production efficiency and quality are improved.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of computer technology, and in particular to a display panel production scheduling method and system based on intelligent agent collaboration. Background Art

[0002] In the display panel manufacturing sector, with the continuous expansion of production scale and the increasing complexity of the production process, the requirements for production scheduling are becoming increasingly stringent. Traditional display panel production scheduling methods rely primarily on manual decision-making based on experience, making them difficult to adapt to the demands of modern large-scale production. Under this traditional approach, information flow between various links involved in the production process is poor, resulting in delayed and inaccurate information. For example, the operating status of equipment cannot be promptly fed back to dispatchers, making it difficult for them to make reasonable scheduling decisions based on actual conditions. This can easily lead to idle or overused equipment, reducing production efficiency.

[0003] At the same time, traditional approaches lack a scientific basis for task allocation, often employing fixed allocation patterns that fail to dynamically adjust based on the urgency of orders and the actual load of agents. This results in some urgent orders not being processed promptly, while some agents are overloaded with tasks, leading to task backlogs and production delays.

[0004] Furthermore, traditional approaches lack effective conflict resolution mechanisms for agent collaboration. When conflicts arise among multiple agents over resource allocation and task scheduling, manual intervention is often required to resolve them. This is not only inefficient but also prone to subjective and irrational decision-making. These issues severely restrict the efficiency and quality of display panel production and increase production costs. Therefore, a more intelligent and efficient production scheduling method is urgently needed to address these issues. Summary of the Invention

[0005] Embodiments of the present invention provide a display panel production scheduling method and system based on intelligent agent collaboration.

[0006] In a first aspect, an embodiment of the present invention provides a display panel production scheduling method based on agent collaboration, which is applied to a display panel production scheduling system. The method includes: Performing a real-time state collaborative update operation on multiple agents in a display panel production system to obtain a collaborative state set of each agent, wherein the collaborative state set includes agent identification information and corresponding production-related state descriptions; Generate a collaborative strategy set between agents based on the collaborative state set, wherein the collaborative strategy set includes information interaction rules between agents and task allocation priority description; Invoke a pre-configured negotiation mechanism to perform conflict detection and negotiation on the collaboration strategy set, and generate an optimized collaboration strategy after conflict resolution; Dynamic scheduling operations for display panel production are executed according to the optimized collaboration strategy to generate a production scheduling instruction set including a process execution sequence and an equipment allocation plan.

[0007] In a second aspect, an embodiment of the present invention provides a display panel production scheduling system, including: processor; a storage device having a computer program stored thereon, When the computer program is executed by the processor, the processor implements any of the display panel production scheduling methods based on intelligent agent collaboration.

[0008] An embodiment of the present invention provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, the steps of the display panel production scheduling method based on intelligent agent collaboration are implemented.

[0009] It can be seen that the embodiments of the present invention significantly improve the efficiency and quality of display panel production scheduling, and realize the intelligent, dynamic and efficient production process.

[0010] First, by performing real-time status collaborative update operations on multiple intelligent agents in the display panel production system, the production-related status of each intelligent agent is fully and timely grasped, providing an accurate and real-time data basis for subsequent scheduling decisions, enabling the scheduling system to formulate strategies based on accurate information, avoiding production chaos caused by delayed or inaccurate information.

[0011] Secondly, the inter-agent collaboration strategy set generated from the collaborative state set clarifies the information exchange rules and task allocation priorities between agents, effectively coordinating the work of each agent. Reasonable information exchange rules ensure that key information can be shared between agents in a timely and accurate manner, improving the efficiency of information flow; scientific task allocation priorities ensure that tasks are properly assigned to the appropriate agents, avoiding unreasonable task accumulation and waste of resources.

[0012] Furthermore, a pre-configured negotiation mechanism is used to detect and negotiate conflicts between collaborative strategies, promptly identifying and resolving potential conflicts that may arise during the collaborative process. This not only prevents conflicts from hindering production progress but also improves collaborative efficiency and overall production stability by optimizing collaborative strategies.

[0013] Finally, dynamic scheduling of display panel production is executed based on the optimized collaborative strategy, generating a set of production scheduling instructions that includes the process execution sequence and equipment allocation plan. This set of instructions can be dynamically adjusted based on actual production conditions, ensuring a reasonable process execution sequence and effective equipment allocation. Dynamic scheduling enables the production process to quickly respond to various changes, such as equipment failures and order changes, improving production flexibility and adaptability. Ultimately, it achieves optimal scheduling for display panel production, improving production efficiency and product quality, and reducing production costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 A flowchart of a display panel production scheduling method based on intelligent agent collaboration provided by an embodiment of the present invention.

[0015] Figure 2 A schematic diagram of the basic structure of a display panel production scheduling system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0016] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the embodiments of the present invention are further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0017] See also Figure 1 As shown in FIG, this figure is a flow chart of a display panel production scheduling method based on intelligent agent collaboration provided by an embodiment of the present invention, which can be applied to a display panel production scheduling system. Figure 1 As shown, the method may include steps 110 to 140.

[0018] Step 110: Perform a real-time state collaborative update operation on multiple agents in the display panel production system to obtain a collaborative state set of each agent, wherein the collaborative state set includes agent identification information and corresponding production-related state descriptions.

[0019] In the display panel production scheduling scenario of an embodiment of the present invention, multiple agents are deployed in the display panel production system, each of which is responsible for different production tasks, such as equipment operation, material transportation, and quality inspection. To accurately grasp the real-time status of each agent, the display panel production scheduling system initiates a real-time status collaborative update operation.

[0020] Specifically, the display panel production scheduling system establishes a stable data transmission channel with each agent, continuously collecting agent status information through sensors, monitoring equipment, and communication interfaces. For agents responsible for equipment operation, information such as the current operating mode, speed, and any abnormal alarms is collected. For agents responsible for material transportation, information such as the material's transportation location, transportation progress, and remaining transportation volume is collected. For agents responsible for quality inspection, information such as the number of inspected products, the number of qualified products, and the number of rejected products is collected.

[0021] After collecting this raw information, the display panel production scheduling system performs preliminary cleaning and organization. Noisy, duplicate, and invalid data are removed to ensure data accuracy and completeness. Each agent's identification information is then associated with its corresponding production status description and stored in a data structure to form a collaborative status set. Each record in this set comprehensively reflects the identity of an agent and its current production status.

[0022] Step 120: Generate a collaborative strategy set between agents based on the collaborative state set, where the collaborative strategy set includes information interaction rules and task allocation priority descriptions between agents.

[0023] After obtaining the collaborative state set for each agent, the display panel production scheduling system needs to further develop a set of collaborative strategies between agents. The display panel production scheduling system analyzes the information in the collaborative state set, taking into account the relationships between different agents and the overall requirements of the production process. The development of information exchange rules ensures that necessary information can be shared between agents in a timely and accurate manner, avoiding production problems caused by information delays or inconsistencies. The task allocation priority description specifies how tasks should be assigned to appropriate agents in different situations to ensure that production tasks are completed on time and with high quality.

[0024] In an optional embodiment, step 120 includes: Step 121: parse the production-related status description of each agent in the collaborative status set, and extract a subset of status features including equipment operating status, order completion progress, and process waiting queue length.

[0025] The display panel production scheduling system will conduct an in-depth analysis of the collaborative status set, and filter out information related to the equipment operating status, order completion progress, and process waiting queue length from the production-related status description of each intelligent entity.

[0026] Equipment operating status is analyzed in detail, including whether the equipment is in normal operation, undergoing repair, or in standby mode, as well as indicators such as operating efficiency and energy consumption. Order completion progress is analyzed to determine the amount of work completed, the remaining work, and the estimated completion time for each order. The length of the process queue is analyzed by counting the number of tasks waiting to be processed before each process, reflecting the busyness and load of that process.

[0027] The filtered information is sorted and classified to form a status feature subset that can highlight the key information in the panel production process.

[0028] Step 122: Perform availability evaluation processing on the device operating status in the status feature subset to generate a time distribution matrix of idle periods and busy periods.

[0029] The display panel production scheduling system performs availability assessment on the equipment operating status in the status feature subset, taking into account factors such as the equipment's maintenance plan, fault history, and current operating tasks.

[0030] By analyzing the equipment's operating records and maintenance schedules, we can determine the equipment's availability during different time periods. If the equipment is undergoing maintenance or has a planned downtime, it is unavailable during that time. If the equipment is performing a task, it is busy during that time. Only when the equipment has no tasks and is not undergoing maintenance is it idle.

[0031] Arrange the idle and busy periods of the equipment in chronological order to form a time distribution matrix. This matrix accurately represents the status of the equipment in different time periods and can be used to reasonably arrange production tasks and avoid excessive use or idleness of equipment.

[0032] Step 123: Prioritize the order completion progress in the status feature subset and generate an order priority sequence based on the order delivery time requirement and product type complexity.

[0033] The display panel production scheduling system prioritizes the completion progress of orders within a subset of status features. It considers two key factors: order delivery time requirements and product complexity. Orders with tighter delivery time requirements are given higher priority; orders with more complex product types are also given a higher priority due to their greater production difficulty. By comprehensively evaluating these two factors, all orders are sorted and a priority sequence is generated. During the sorting process, each order's delivery time and product complexity are quantitatively assessed. For example, delivery time is scored based on its proximity to the current time, and product complexity is scored based on factors such as process difficulty and required materials. Orders are then sorted based on these scores to ensure that high-priority orders are processed first.

[0034] Step 124: Perform load balancing analysis on the length of the waiting queue of the process in the state feature subset, and calculate the average waiting time and queue backlog rate of the processes corresponding to different intelligent agents.

[0035] The display panel production scheduling system performs a load balancing analysis on the waiting queue lengths of the processes in the state feature subset. It counts the waiting queue lengths for each agent's corresponding process and calculates the average waiting time based on the processing speed of that process. The average waiting time reflects how long the tasks before a process have to wait before being processed. At the same time, the queue backlog rate is calculated, which is the ratio of the waiting queue length to the maximum load capacity of the process. This ratio can be used to measure the load pressure of the process. By comparing the average waiting time and queue backlog rate of the processes corresponding to different agents, it can be found which processes are overloaded and which processes have excess processing capacity. This can be used to rationally allocate tasks and achieve load balancing.

[0036] Step 125: Construct information interaction rules based on the time distribution matrix, the order priority sequence and the average waiting time, wherein the information interaction rules include the state data type and interaction frequency parameters shared in real time between intelligent agents.

[0037] The display panel production scheduling system constructs information exchange rules based on the previously generated time distribution matrix, order priority sequence, and average waiting time. When determining the types of status data to be shared in real time between agents, critical information from the production process is considered. For example, information such as equipment operating status, order priority, and process waiting time needs to be shared promptly. This information helps agents understand the status of other agents and facilitate better coordination. The interaction frequency parameter is determined based on actual production needs and equipment performance. Critical information, such as equipment failure information and urgent order information, requires real-time sharing, resulting in a higher interaction frequency. For relatively stable information, such as long-term equipment operating statistics, the interaction frequency can be appropriately reduced. By properly setting information exchange rules, smooth information flow between agents can be ensured, improving production scheduling efficiency.

[0038] Step 126: Generate a task allocation priority description in combination with the queue backlog rate and the order priority sequence, wherein the task allocation priority description includes a priority allocation rule for urgent orders and a priority acceptance rule for low-load agents.

[0039] The display panel production scheduling system combines queue backlog rates and order priority sequences to generate a description of task assignment priorities. Urgent orders, due to their tight delivery deadlines, are prioritized for assignment to suitable agents. When selecting an agent, the agent's processing capacity, current load, and degree of compatibility with the order are considered. At the same time, the rule of prioritizing low-load agents is followed. For agents with low queue backlog rates, this indicates that their current load is lighter and they have more processing power to take on new tasks. Therefore, tasks are prioritized for these low-load agents to achieve a balanced distribution of tasks and improve overall production efficiency.

[0040] Step 130: calling a pre-configured negotiation mechanism to perform conflict detection and negotiation processing on the collaboration strategy set, and generating an optimized collaboration strategy after conflict resolution.

[0041] The display panel production scheduling system calls a pre-configured negotiation mechanism to perform conflict detection and negotiation on the collaborative strategy set. During the collaborative process of intelligent agents, various conflicts may arise, such as conflicts in information exchange rules and conflicts in task allocation.

[0042] The negotiation mechanism conducts a detailed inspection of each rule and description in the collaborative strategy set. For information exchange rules, it checks whether there are conflicts in the frequency and data types of interactions between different agents. For task allocation priority descriptions, it checks whether multiple tasks are assigned to the same agent simultaneously, or whether a single agent is overloaded with tasks.

[0043] When a conflict is detected, the negotiation mechanism initiates a negotiation process. This involves the relevant agents exchanging information and negotiating to find the best solution to the conflict. For example, if the frequency of information exchange between two agents is inconsistent, resulting in data delays, the frequency of exchange will be adjusted through negotiation to achieve consistency.

[0044] After conflict detection and negotiation, an optimized collaboration strategy is generated after conflict resolution. This strategy can better coordinate the work between intelligent agents, avoid conflicts, and improve production efficiency.

[0045] As an implementation, step 130 includes: Step 131: performing conflict type identification processing on the information interaction rules in the collaborative strategy set, extracting state data lag conflicts caused by interaction frequency mismatch and information understanding deviation conflicts caused by inconsistent shared data types.

[0046] The display panel production scheduling system identifies conflict types within the information interaction rules within the collaborative strategy set. It carefully analyzes two key factors within these rules: interaction frequency and shared data type. Regarding interaction frequency, the system compares the consistency of interaction frequencies across agents. If the interaction frequency of a particular agent is too high or too low, this can lead to a lag in status data. For example, if one agent sends data at a high frequency while another receives data at a low frequency, the receiver may not receive the latest data in a timely manner, leading to biased production decisions. Regarding shared data types, the system checks the consistency of the shared data types across agents. If the data type sent by one agent is inconsistent with the data type expected by another agent, this can lead to information misunderstanding. For example, if one agent sends equipment energy consumption data while another agent expects to receive equipment speed data, the receiver may misunderstand the information, impacting production coordination. These identified conflict types are organized and recorded for subsequent negotiation and resolution.

[0047] Step 132: Conflict scenario location processing is performed on the task allocation priority descriptions in the collaborative strategy set to identify equipment competition conflicts caused by overlapping allocation of emergency orders and task backlog conflicts caused by insufficient capacity of low-load intelligent agents to undertake tasks.

[0048] The display panel production scheduling system locates conflict scenarios for the task allocation priority descriptions in the collaborative strategy set. It focuses on the allocation of urgent orders and the acceptance capacity of low-load agents. For overlapping allocation of urgent orders, it checks whether multiple urgent orders are assigned to the same device or the same agent at the same time. If this happens, it will cause equipment competition conflicts, and the equipment may not be able to process multiple urgent orders at the same time, thus affecting the delivery time of the orders. For insufficient acceptance capacity of low-load agents, it will analyze whether the low-load agents can complete the new tasks on time after taking them on. If the processing capacity of the low-load agent is limited and it cannot complete the newly assigned tasks within the specified time, it will cause task backlog conflicts and affect the overall production progress. When locating conflict scenarios, the agent identification, task identification and specific circumstances involved in the conflict will be recorded in detail.

[0049] In an optional embodiment, the conflict scenario location processing of the task allocation priority description in the collaboration strategy set, identifying equipment competition conflicts caused by overlapping allocation of urgent orders and task backlog conflicts caused by insufficient capacity of low-load agents to undertake tasks, includes: Step 1321: Extract the target agent allocation list of the urgent order in the task allocation priority description, and count the number of times the same device is simultaneously allocated to multiple urgent orders as the frequency of device overlapping allocation.

[0050] The production scheduling system displays a list of target agents assigned to emergency orders from the task assignment priority description. It then examines which agents are assigned to each emergency order and the equipment used by those agents. It then counts the number of times the same equipment is assigned to multiple emergency orders simultaneously. This number is known as the frequency of equipment overlap, reflecting the intensity of equipment competition. A high frequency of equipment overlap indicates that the equipment is being overused in emergency order allocation, potentially leading to equipment contention conflicts.

[0051] Step 1322: Calculate the severity parameter of the device contention conflict based on the device overlapping allocation frequency and the maximum number of orders that the device can process at one time.

[0052] The display panel production scheduling system calculates the severity parameter of equipment contention conflicts based on the frequency of equipment overlapping allocations and the maximum number of orders that the equipment can process at a single time. The maximum number of orders that a device can process at a single time is an inherent property of the device, which indicates the upper limit of the number of orders that the device can process at the same time. By comparing the frequency of equipment overlapping allocations and the maximum number of orders that the device can process at a single time, the severity of equipment contention conflicts can be assessed. If the frequency of equipment overlapping allocations exceeds the maximum number of orders that the device can process at a single time, it means that the equipment contention conflict is relatively serious; if the frequency of equipment overlapping allocations is close to the maximum number of orders that the device can process at a single time, it means that there is a certain risk of equipment contention conflicts; if the frequency of equipment overlapping allocations is much less than the maximum number of orders that the device can process at a single time, it means that the equipment contention conflict is relatively mild.

[0053] Step 1323: Extract the process processing capability parameters of the low-load agent in the task allocation priority description, wherein the process processing capability parameters include the number of processes that can be processed per unit time and the maximum continuous working time.

[0054] The display panel production scheduling system extracts process processing capacity parameters for low-load agents in the task allocation priority description. It focuses on two key parameters: the number of processes that the low-load agent can process per unit time and the maximum continuous working time. The number of processes that can be processed per unit time reflects the number of processes that the agent can complete per unit time, which reflects the agent's processing efficiency. The maximum continuous working time indicates the maximum time the agent can continuously work, after which the agent may need to rest or undergo maintenance. By extracting these parameters, the processing capacity of the low-load agent can be accurately assessed.

[0055] Step 1324: Count the number of tasks assigned to the low-load agent and the total task processing time, and calculate the difference between the total task processing time and the maximum continuous working time as the remaining capacity.

[0056] The production scheduling system on the display panel counts the number of tasks assigned to low-load agents and the total processing time of these tasks. It examines each low-load agent's current tasks and calculates their total processing time. It then subtracts the total processing time from the maximum continuous working time to determine the remaining capacity. This represents the amount of time the low-load agent can still handle tasks without exceeding the maximum continuous working time. A large remaining capacity indicates that the low-load agent has sufficient processing capacity to take on new tasks. A small remaining capacity indicates that the low-load agent's processing capacity is nearing its limit, and new tasks require caution.

[0057] Step 1325: When the remaining acceptance capacity is less than the preset acceptance capacity value, it is determined that there is a task backlog conflict caused by insufficient acceptance capacity of the low-load intelligent agent, and the intelligent agent identifier and task identifier involved in the task backlog conflict are recorded.

[0058] The display panel production scheduling system compares the remaining capacity with the preset capacity. This is a threshold set based on production experience and equipment performance. If the remaining capacity falls below the preset capacity, it indicates that the low-load agent's capacity is insufficient and may be unable to complete the newly assigned task on time, leading to a task backlog conflict. When a task backlog conflict is identified, the agent and task IDs involved are recorded in detail. This record helps facilitate subsequent negotiation and task adjustments.

[0059] Step 1326: Combine the severity parameter of the device contention conflict and the scope of the task backlog conflict to generate a conflict scene positioning result including conflict positioning coordinates and impact degree.

[0060] The display panel production scheduling system combines the severity parameters of equipment contention conflicts and the scope of task backlog conflicts to generate conflict scenario positioning results. The conflict positioning coordinates clearly identify the specific location of the conflict, including the agents and equipment involved. The degree of impact is determined based on the severity parameters of the equipment contention conflicts and the scope of the task backlog conflicts. For example, if the equipment contention conflict is severe and the task backlog conflict involves multiple agents and tasks, the impact will be higher; if the equipment contention conflict is mild and the task backlog conflict only involves a few agents and tasks, the impact will be lower. This conflict scenario positioning result can provide detailed information for subsequent negotiation and processing, helping to quickly resolve the conflict.

[0061] Step 133: Input the state data delay conflict, the information understanding deviation conflict, the device competition conflict, and the task backlog conflict into a preconfigured conflict classification model to generate a conflict analysis result including a conflict type identifier and an impact range.

[0062] The display panel production scheduling system inputs conflicts related to status data lag, information misunderstanding, equipment competition, and task backlog into a preconfigured conflict classification model. This conflict classification model is a trained neural network model that classifies and analyzes the input conflict information. The input data is received at the model's input layer and then processed by the hidden layer. In the hidden layer, the model extracts and analyzes the characteristics of different types of conflicts to identify the key factors of the conflict. Finally, at the output layer, the model generates conflict analysis results that include a conflict type identifier and impact scope. The conflict type identifier identifies the specific type of conflict, such as status data lag or equipment competition; the impact scope describes the degree and extent of the conflict's impact on the production process.

[0063] Step 134: Based on the conflict analysis results, the auction negotiation sub-mechanism is called, and the intelligent agents involved in the conflict analysis results submit bidding parameters including equipment availability period, process processing efficiency and resource consumption cost, and the task allocation priority is adjusted according to the comprehensive evaluation results of the bidding parameters.

[0064] The display panel production scheduling system invokes the auction negotiation sub-mechanism based on the conflict analysis results. Agents involved in the conflict analysis are required to submit bidding parameters, including equipment availability, process efficiency, and resource costs. Equipment availability indicates the time period during which the agent's equipment is available; process efficiency reflects the speed at which the agent completes a process; and resource costs include costs such as energy consumption and raw material consumption. The display panel production scheduling system conducts a comprehensive evaluation of these bidding parameters, considering factors such as whether the equipment availability matches the task's time requirements, whether the process efficiency is high, and whether the resource costs are low. Based on this comprehensive evaluation, the task allocation priority is adjusted. Agents with excellent bidding parameters have their task allocation priority increased, resulting in higher priority allocation; agents with poor bidding parameters have their task allocation priority decreased.

[0065] Step 135: Call the contract network negotiation sub-mechanism to negotiate and revise the information interaction rules. The dominant intelligent agent publishes the interaction rules to adjust the bidding information. The non-dominant intelligent agent responds and feeds back the constraints including data processing capabilities and communication bandwidth limitations. The information interaction rules are optimized based on the compatibility analysis results of the constraints.

[0066] The display panel production scheduling system invokes the contract network negotiation sub-mechanism to negotiate and revise information exchange rules. A leading agent, responsible for global coordination, is selected as the lead agent. The leading agent generates bidding information for interaction rule adjustments, including requirements for interaction data type and interaction frequency adjustments. This information is then broadcast to non-leading agents. Non-leading agents generate constraint feedback based on their data processing capabilities and communication bandwidth limitations. Data processing capability indicates the type and volume of data an agent can process; communication bandwidth limits specify the amount of data an agent can transmit per unit time. The display panel production scheduling system performs a compatibility analysis on the constraint feedback. It calculates the degree of compatibility between interaction data type adjustments and each agent's data processing capabilities, as well as the degree of compatibility between interaction frequency adjustments and each agent's communication bandwidth limits. Based on the compatibility analysis results, the information exchange rules are optimized. For example, agents with high compatibility and adaptability are prioritized for interaction rule adjustments. Further negotiation is conducted with agents with low compatibility and adaptability to find a suitable solution.

[0067] As a preferred embodiment, step 135 includes: Step 1351: The intelligent agent responsible for global coordination in the current production stage is selected as the leading intelligent agent, and the leading intelligent agent generates interaction rule adjustment bidding information including interaction data type adjustment requirements and interaction frequency adjustment requirements.

[0068] The display panel production scheduling system selects a leading agent responsible for global coordination during the current production phase. This leading agent possesses more comprehensive information and stronger coordination capabilities. Based on the actual production process and requirements, the leading agent generates interaction rules to adjust bidding information. Interaction data type adjustments consider which data types need to be added, reduced, or modified; interaction frequency adjustments are determined based on data timeliness requirements and equipment performance. These adjustment requirements are organized into bidding information for dissemination to non-leading agents.

[0069] Step 1352: Broadcast the interaction rule adjustment bidding information to the non-dominant intelligent agents, and the non-dominant intelligent agents generate constraint feedback information based on their own data processing capabilities and communication bandwidth limitations.

[0070] The dominant agent broadcasts the bidding information, adjusting the interaction rules, to all non-dominant agents. Upon receiving the bidding information, the non-dominant agents generate constraint feedback based on their own data processing capabilities and communication bandwidth limitations. Data processing capabilities determine the type and volume of data an agent can process. If the type of interaction data in the bidding information exceeds the agent's data processing capabilities, the agent may not be able to process the data properly. Communication bandwidth limits the amount of data an agent can transmit per unit time. Excessive interaction frequency can lead to communication congestion. Non-dominant agents then feed these constraints back to the dominant agent for subsequent negotiation.

[0071] Step 1353: Perform compatibility analysis on the constraint feedback information to calculate the matching degree between the interaction data type adjustment requirement and the data processing capability of each agent, as well as the adaptability between the interaction frequency adjustment requirement and the communication bandwidth limitation of each agent.

[0072] The display panel production scheduling system performs a compatibility analysis on constraint feedback. It compares the interaction data type adjustment requirements with the data processing capabilities of each agent and calculates the degree of compatibility. If the interaction data type adjustment requirements fully match the agent's data processing capabilities, the degree of compatibility is high; if they partially match, the degree of compatibility is medium; if they do not match at all, the degree of compatibility is low.

[0073] Similarly, the interaction frequency adjustment requirements are compared with the communication bandwidth limit of each agent to calculate the fitness. If the interaction frequency adjustment requirements are within the communication bandwidth limit of the agent, the fitness is high; if they are close to the communication bandwidth limit, the fitness is medium; if they exceed the communication bandwidth limit, the fitness is low.

[0074] Step 1354: Filter the agents whose matching degree and adaptability both exceed the preset adaptability as rule-accepting agents, and calculate the average data processing capability and communication bandwidth limit of the rule-accepting agents.

[0075] The display panel production scheduling system selects agents whose matching and adaptability exceed a preset adaptability threshold and defines them as rule-accepting agents. The preset adaptability threshold is a pre-set threshold used to determine whether an agent accepts the adjusted interaction rules. Statistics are collected on the data processing capabilities and communication bandwidth limitations of the rule-accepting agents. The mean data processing capabilities and communication bandwidth limitations of these agents are calculated, reflecting the overall processing and communication capabilities of the rule-accepting agents.

[0076] Step 1355: Adjust the interaction data type based on the mean value of the data processing capability, retain the data type that can be processed by each agent and eliminate the data type that does not match the data processing capability.

[0077] The display panel production scheduling system adjusts the interaction data type based on the average data processing capability. It analyzes each data type to see if it can be supported by the data processing capabilities of the rule-accepting agents. It retains data types that can be processed by all agents, ensuring that all rule-accepting agents can process the interaction information properly. It then eliminates data types that do not match the data processing capabilities to avoid information processing issues caused by data type mismatches.

[0078] Step 1356: Adjust the interaction frequency based on the mean value of the communication bandwidth limit, and set the interaction frequency to a maximum supportable frequency that does not exceed the communication bandwidth limit of each agent.

[0079] The display panel production scheduling system adjusts the interaction frequency based on the average communication bandwidth limit. It determines a maximum supportable frequency that does not exceed the communication bandwidth limit of each agent. Setting the interaction frequency to this maximum supportable frequency ensures timely information exchange without causing communication congestion, thereby improving the efficiency and stability of information exchange.

[0080] Step 1357: Generate optimized information interaction rules including the adjusted interaction data type and the adjusted interaction frequency.

[0081] The display panel production scheduling system organizes the adjusted interaction data types and interaction frequencies to generate optimized information interaction rules, which can better adapt to the data processing capabilities and communication bandwidth limitations of the intelligent body and improve the effect of information interaction.

[0082] Step 136: Integrate the processing results of the auction negotiation sub-mechanism and the contract network negotiation sub-mechanism to generate an optimized collaboration strategy after conflict resolution.

[0083] The display panel production scheduling system integrates the results of the auction negotiation sub-mechanism and the contract network negotiation sub-mechanism. The adjusted task allocation priorities in the auction negotiation sub-mechanism are combined with the optimized information exchange rules in the contract network negotiation sub-mechanism. This fusion generates an optimized collaboration strategy after conflict resolution. This strategy fully considers the processing capabilities, resource consumption, and information exchange needs of the agents, making collaboration between agents more efficient and stable.

[0084] Step 140: Execute dynamic scheduling operations for display panel production according to the optimized collaboration strategy, and generate a production scheduling instruction set including a process execution sequence and an equipment allocation plan.

[0085] The display panel production scheduling system dynamically schedules display panel production based on an optimized collaboration strategy. This system combines the information exchange rules and task allocation priorities within the optimized collaboration strategy to schedule the production process in real time. It determines the execution order and timing of each process, forming a process execution sequence. Furthermore, it allocates equipment to appropriate processes based on equipment availability and task requirements, creating an equipment allocation plan. The process execution sequence and equipment allocation plan are organized and integrated to generate a production scheduling instruction set, which contains detailed production scheduling information and guides the actual production operations of the display panel production system.

[0086] In one exemplary embodiment, step 140 includes: Step 141: extract the information interaction rules in the optimized collaboration strategy, and obtain the real-time updated status data of each intelligent agent based on the information interaction rules. The real-time updated status data includes the current operating status of the equipment and the real-time length of the process queue.

[0087] The display panel production scheduling system extracts information exchange rules from the optimized collaborative strategy. Based on these rules, it interacts with each agent and obtains real-time status updates. The system monitors the current operating status of equipment, including whether it is operating normally, undergoing repairs, or in standby mode. The system also calculates the real-time length of the process queue by counting the number of tasks awaiting processing before each process. This real-time status data reflects the agent's current state, providing the latest information for subsequent scheduling decisions.

[0088] Step 142: Extract the task allocation priority description in the optimized collaboration strategy, perform preliminary allocation processing on the current production tasks to be scheduled according to the emergency order priority allocation rule and the low-load agent priority acceptance rule in the task allocation priority description, and generate a pre-allocation list containing candidate agents and candidate equipment.

[0089] The display panel production scheduling system extracts the task allocation priority description from the optimized collaborative strategy. Based on the priority allocation rules for urgent orders and the priority allocation rules for low-load agents, it makes a preliminary allocation of production tasks. For urgent orders, suitable candidate agents and equipment are prioritized for allocation. For other tasks, low-load agents and available equipment are prioritized. These preliminary allocation results are compiled to generate a pre-allocation list of candidate agents and equipment.

[0090] Step 143: Based on the current operating status of the devices in the real-time updated status data, devices in a faulty or maintenance state are excluded, and the candidate device set in the pre-allocated list is updated.

[0091] The production scheduling system displays the current operating status of devices in real-time status data and updates the candidate device set in the pre-assignment list. The system checks the current status of each candidate device and excludes any devices that are experiencing a failure or maintenance condition. This ensures that all subsequently assigned devices are available, improving production reliability.

[0092] Step 144: Based on the real-time length of the process queue in the real-time updated status data, adjust the priority of the candidate agents in the pre-allocated list and select the agent whose queue length is less than the average queue length.

[0093] The display panel production scheduling system uses the real-time process queue lengths from real-time status data to adjust the priorities of candidate agents in the pre-assigned list. First, the average process queue lengths of all agents are calculated as the average queue length. The real-time process queue length of each candidate agent is then compared with the average queue length. Agents with queue lengths shorter than the average queue length are selected and given a higher priority. These agents are currently less loaded and have more processing power to take on new tasks.

[0094] Optionally, step 144 includes: Step 1441: Calculate the average of the real-time lengths of the process queues of all agents as the average queue length.

[0095] The production scheduling system displays statistics on the real-time lengths of all agent process queues. These lengths are added together and divided by the number of agents to obtain the average queue length. This average queue length reflects the average busyness of process queues across the entire production system.

[0096] Step 1442: Extract the real-time length of the process queue of each candidate agent in the pre-allocated list, mark the agents whose queue length is less than the average queue length as first-priority agents, and mark the agents whose queue length is greater than or equal to the average queue length as second-priority agents.

[0097] The production scheduling system displays the real-time queue lengths of each candidate agent from the pre-assigned list. These lengths are compared with the average queue length. Agents with queue lengths less than the average queue length are marked as first-priority agents. These agents are currently lightly loaded and have a higher priority for taking on new tasks. Agents with queue lengths greater than or equal to the average queue length are marked as second-priority agents. These agents are relatively heavily loaded and have a lower priority.

[0098] Step 1443: Review the processing capability of the first-priority agent, extract the number of processes that can be processed per unit time and the current remaining processing time of the first-priority agent, and calculate the number of additional tasks that the first-priority agent can undertake.

[0099] The production scheduling system on the display panel reviews the processing capacity of the first-priority agents. It extracts the number of processes they can process per unit time and the current remaining processing time. Based on this number of processes and the current remaining processing time, it calculates the number of additional tasks the first-priority agent can take on. This number reflects the number of tasks the first-priority agent can still take on without exceeding its processing capacity.

[0100] Step 1444: Perform task backlog risk assessment on the second-priority agent, and predict the task completion time based on the second-priority agent queue length and processing capacity per unit time. If the predicted completion time exceeds the order delivery time, it will be marked as an unavailable agent.

[0101] The production scheduling system on the display panel assesses the backlog risk of second-priority agents. Based on their queue length and processing capacity per unit time, it predicts how long they will need to complete their current tasks. The predicted completion time is compared with the order delivery time. If the predicted completion time exceeds the order delivery time, it indicates that the agent may not be able to complete the task on time, posing a backlog risk. Such agents are marked as unavailable and are not considered for subsequent task assignments.

[0102] Step 1445: The agents in the first priority group whose number of additional tasks they can undertake is greater than zero are selected as priority candidates and sorted from high to low according to the number of additional tasks they can undertake.

[0103] The display panel production scheduling system selects agents from the first-priority group whose number of available additional tasks is greater than zero as priority candidates. These agents are then sorted from highest to lowest by the number of additional tasks they can take on. This sorted list clearly identifies each priority candidate's capacity, facilitating the proper allocation of tasks.

[0104] Step 1446: The second-priority agents that are not marked as unavailable are selected as alternative candidates and sorted from low to high according to the difference between the predicted completion time and the order delivery time.

[0105] The display panel production scheduling system selects agents from the second-priority group that are not marked as unavailable as backup candidates. These agents, despite being heavily loaded, still have some processing capacity. These backup candidates are ranked from lowest to highest based on the difference between their predicted completion time and the order delivery time. The smaller the difference, the more likely the agent is to complete the task on time. This ranking can be used to select backup candidates when the number of priority candidates is insufficient.

[0106] Step 1447: Generate an adjusted candidate agent priority list including a priority candidate ranking and an alternative candidate ranking.

[0107] The display panel production scheduling system organizes the priority candidate rankings and the alternative candidate rankings to generate an adjusted candidate agent priority list, which can comprehensively characterize the priority and acceptance capacity of each agent and provide an accurate reference for task allocation.

[0108] Step 145: Based on the adjusted candidate agents and the adjusted candidate equipment set, the production tasks are sorted in the time dimension according to the order delivery time requirements to generate a process execution timeline including the task start time and task end time.

[0109] The display panel production scheduling system sorts production tasks by time based on the adjusted candidate agents and equipment set. It considers the delivery time requirements for each order and arranges tasks in chronological order. It determines the start and end times for each task, creating a process execution timeline. This timeline visually displays the timing of each task, allowing for the rational allocation of equipment and coordination of agent work.

[0110] Step 146: Perform equipment allocation matching processing according to the process execution time axis and the equipment time distribution matrix to obtain an equipment allocation matching result.

[0111] The display panel production scheduling system performs equipment allocation matching based on the process execution timeline and the equipment's time distribution matrix. For each task in the process execution timeline, the equipment's time distribution matrix is ​​checked to determine which equipment is available during the task's execution time period. For each task time interval, the number of matching equipment is counted. If the number of matching equipment is greater than zero, the equipment with the greatest coverage between the idle period and the task time interval is selected as the allocation equipment; if the number of matching equipment is zero, the end time of the busy period in the equipment's time distribution matrix is ​​extracted, and the time difference between the start time of the task time interval and the end time of the busy period is calculated as the waiting time. The equipment with the shortest waiting time is selected as the delayed allocation equipment, and the start time of the task in the process execution timeline is adjusted to the end time of the busy period. These matching results are sorted to obtain the equipment allocation matching results.

[0112] In one exemplary embodiment, step 146 includes: Step 1461: Extract the start time and end time of each task in the process execution timeline to generate a task time interval set; extract the idle period of each device in the time distribution matrix of the device to generate a device idle time interval set; perform intersection calculation on the task time interval set and the device idle time interval set to identify task-device matching pairs with time overlap.

[0113] The display panel production scheduling system extracts the start and end times of each task in the process execution timeline and combines these times into a set of task time intervals. Simultaneously, it extracts the idle periods of each device from the device's time distribution matrix to form a set of device idle time intervals.

[0114] The intersection of these two sets is calculated. Each task time interval is compared with each device idle time interval to identify any overlaps. These overlaps indicate that the device is available during the task execution period, forming a task-device matching pair.

[0115] Step 1462: For each task time interval, count the number of matching devices. If the number of matching devices is greater than zero, select the device with the largest coverage of the idle period and the task time interval as the allocation device; if the number of matching devices is zero, extract the end time of the busy period in the time distribution matrix of the device, and calculate the time difference between the start time of the task time interval and the end time of the busy period as the waiting time; select the device with the shortest waiting time as the delayed allocation device, and adjust the start time of the task to the end time of the busy period in the process execution timeline; generate a device allocation matching result including directly matched devices and delayed allocation devices.

[0116] The display panel production scheduling system counts the number of matching devices for each task time interval. If the number of matching devices is greater than zero, the coverage of each device's idle period with the task time interval is calculated, and the device with the largest coverage is selected as the allocation device, so as to maximize the use of the device's idle time. If the number of matching devices is zero, the end time of the busy period in the device's time distribution matrix is ​​extracted. The time difference between the start time of the task time interval and the end time of the busy period is calculated, and this time difference is the waiting time. The device with the shortest waiting time is selected as the delayed allocation device, and the start time of the task is adjusted to the end time of the busy period in the process execution timeline to ensure that the task can be processed as soon as possible. The information of the directly matched devices and the delayed allocation devices is sorted to generate the device allocation matching results.

[0117] Step 147: The process execution timeline and the equipment allocation matching result are integrated to generate a production scheduling instruction set including the process execution sequence, task start and end time, and corresponding equipment identification.

[0118] The display panel production scheduling system integrates the process execution timeline and equipment allocation matching results. It integrates the execution order, start and end times, and corresponding equipment identifiers of each task to form a production scheduling instruction set. This set contains detailed production scheduling information, which can guide the actual production operations of the display panel production system and ensure that each task is executed at the appropriate time and equipment.

[0119] In yet another embodiment, step 140 further includes: Step 148: During the scheduling process, the status changes of each intelligent agent are monitored in real time. When a sudden equipment failure is detected, the target task identifier affected by the faulty equipment is extracted; the affected tasks corresponding to the target task identifier are reallocated, the task allocation priority description in the optimized collaboration strategy is called, and the remaining available intelligent agents and equipment are selected as alternative undertaking objects; the difference between the task processing time of the alternative undertaking object and the original planned processing time is calculated as the scheduling adjustment delay amount. If the scheduling adjustment delay amount exceeds the preset delay amount, the order priority re-evaluation process is triggered.

[0120] The display panel production scheduling system monitors the status changes of each intelligent agent in real time during the scheduling process. Through sensors and monitoring equipment, the status information of the intelligent agent is continuously collected. When a sudden equipment failure is detected, the target task identification affected by the faulty equipment will be quickly extracted. These identifications can clearly indicate which tasks are affected by the faulty equipment. The affected tasks will be reallocated. The task allocation priority description in the optimized collaboration strategy will be called to select a suitable alternative undertaking object from the remaining available intelligent agents and equipment. The difference between the task processing time of the alternative undertaking object and the original planned processing time is calculated to obtain the scheduling adjustment delay. If the scheduling adjustment delay exceeds the preset delay, it means that the failure has a greater impact on the production progress, which will trigger the order priority re-evaluation process to readjust the order priority to ensure that important orders can be completed on time.

[0121] Step 149: When a change in order demand is detected, the product type and new delivery time requirement of the change order are extracted; the change order is inserted into the specified position of the process execution sequence, and the execution order of related tasks is adjusted according to the emergency order priority allocation rule in the optimized collaboration strategy; the adjusted process execution sequence is subjected to time conflict detection processing, and if there is time overlap, the negotiation mechanism in the optimized collaboration strategy is called to perform secondary conflict resolution processing; and a production scheduling instruction set containing alternative allocation information of faulty equipment and order change adjustment information is generated.

[0122] When the display panel production scheduling system detects a change in order demand, it extracts the product type and new delivery time requirement of the change order. This information clarifies the specific content of the order change. The change order is then inserted into a designated position in the process execution sequence. This designated position is determined based on the new delivery time requirement and product type. Then, based on the priority allocation rules for urgent orders within the optimized collaboration strategy, the execution order of related tasks is adjusted to ensure timely processing of the change order. The adjusted process execution sequence is then checked for time conflicts. Time overlap between tasks is checked, indicating a potential conflict. When a time conflict is detected, the negotiation mechanism within the optimized collaboration strategy is invoked for secondary conflict resolution. Relevant agents negotiate to find the best solution to the conflict. Finally, a production scheduling instruction set is generated, containing information on alternate allocations for faulty equipment and order change adjustments. This set reflects the latest production scheduling status and ensures smooth production.

[0123] In a non-limiting embodiment, step 140 also includes: collecting real-time execution feedback data of each intelligent agent during the execution of the production scheduling instruction set; performing deviation feature extraction processing on the real-time execution feedback data to generate a deviation feature vector including the process time deviation rate, load fluctuation amplitude and material consumption deviation coefficient; inputting the deviation feature vector into a pre-trained adaptive optimization model, adjusting the time parameters of the process execution sequence and the load parameters of the equipment allocation plan through the adaptive optimization model, and generating optimization adjustment parameters including time parameter correction values ​​and load parameter correction values; iteratively updating the process execution sequence and equipment allocation plan in the production scheduling instruction set based on the optimization adjustment parameters to generate an iterative optimization scheduling instruction set after deviation correction.

[0124] The display panel production scheduling system collects real-time execution feedback data from each agent as it executes the production scheduling instruction set. Sensors and data acquisition devices are used to collect information about each agent's actual execution status, such as the actual execution time of each process, equipment load, and material consumption. Deviation features are extracted from this real-time execution feedback data to calculate the process time deviation rate (the ratio of the deviation between the actual process execution time and the planned execution time); the load fluctuation amplitude (the amplitude of equipment load fluctuation over different time periods); and the material consumption deviation coefficient (the degree of deviation between actual and planned material consumption). These deviation features are combined into a deviation feature vector, which is then input into a pre-trained adaptive optimization model. This model is a neural network model trained on a large amount of data. Based on the input deviation feature vector, it adjusts the time parameters of the process execution sequence and the load parameters of the equipment allocation plan. The adaptive optimization model outputs optimization adjustment parameters, including correction values ​​for the time and load parameters. Based on these optimization adjustment parameters, the process execution sequence and equipment allocation plan in the production scheduling instruction set are iteratively updated. Through continuous iterative updates, deviations are gradually eliminated, resulting in an iteratively optimized scheduling instruction set with corrected deviations.

[0125] In a non-limiting embodiment, step 140 also includes: collecting real-time data of multi-dimensional optimization indicators of the display panel production system; performing feature fusion processing on the real-time data of the multi-dimensional optimization indicators to generate a multi-objective feature matrix including an energy efficiency-loss-synergy correlation matrix; inputting the multi-objective feature matrix into a preconfigured multi-objective collaborative optimization model, and performing multi-objective trade-off adjustment on the resource configuration parameters of the process execution sequence priority and the equipment allocation plan in the production scheduling instruction set through the multi-objective collaborative optimization model to generate multi-objective optimization parameters including a priority adjustment coefficient and a resource configuration correction parameter; based on the multi-objective optimization parameters, collaboratively updating the process execution sequence and the equipment allocation plan in the production scheduling instruction set to generate a collaborative optimization scheduling instruction set after multi-objective balance.

[0126] The display panel production scheduling system collects real-time data on multi-dimensional optimization indicators for the display panel production system. These indicators include energy efficiency, loss, and collaboration. For example, energy efficiency reflects the energy utilization efficiency of equipment, loss reflects material and equipment loss during the production process, and collaboration measures the efficiency of collaboration between agents. Feature fusion processing is performed on the real-time data of these multi-dimensional optimization indicators. The data from different dimensions is integrated and analyzed to generate a multi-objective feature matrix, including an energy efficiency-loss-synergy correlation matrix. This matrix reflects the relationships between different optimization indicators. This multi-objective feature matrix is ​​input into a preconfigured multi-objective collaborative optimization model, a neural network model specifically designed for multi-objective optimization problems. This model performs multi-objective trade-off adjustments on the priority of the process execution sequence and the resource allocation parameters of the equipment allocation plan within the production scheduling instruction set. The multi-objective collaborative optimization model comprehensively considers multiple objectives, including energy efficiency, loss, and collaboration, to find an optimal solution. It outputs multi-objective optimization parameters, including priority adjustment coefficients and resource allocation correction parameters. Based on the multi-objective optimization parameters, the process execution sequence and equipment allocation plan within the production scheduling instruction set are collaboratively updated. Through collaborative updating, the balance of multiple objectives is achieved, and a collaborative optimization scheduling instruction set after the multi-objective balance is generated.

[0127] In a non-limiting embodiment, step 140 is followed by: collecting real-time data of disturbance events of each intelligent agent in the production process; performing feature quantization processing on the real-time data of the disturbance events to generate a disturbance feature vector including an event type identifier, an impact range vector, and a duration parameter; inputting the disturbance feature vector into a pre-trained disturbance impact assessment model, and calculating the impact assessment result of each disturbance event on the production scheduling instruction set through the disturbance impact assessment model, wherein the impact assessment result includes an affected scheduling instruction identifier, an impact severity coefficient, and a production target deviation; calling a preset disturbance response optimization model based on the impact assessment result to generate an adaptive adjustment strategy including a process execution sequence adjustment strategy and an equipment reallocation strategy; and modifying the process execution sequence and equipment allocation plan in the production scheduling instruction set according to the adaptive adjustment strategy to generate an optimized scheduling instruction set after disturbance adaptation.

[0128] The display panel production scheduling system collects real-time data on disturbance events from various agents during the production process. Disturbance events may include equipment failures, raw material supply disruptions, and order changes. Relevant information about these events is collected in real time through various monitoring methods. The real-time disturbance event data is quantified to identify event types, such as equipment failures and order changes. The event's impact vector is calculated to reflect the agents and processes affected by the event. The event's duration parameter, representing the duration from the event's onset to its end, is also calculated. These quantified features are combined into a disturbance feature vector, which is then fed into a pre-trained disturbance impact assessment model. This model, a neural network trained with extensive historical data, calculates the impact of each disturbance event on the production scheduling instruction set based on the input disturbance feature vector. The impact assessment results include the affected scheduling instruction identifier, identifying which scheduling instructions are affected by the disturbance event; the impact severity coefficient, which measures the event's impact on the production process; and the production target deviation, which reflects the degree to which the event caused the production target to deviate. Based on the impact assessment results, a preset disturbance response optimization model is invoked. This model generates an adaptive adjustment strategy, including a process execution sequence adjustment strategy and an equipment reallocation strategy, based on the impact assessment results. Based on this adaptive adjustment strategy, the process execution sequence and equipment allocation plan in the production scheduling instruction set are modified. This modification enables the production schedule to adapt to the impact of the disturbance event, generating a disturbance-adapted optimized scheduling instruction set.

[0129] In actual application, the training and application process of the neural network model can be implemented based on the industrial Internet of Things data acquisition framework and open source deep learning libraries (such as TensorFlow or PyTorch) in existing technologies.

[0130] Specifically, for the conflict classification model, a supervised learning method can be used to construct a multi-layer perceptron structure, and a conflict sample data set in the production history (including features such as the frequency of status data lag and the severity of equipment competition) can be used for training. The weight parameters are optimized through the back-propagation algorithm so that the model can accurately output the conflict type identification and impact range.

[0131] For the adaptive optimization model, the reinforcement learning mechanism can be used to process real-time execution feedback data, take the process time deviation rate and load fluctuation amplitude as state input, define scheduling efficiency as the reward function, and train the intelligent agent to generate time parameter correction values ​​to optimize the process sequence.

[0132] To solve the problem of inconsistent dimensions, we can refer to industrial standardization protocols (such as OPC UA), clarify that time parameters are unified into minutes (such as average waiting time and task processing time), and introduce dimensionless processing when calculating device competition conflict severity parameters (such as the ratio of normalized overlapping allocation frequency to the maximum processing capacity of the device) to ensure data comparability.

[0133] For multi-objective collaborative optimization models, the input features of the energy efficiency-loss-synergy correlation matrix can be integrated based on evolutionary algorithms (such as NSGA-II), and resource configuration correction parameters can be generated through multi-objective Pareto optimal search, so that the output can simultaneously meet the requirements of minimizing energy consumption and maximizing production synergy.

[0134] For the disturbance impact assessment model, a time series prediction network (such as LSTM) can be used to analyze the correlation between the duration of a disturbance event and the impact range vector. During training, historical fault datasets are used to simulate deviations from production targets, outputting a quantitative impact severity coefficient for adaptive adjustment. Furthermore, the model training process can partition the dataset using cross-validation techniques, employing early stopping strategies to prevent overfitting, and deploying it on edge computing nodes for real-time inference, ultimately forming a complete and reproducible intelligent scheduling closed loop.

[0135] In this embodiment of the present invention, the inputs and outputs of the relevant neural network models form a hierarchical decision-making chain. The base layer's input is derived from real-time production data: the conflict classification model receives raw conflict data generated by state collaborative updates (such as lagged data caused by interaction frequency mismatches and overlapping allocation signals for urgent orders). The hidden layer extracts nonlinear correlation features between conflicts and outputs high-level semantic conflict analysis results (including type identification and impact range). This result directly serves to adjust tasks within the auction negotiation sub-mechanism. At the advanced level, the adaptive optimization model uses deviation feature vectors from real-time execution feedback as input (such as process time deviation rate and material consumption deviation coefficient). The feature fusion layer quantifies the degree of production deviation and outputs time parameter corrections and load parameter corrections to drive iterative optimization of the process sequence. At the target optimization layer, the multi-objective collaborative optimization model receives a correlation matrix (energy efficiency, loss, and collaboration data) fused from multiple dimensions. The output layer performs multi-objective trade-off calculations and generates priority adjustment coefficients to reconfigure equipment resources. In the disturbance response layer, the disturbance impact assessment model inputs the disturbance feature vector (event type identifier, duration) into the timing analysis module, outputs the impact assessment result (including production target deviation), and then triggers the disturbance response optimization model to generate a process sequence adjustment strategy.

[0136] As you can understand, all of the above models form a closed loop: low-level inputs (status data, execution deviations) are transformed into optimization parameters (time corrections, priority coefficients) through feature abstraction. These outputs are then fed into the scheduling layer as high-level instructions (such as adjusting the process execution timeline), ultimately achieving hierarchical decision-making coordination from data collection to policy generation. This ensures that the entire logic strictly adheres to the production context characteristics of the input data (such as equipment idle time and order delivery time), and that the output decisions maintain a causal relationship with production goals (load balancing, urgent order delivery).

[0137] In summary, the embodiments of the present invention significantly improve the efficiency and quality of display panel production scheduling, and realize the intelligent, dynamic and efficient production process.

[0138] First, by performing real-time status collaborative update operations on multiple intelligent agents in the display panel production system, the production-related status of each intelligent agent is fully and timely grasped, providing an accurate and real-time data basis for subsequent scheduling decisions, enabling the scheduling system to formulate strategies based on accurate information, avoiding production chaos caused by delayed or inaccurate information.

[0139] Secondly, the inter-agent collaboration strategy set generated from the collaborative state set clarifies the information exchange rules and task allocation priorities between agents, effectively coordinating the work of each agent. Reasonable information exchange rules ensure that key information can be shared between agents in a timely and accurate manner, improving the efficiency of information flow; scientific task allocation priorities ensure that tasks are properly assigned to the appropriate agents, avoiding unreasonable task accumulation and waste of resources.

[0140] Furthermore, a pre-configured negotiation mechanism is used to detect and negotiate conflicts between collaborative strategies, promptly identifying and resolving potential conflicts that may arise during the collaborative process. This not only prevents conflicts from hindering production progress but also improves collaborative efficiency and overall production stability by optimizing collaborative strategies.

[0141] Finally, dynamic scheduling of display panel production is executed based on the optimized collaborative strategy, generating a set of production scheduling instructions that includes the process execution sequence and equipment allocation plan. This set of instructions can be dynamically adjusted based on actual production conditions, ensuring a reasonable process execution sequence and effective equipment allocation. Dynamic scheduling enables the production process to quickly respond to various changes, such as equipment failures and order changes, improving production flexibility and adaptability. Ultimately, it achieves optimal scheduling for display panel production, improving production efficiency and product quality, and reducing production costs.

[0142] See also Figure 2 As shown in FIG. 1 , this figure is a schematic diagram of the basic structure of a display panel production scheduling system 200 provided by an embodiment of the present invention. The display panel production scheduling system 200 includes: Processor 201; a storage device 202 having a computer program 2020 stored thereon; When the computer program 2020 is executed by the processor 201, the processor 201 implements any of the display panel production scheduling methods based on intelligent agent collaboration.

[0143] Based on the above, a readable storage medium is provided, on which a program or instruction is stored. When the program or instruction is executed by a processor, the steps of the above method are implemented.

[0144] It should be noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. For the systems or devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method description.

Claims

1. A display panel production scheduling method based on intelligent agent collaboration, characterized in that: include: Performing a real-time state collaborative update operation on multiple agents in a display panel production system to obtain a collaborative state set of each agent, wherein the collaborative state set includes agent identification information and corresponding production-related state descriptions; Generate a collaborative strategy set between agents based on the collaborative state set, wherein the collaborative strategy set includes information interaction rules between agents and task allocation priority description; Invoke a pre-configured negotiation mechanism to perform conflict detection and negotiation on the collaboration strategy set, and generate an optimized collaboration strategy after conflict resolution; Dynamic scheduling operations for display panel production are executed according to the optimized collaboration strategy to generate a production scheduling instruction set including a process execution sequence and an equipment allocation plan.

2. The method according to claim 1, characterized in that The generation of a collaborative strategy set between agents based on the collaborative state set, wherein the collaborative strategy set includes information interaction rules and task allocation priority descriptions between agents, including: Parsing the production-related status description of each agent in the collaborative state set, and extracting a subset of state features including equipment operating status, order completion progress, and process waiting queue length; Performing availability evaluation on the equipment operating status in the status feature subset to generate a time distribution matrix of idle periods and busy periods; Prioritizing the order completion progress in the status feature subset, and generating an order priority sequence based on order delivery time requirements and product type complexity; Perform load balancing analysis on the length of the waiting queue of the process in the state feature subset, and calculate the average waiting time and queue backlog rate of the corresponding processes of different agents; Constructing information interaction rules based on the time distribution matrix, the order priority sequence, and the average waiting time, wherein the information interaction rules include state data types and interaction frequency parameters shared in real time between agents; A task allocation priority description is generated in combination with the queue backlog rate and the order priority sequence, and the task allocation priority description includes a priority allocation rule for urgent orders and a priority acceptance rule for low-load intelligent agents.

3. The method according to claim 1, characterized in that The calling of a pre-configured negotiation mechanism to perform conflict detection and negotiation on the collaboration strategy set to generate an optimized collaboration strategy after conflict resolution includes: Perform conflict type identification processing on the information interaction rules in the collaborative strategy set, extracting state data lag conflicts caused by interaction frequency mismatch and information understanding deviation conflicts caused by inconsistent shared data types; Perform conflict scenario location processing on the task allocation priority descriptions in the collaborative strategy set, identifying equipment competition conflicts caused by overlapping allocation of urgent orders and task backlog conflicts caused by insufficient capacity of low-load agents to undertake tasks; Inputting the state data delay conflict, the information understanding deviation conflict, the equipment competition conflict, and the task backlog conflict into a preconfigured conflict classification model to generate a conflict analysis result including a conflict type identifier and an impact range; Based on the conflict analysis results, the auction negotiation sub-mechanism is invoked, and the agents involved in the conflict analysis results submit bidding parameters including equipment availability time, process processing efficiency, and resource consumption cost, and the task allocation priority is adjusted according to the comprehensive evaluation results of the bidding parameters; The contract network negotiation sub-mechanism is called to negotiate and amend the information interaction rules. The leading agent publishes the interaction rules to adjust the bidding information. The non-leading agents respond and feedback the constraints including data processing capacity and communication bandwidth limitations. The information interaction rules are optimized based on the compatibility analysis results of the constraints. The processing results of the auction negotiation sub-mechanism and the contract network negotiation sub-mechanism are integrated to generate an optimized collaboration strategy after conflict resolution.

4. The method according to claim 3, characterized in that The conflict scenario positioning processing of the task allocation priority description in the collaborative strategy set, identifying the equipment competition conflict caused by overlapping allocation of urgent orders and the task backlog conflict caused by insufficient capacity of low-load intelligent agents, includes: Extract the target agent allocation list of the urgent order in the task allocation priority description, and count the number of times the same device is simultaneously allocated to multiple urgent orders as the device overlapping allocation frequency; Calculate a severity parameter of the device contention conflict based on the device overlapping allocation frequency and the maximum number of orders processed by the device at a single time; Extracting process processing capability parameters of the low-load agent in the task allocation priority description, wherein the process processing capability parameters include the number of processes that can be processed per unit time and the maximum continuous working time; Counting the number of tasks assigned to the low-load agent and the total task processing time, and calculating the difference between the total task processing time and the maximum continuous working time as the remaining capacity; When the remaining acceptance capacity is less than the preset acceptance capacity value, it is determined that there is a task backlog conflict caused by insufficient acceptance capacity of the low-load agent, and the agent identifier and task identifier involved in the task backlog conflict are recorded; Combined with the severity parameter of the equipment contention conflict and the scope of the task backlog conflict, a conflict scene positioning result including conflict positioning coordinates and impact degree is generated.

5. The method according to claim 3, characterized in that The contract network negotiation sub-mechanism is called to negotiate and amend the information interaction rules. The leading agent publishes the interaction rules to adjust the bidding information. The non-leading agents respond and feedback the constraints including data processing capacity and communication bandwidth limitations. The information interaction rules are optimized based on the compatibility analysis results of the constraints, including: The agent responsible for global coordination in the current production phase is selected as the leading agent, and the leading agent generates interaction rule adjustment bidding information including interaction data type adjustment requirements and interaction frequency adjustment requirements; Broadcasting the interactive rule adjustment bidding information to non-leading agents, which generate constraint feedback information based on their own data processing capabilities and communication bandwidth limitations; Performing compatibility analysis on the constraint feedback information to calculate the degree of compatibility between the interaction data type adjustment requirements and the data processing capabilities of each agent, as well as the degree of compatibility between the interaction frequency adjustment requirements and the communication bandwidth limitations of each agent; Screening agents whose matching degree and adaptability both exceed a preset adaptability as rule-accepting agents, and calculating the average data processing capability and communication bandwidth limit of the rule-accepting agents; Adjusting the interaction data type based on the mean of the data processing capabilities, retaining the data type that can be processed by each agent and eliminating the data type that does not match the data processing capabilities; Adjusting the interaction frequency based on the mean of the communication bandwidth limit to set the interaction frequency to a maximum supportable frequency that does not exceed the communication bandwidth limit of each agent; Generate optimized information interaction rules including adjusted interaction data types and adjusted interaction frequencies.

6. The method according to claim 1, wherein The step of executing the dynamic scheduling operation for display panel production according to the optimized collaboration strategy to generate a production scheduling instruction set including a process execution sequence and an equipment allocation plan includes: Extracting information interaction rules from the optimization collaboration strategy, and obtaining real-time updated status data of each agent based on the information interaction rules, wherein the real-time updated status data includes the current operating status of the equipment and the real-time length of the process queue; Extract the task allocation priority description in the optimized collaboration strategy, perform preliminary allocation processing on the current production tasks to be scheduled according to the urgent order priority allocation rule and the low-load agent priority acceptance rule in the task allocation priority description, and generate a pre-allocation list containing candidate agents and candidate equipment; In combination with the current operating status of the devices in the real-time updated status data, devices in a faulty or maintenance state are excluded, and the candidate device set in the pre-allocated list is updated; In combination with the real-time length of the process queue in the real-time updated status data, the priority of the candidate agents in the pre-assigned list is adjusted to select the agent whose queue length is less than the average queue length; Based on the adjusted candidate agents and the adjusted candidate equipment set, the production tasks are sorted in the time dimension according to the order delivery time requirements, and a process execution timeline including the task start time and task end time is generated; Perform equipment allocation matching processing according to the process execution time axis and the equipment time distribution matrix to obtain an equipment allocation matching result; The process execution timeline and the equipment allocation matching result are integrated to generate a production scheduling instruction set including the process execution sequence, task start and end time and corresponding equipment identification.

7. The method according to claim 6, characterized in that The step of adjusting the priority of candidate agents in the pre-allocated list in combination with the real-time length of the process queue in the real-time updated status data, and selecting an agent with a queue length less than an average queue length, includes: Calculate the average of the real-time lengths of the process queues of all agents as the average queue length; Extracting the real-time process queue length of each candidate agent in the pre-assigned list, marking agents with queue lengths less than the average queue length as first-priority agents, and agents with queue lengths greater than or equal to the average queue length as second-priority agents; Perform a processing capacity review on the first-priority agent, extract the number of processes that the first-priority agent can process per unit time and the current remaining processing time, and calculate the number of additional tasks that the first-priority agent can undertake; Perform a task backlog risk assessment on the second-priority agent and predict the task completion time based on the second-priority agent's queue length and processing capacity per unit time. If the predicted completion time exceeds the order delivery time, the agent is marked as unavailable; The agents with more than zero additional tasks that can be undertaken among the first-priority agents are selected as priority candidates and sorted from high to low according to the number of additional tasks that can be undertaken; The agents in the second priority group that are not marked as unavailable are selected as candidates and sorted from low to high according to the difference between the predicted completion time and the order delivery time; Generate an adjusted candidate agent priority list including a preferred candidate ranking and an alternative candidate ranking.

8. The method according to claim 6, characterized in that The device allocation matching process is performed according to the process execution time axis and the time distribution matrix of the device to obtain the device allocation matching result, including: Extracting the start time and end time of each task in the process execution timeline to generate a task time interval set; Extracting the idle time period of each device in the time distribution matrix of the devices to generate a device idle time interval set; Performing intersection calculation on the task time interval set and the device idle time interval set to identify task-device matching pairs with overlapping time periods; For each task time interval, count the number of matching devices. If the number of matching devices is greater than zero, select the device with the largest coverage of idle time and task time interval as the assigned device. If the number of matching devices is zero, extract the end time of the busy period in the time distribution matrix of the device, and calculate the time difference between the start time of the task time interval and the end time of the busy period as the waiting time; Select the device with the shortest waiting time as the delayed allocation device, and adjust the start time of the task in the process execution timeline to the end time of the busy period; Generates device allocation matching results that include directly matched devices and delayed allocated devices.

9. The method according to claim 1, characterized in that The step of executing the dynamic scheduling operation of display panel production according to the optimized collaboration strategy to generate a production scheduling instruction set including a process execution sequence and an equipment allocation plan further includes: Monitor the state changes of each agent in real time during the scheduling process. When a sudden equipment failure is detected, extract the target task identifier affected by the faulty equipment. Reassign the affected tasks corresponding to the target task identifier, call the task assignment priority description in the optimized collaboration strategy, and select the remaining available agents and devices as alternative undertaking objects; Calculate the difference between the task processing time of the replacement undertaking object and the original planned processing time as the scheduling adjustment delay amount. If the scheduling adjustment delay amount exceeds the preset delay amount, the order priority re-evaluation process is triggered; When a change in order requirements is detected, the product type and new delivery time requirements of the changed order are extracted; Inserting the change order into a designated position of the process execution sequence, and adjusting the execution order of related tasks according to the emergency order priority allocation rule in the optimized collaboration strategy; Performing time conflict detection on the adjusted process execution sequence, and if there is time overlap, calling the negotiation mechanism in the optimization collaboration strategy to perform secondary conflict resolution; Generate a set of production scheduling instructions containing replacement allocation information for faulty equipment and order change adjustment information.

10. A display panel production scheduling system, characterized in that: include: processor; A storage device having a computer program stored thereon, wherein when the computer program is executed by the processor, the processor implements the display panel production scheduling method based on intelligent agent collaboration as described in any one of claims 1-9.

Images