Microswitch production line intelligent scheduling method and system based on graph neural network

Abstract

The invention relates to the technical field of micro switch production line intelligent scheduling methods, in particular to a micro switch production line intelligent scheduling method and system based on a graph neural network, and the method comprises the steps: obtaining an attribute set of nodes, and constructing a node graph of a production line; the dynamic representation module is used for dynamically representing real-time states of nodes; generating a multi-scale state prediction model based on the aggregation weight matrix, and obtaining a state prediction matrix for predicting the state change trend of the equipment; based on the state prediction matrix, constructing a nonlinear global optimization scheduling strategy, and further optimizing task allocation and resource allocation in the production line to obtain a global optimal scheduling scheme; constructing a layered feedback correction mechanism based on the global optimal scheduling scheme; based on a hierarchical feedback correction mechanism, an intelligent maintenance trigger signal is generated for identifying potential faults and triggering preventive maintenance, deep modeling and state prediction are performed on each node of the production line, and efficient task scheduling and equipment maintenance management are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent scheduling methods for micro switch production lines, and more specifically, to an intelligent scheduling method for micro switch production lines based on a graph neural network and a system thereof. Background Art

[0002] With the continuous development of industrial automation and intelligent technology, the production and manufacturing industry of micro switches has increasingly higher requirements for efficiency, quality and maintenance management. However, traditional production line scheduling methods have many limitations when facing complex equipment and process management, especially in the efficient operation and continuous maintenance of micro switch production lines. These traditional methods are difficult to provide accurate and flexible support. Current technologies mainly rely on static planning or rule-based scheduling mechanisms for scheduling optimization. These methods lack the ability of dynamic scheduling and real-time adjustment, and are difficult to adapt to production needs under multi-node and complex process conditions. Especially in production lines with high correlation between nodes, traditional scheduling methods are difficult to achieve ideal results in resource allocation, task priority control and equipment maintenance, resulting in limited overall production efficiency and low equipment utilization.

[0003] Among the existing scheduling methods for micro switch production lines, typical planning is based on static task allocation or linear scheduling algorithms. These methods usually allocate tasks to each node device in sequence through preset rules or established task scheduling methods. However, the efficiency of this scheduling method is heavily dependent on the preset planning structure. Once the equipment status or task priority in the production environment changes, the scheduling scheme lacks sufficient flexibility to adapt to new production needs. For example, when a key device on the production line has a reduced operating efficiency due to overload, the traditional scheduling method cannot dynamically evaluate the impact of this node on the entire production line, nor can it quickly adjust the task allocation of subsequent nodes, which causes a chain reaction in the production process due to the delay of the bottleneck node, ultimately affecting the overall production efficiency. In addition, the existing scheduling schemes usually do not include preventive maintenance for equipment failures, and often rely on regular equipment maintenance or repair after sudden failures, lacking active status monitoring and maintenance measures. This passive maintenance method affects the continuity of the production line, increases equipment downtime and maintenance costs, and further restricts production efficiency.

[0004] In order to solve these problems, some intelligent scheduling methods have tried to introduce machine learning algorithms in recent years to improve the adaptability and intelligence of scheduling systems. However, existing methods are usually limited by the simplicity of algorithm models and insufficient computing resources, and fail to fully explore the correlation between equipment nodes on the production line. Such methods are mostly adjusted based on the individual status of the equipment, fail to effectively handle the dependencies between nodes, and fail to establish a complete node status prediction mechanism. When dealing with complex scheduling requirements, it is difficult for existing technologies to achieve deep modeling of the mutual dependencies and task transfer relationships between nodes on the production line, resulting in a lack of global optimization capabilities for scheduling schemes. In addition, due to the failure to provide fault prediction and status feedback, the existing scheduling methods have limited effects on production line maintenance, and it is difficult to prevent and respond quickly to equipment failures. Summary of the invention

[0005] In view of the above problems, the present invention proposes an intelligent scheduling method and system for a micro switch production line based on graph neural network.

[0006] The present invention provides a micro switch production line intelligent scheduling method based on a graph neural network, comprising:

[0007] A node graph structure is constructed based on production line equipment, workstations and work-in-progress, with the equipment status, location and task allocation as static attributes and the relationship and interaction information between nodes as dynamic attributes, to obtain the attribute set of the nodes and construct a node graph of the production line;

[0008] An aggregation weight matrix is ​​generated based on an adaptive node weight aggregation algorithm to dynamically characterize the real-time status of the node;

[0009] Based on the aggregate weight matrix, a multi-scale state prediction model is generated to obtain a state prediction matrix for predicting a state change trend of the device;

[0010] Based on the state prediction matrix, a nonlinear global optimization scheduling strategy is constructed to further optimize the task allocation and resource allocation in the production line to obtain a global optimal scheduling solution;

[0011] Based on the global optimal scheduling scheme, a hierarchical feedback correction mechanism is constructed to further dynamically correct the scheduling results to improve the stability and robustness of the scheduling;

[0012] Based on the hierarchical feedback correction mechanism, an intelligent maintenance trigger signal is generated to identify potential faults and trigger preventive maintenance.

[0013] Preferably, the adaptive node weight aggregation algorithm assigns a dynamic weight w to each state parameter in the device state set. k , get the comprehensive status value of the node where wk Indicates the dynamic importance of state parameters, s k Indicates the current state of the node, and n is the number of states.

[0014] Preferably, the multi-scale state prediction model is based on the node aggregation weight matrix W agg And the dependency function f(p ij (t),s j ), generate the state prediction matrix P at time t t , where f(p ij (t),s j ) represents the state dependence of node i on node j, and further obtains the predicted state p ij (t+1)=α·W agg (v i )·f(p ij (t),s j ).

[0015] Preferably, the nonlinear global optimization scheduling strategy is based on the state prediction matrix P t and the global optimization function F opt , and further obtain the optimal scheduling solution, where the optimization function g(v j ,t i ) represents the task execution cost, m is the number of tasks, and n is the number of nodes.

[0016] Preferably, the hierarchical feedback correction mechanism adjusts the scheduling scheme by generating a feedback correction matrix H, so as to further improve the scheduling stability and robustness, wherein β is the feedback coefficient.

[0017] Preferably, the intelligent maintenance module generates a maintenance trigger matrix based on the feedback correction matrix Used for preventive maintenance of equipment, where m i (t+1) = γ·h ii (t+1)·p ii (t+1), when m i When (t+1)>threshold, the maintenance of device i is triggered.

[0018] Preferably, the intelligent scheduling method further includes periodically updating the maintenance trigger threshold threshold through a scheduling feedback mechanism, so as to adjust the maintenance trigger condition according to the real-time status of the production line to implement an adaptive fault maintenance strategy.

[0019] The system using the micro switch production line intelligent scheduling method comprises:

[0020] A graph structure building module, used to build the node graph structure based on static and dynamic properties of production line equipment;

[0021] Adaptive node weight aggregation module, used to generate the aggregation weight matrix W based on the real-time status of the node agg ; A multi-scale state prediction module for generating a state prediction matrix P based on the aggregation weight matrix t ;

[0022] A nonlinear global optimization scheduling module is used to generate an optimal scheduling solution based on the state prediction matrix.

[0023] A hierarchical feedback correction module, used to generate a feedback correction matrix H based on the scheduling scheme;

[0024] An intelligent maintenance module is used to generate a maintenance trigger matrix based on the feedback correction matrix and perform preventive equipment maintenance.

[0025] Preferably, the adaptive node weight aggregation module also includes a dynamic update mechanism for automatically adjusting the node weight coefficient w based on the change of the production line status. k , to ensure the dynamic adaptability of the aggregation weight matrix.

[0026] Preferably, the intelligent maintenance module further comprises a maintenance update module, which regularly updates the maintenance trigger threshold and trigger coefficient to ensure the real-time and flexibility of the fault preventive maintenance of the production line under different conditions.

[0027] The invention has the following beneficial effects:

[0028] The present invention uses an intelligent scheduling system based on a graph neural network to perform in-depth modeling and state prediction on each node of the production line, thereby achieving efficient task scheduling and equipment maintenance management. Specifically, the present invention uses the structural characteristics of the graph neural network to construct a graph structure model that reflects the status and mutual relationships of node equipment in the production line. On this basis, an adaptive node weight aggregation algorithm is used to dynamically adjust the state weight of the node to ensure that the scheduling system can reflect the operating status and importance of each node device in real time. This adaptive adjustment mechanism effectively solves the problem that traditional methods cannot cope with dynamic changes in equipment load, ensuring that key nodes obtain higher task priorities, thereby improving overall scheduling efficiency.

[0029] At the same time, the multi-scale state prediction model of the present invention predicts the state change trend of each node through state aggregation and node dependency modeling. This function enables the scheduling system to adjust task allocation and resource allocation in advance to cope with potential changes in equipment status, thereby maintaining efficient operation of the production line when equipment load changes. Compared with the traditional method that relies on the limitations of fixed scheduling rules, the present invention significantly enhances the dynamic response capability of the scheduling system through the prediction function, making task scheduling more flexible and accurate.

[0030] In terms of global optimization, the present invention constructs a nonlinear global optimization scheduling strategy based on the state prediction matrix, so that the system can achieve optimal scheduling in a global range in a multi-node, multi-process production line. This optimization strategy dynamically adjusts the state and load of the node through a feedback correction mechanism to ensure the continuous optimization and stability of the scheduling scheme. In traditional methods, due to the lack of feedback and correction capabilities for production status, production delays and scheduling imbalances usually occur under sudden conditions. The hierarchical feedback correction mechanism of the present invention effectively solves the contradiction between scheduling stability and flexibility, so that the scheduling scheme can quickly restore balance when responding to sudden changes.

[0031] In addition, the intelligent maintenance trigger module generates a maintenance trigger signal by predicting the node status and analyzing the feedback information, thereby achieving preventive maintenance of potential faults. Unlike the prior art that relies on passive maintenance, the intelligent maintenance function of the present invention can make timely adjustments before the equipment status becomes abnormal, avoid equipment failures, and improve the continuity and stability of the production line. Overall, the present invention achieves synergy effects that are difficult to achieve with prior art in multiple aspects such as graph structure modeling, state prediction, global optimization, and fault prevention. Through the complementarity and superposition of various modules, the scheduling efficiency and reliability of the production line are significantly improved, providing an innovative solution for the intelligent management of micro switch production lines. . BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a system logic block diagram of the present invention. DETAILED DESCRIPTION

[0033] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation methods, structures, features and effects thereof are described in detail below in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0034] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0035] Please refer to Figure 1 The present invention provides an intelligent scheduling method and system for a micro switch production line based on a graph neural network. Through the coordinated action of six modules, namely, graph structure construction of each node of the micro switch production line, real-time dynamic characterization of the state, multi-scale state prediction, global optimization scheduling, hierarchical feedback correction, and intelligent maintenance triggering, efficient scheduling of production tasks and timely prevention of potential failures are achieved, thereby significantly improving production efficiency and equipment reliability.

[0036] The method of the present invention constructs a node graph structure based on the micro switch production line equipment, workstations and work-in-progress, with the state, position and task allocation of each node as static attributes, and the relationship and interaction information between nodes as dynamic attributes. The construction of the node graph structure can make each node of the production line show the mutual state and interaction relationship in an intuitive and systematic form in the graph, and the node set can be represented as V, where v i Represents each node in the production line. Preferably, in one embodiment of the present invention, the graph structure building module 1 can select the occupied time, equipment capacity and working area of ​​the node equipment as static attributes, and the sequence, process connection and task alternation between nodes as dynamic attributes. Such attribute division not only makes the graph structure information richer, but also can be adaptively adjusted when the production line load changes, ensuring the accuracy of scheduling.

[0037] As a preferred embodiment, the present invention adopts an adaptive node weight aggregation algorithm to generate an aggregation weight matrix to dynamically characterize the real-time state of the node. Specifically, by assigning a dynamic weight w to each state parameter of the node k , calculate the comprehensive state value where w k is the weight of the state parameter, representing the influence of the parameter on the current node state, s k is the state parameter of the node, and n is the number of state parameters. Preferably, in one embodiment of the present invention, the important state parameters of the node device, such as the running time, load and process completion status, are assigned weights of 0.3, 0.4 and 0.3 respectively to balance the influence of each attribute in the comprehensive state. The effect of this weight distribution is significant. By using an adaptive aggregate weight matrix, the method can timely reflect the key state and service load of the node device, making subsequent scheduling more efficient and flexible.

[0038] As a preferred embodiment, based on the generated aggregate weight matrix, the present invention constructs a multi-scale state prediction model to obtain the state prediction matrix P at the current time t t The prediction matrix is ​​used to predict the state change trend of the node device in the future. Each element p ij(t+1) represents the influence of node i on the state of node j at time t+1. The calculation formula is p ij (t+1)=α·W agg (v i )·f(p ij (t),s j ), where α is the adjustment factor, preferably between 0.5 and 1.5, to enhance the balance of state adjustment between different nodes; f(p ij (t),s j ) is the dependency function of node i on node j, and p is predicted by the state at time t ij (t) and the state parameter s of node j j In practical applications, experience shows that it is better to set a larger adjustment factor for higher load nodes to ensure that high-load tasks are given priority in scheduling, further improving the resource utilization of the production line.

[0039] As a preferred embodiment, the present invention adopts a nonlinear global optimization scheduling strategy based on the state prediction matrix. By introducing the global optimization function Obtain the global optimal scheduling solution for the production line. Among them, the optimization function F opt To minimize the overall cost, g(v j ,t i ) represents node v j Execute Task i The execution cost is preferably the sum of comprehensive parameters such as resource consumption, device idle time or node load; m represents the number of tasks, and n represents the number of nodes. In one embodiment of the present invention, in order to further improve the efficiency of the scheduling scheme, the execution cost of the device node with higher resource consumption is set to be larger, so that tasks with lower resource consumption can be dynamically prioritized in the execution cost calculation, making the overall scheduling resource allocation more efficient and reasonable.

[0040] Based on the optimization scheme, this paper proposes a hierarchical feedback correction mechanism to further improve the stability and robustness of scheduling. By generating a feedback correction matrix H, the scheduling results of the production line are dynamically corrected. where h ij (t+1) represents the correction value of node i to node j at time t+1, and β is the feedback coefficient, which is preferably selected between 0.1 and 1.0 to ensure that the feedback correction amplitude is moderate so that the scheduling process can quickly restore balance under sudden conditions. In the embodiment of the present invention, after using the feedback correction mechanism, the node switching process of the production line is significantly smooth, the frequency of equipment conflicts is significantly reduced, and the stability of the production task is ensured.

[0041] As a preferred embodiment, the present invention proposes an intelligent maintenance trigger module based on feedback correction mechanism. This module is used to identify potential faults in the production line and trigger preventive maintenance by generating a maintenance trigger matrix. Further determine the node maintenance signal m i (t+1) = γ·h ii (t+1)·p ii (t+1), where m i (t+1) represents the maintenance trigger signal for node i at time t+1, γ is the trigger coefficient, and the empirical value is set to 0.1 to 1.0. The adjustment factor can be dynamically adjusted based on the fault history or current load of the node. Preferably, in one embodiment of the present invention, a lower threshold is set for high-load node devices, thereby ensuring that the equipment is maintained in time before a minor fault occurs, reducing the impact of the fault on production.

[0042] As a preferred embodiment, the intelligent maintenance system further includes a scheduling feedback mechanism that can periodically update the maintenance trigger threshold, thereby dynamically adjusting the trigger condition according to the real-time production line status. Preferably, in one embodiment of the present invention, the maintenance trigger threshold is adjusted in real time based on parameters such as the current load of the node and the equipment running time. When the node load exceeds 60% or the continuous running time exceeds 100 hours, the trigger threshold is automatically reduced by 20%, thereby strengthening the maintenance response under high load conditions.

[0043] The present invention also discloses a system for the intelligent scheduling method of a micro switch production line, comprising:

[0044] A graph structure building module 1, used to build the node graph structure based on the static and dynamic attributes of the production line equipment;

[0045] Adaptive node weight aggregation module 2, used to generate the aggregation weight matrix W based on the real-time status of the node agg ;

[0046] The multi-scale state prediction module 3 is used to generate a state prediction matrix P based on the aggregation weight matrix t ;

[0047] A nonlinear global optimization scheduling module 4 is used to generate an optimal scheduling solution based on the state prediction matrix.

[0048] A hierarchical feedback correction module 5, used to generate a feedback correction matrix H based on the scheduling scheme;

[0049] Intelligent maintenance module 6, used to generate a maintenance trigger matrix based on the feedback correction matrix And perform preventive equipment maintenance. In actual application, the modules in the system work together to ensure that the micro switch production line achieves the best results in efficient scheduling and equipment maintenance.

[0050] The adaptive node weight aggregation module 2 also includes a dynamic update mechanism for automatically adjusting the node weight coefficient w based on the change of the production line status. k , to ensure the dynamic adaptability of the aggregation weight matrix.

[0051] The intelligent maintenance module 6 also includes a maintenance update module, which regularly updates the maintenance trigger threshold and trigger coefficient to ensure the real-time and flexibility of the fault preventive maintenance of the production line under different conditions.

[0052] In order to verify the superiority of the micro switch production line intelligent scheduling method and system based on graph neural network of the present invention, the following embodiment shows the experimental results on a real industrial data set. At the same time, a comparative experiment was conducted based on the traditional scheduling method. By comparing the scheduling effects under the same production line conditions, the improvement and superiority of the present invention in the actual production environment were evaluated.

[0053] The embodiment of the present invention is carried out on a micro switch production line. The production line covers more than ten types of equipment nodes and multiple process sites. The data set collects the equipment status, load, task allocation and coupling relationship between equipment. Through the collaborative work of the graph structure construction module, the node weight aggregation module, the multi-scale state prediction model and the nonlinear global optimization scheduling module, intelligent scheduling and fault prevention maintenance can be achieved.

[0054] In the embodiment, the node graph structure is composed of static attributes such as equipment status, load conditions and task allocation, and the process connection and coupling relationship between nodes are dynamic attributes. Then, based on the aggregation weight matrix generated by the adaptive weight aggregation algorithm, dynamic weights are assigned to the load and state parameters of the key nodes. For example, the load parameter weight is set to 0.5, and the task state weight is set to 0.3 to ensure the importance of the key node in scheduling.

[0055] The state prediction model further predicts the state change of the equipment, so that the scheduling system can foresee the state change trend of the equipment and make advance adjustments in the scheduling strategy. In an embodiment of the present invention, the global objective weight of the optimization function is dynamically adjusted based on resource load and equipment idle time, making the scheduling result not only more accurate, but also significantly improving equipment utilization.

[0056] The comparison adopts the traditional rule-based scheduling method, in which the scheduling process is based on static task allocation and fixed equipment resource configuration, and does not involve state prediction or feedback correction mechanism. This method is run on a real data set of the same production line to evaluate the gap with the scheduling method of the present invention in terms of production line efficiency and maintenance rate.

[0057] The test focuses on the following core indicators:

[0058] Production line task completion rate: measures the actual completion of each task by the scheduling method, and records the total amount of tasks completed within the specified time window.

[0059] Equipment utilization rate: measures the utilization degree of equipment in the production line, using the ratio of equipment operating time to total time as an indicator.

[0060] Fault frequency: Based on the system's fault detection function, the average number of failures of the detection equipment during the scheduling process.

[0061] Scheduling stability: The stability of the scheduling scheme in different task scenarios is measured by the variance of task completion time.

[0062] The detection method uses the data recording function of the real-time monitoring system of the production line to record the operating status of the equipment, task completion status, and number of faults in different time periods, and verifies the effectiveness of the intelligent maintenance module with fault repair records.

[0063] The following table shows the average values ​​of various test indicators of the embodiments and comparative examples to verify the advantages of the method of the present invention in multiple key indicators:

[0064] index Method of the present invention Comparative Example Task completion rate (%) 96.2 87.5 Equipment utilization rate (%) 89.4 72.1 Failure frequency (times) 2.1 5.8 Scheduling stability (variance) 4.2 9.7

[0065] In terms of task completion rate, the intelligent scheduling method of the present invention significantly outperforms the traditional method's 87.5% with a high completion rate of 96.2%. This difference is mainly due to the fact that the present invention dynamically adjusts task allocation through real-time prediction and feedback correction, ensuring the smooth completion of key tasks. In terms of equipment utilization, the method of the present invention also shows superiority, reaching a high utilization rate of 89.4%, which is nearly 20% higher than the control group's 72.1%. This is because the present invention makes equipment resource allocation more reasonable through node weight aggregation and nonlinear optimization scheduling, reduces resource waste, and significantly improves the overall efficiency of the production line.

[0066] In terms of fault frequency, the average number of faults in the traditional method is 5.8, while the intelligent maintenance module of the present invention only records 2.1 faults. This shows that through real-time status prediction and intelligent maintenance triggering, the present invention can perform preventive maintenance before potential faults occur, significantly reducing the failure rate. In terms of scheduling stability, the variance of the task completion time of the present invention is 4.2, while that of the traditional method is 9.7. This result shows the efficiency and robustness of the scheduling scheme of the present invention, which can cope with scheduling needs under different load conditions.

[0067] In summary, through the comparison between the embodiments and the comparative examples, it can be seen that the present invention shows significant advantages in multiple key indicators such as task completion rate, equipment utilization rate, fault occurrence frequency and scheduling stability.

[0068] It should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An intelligent scheduling method for micro switch production line based on graph neural network, characterized in that: include: A node graph structure is constructed based on production line equipment, workstations and work-in-progress, with the equipment status, location and task allocation as static attributes and the relationship and interaction information between nodes as dynamic attributes, to obtain the attribute set of the nodes and construct a node graph of the production line; An aggregation weight matrix is ​​generated based on an adaptive node weight aggregation algorithm to dynamically characterize the real-time status of the node; Based on the aggregate weight matrix, a multi-scale state prediction model is generated to obtain a state prediction matrix for predicting a state change trend of the device; Based on the state prediction matrix, a nonlinear global optimization scheduling strategy is constructed to further optimize the task allocation and resource allocation in the production line to obtain a global optimal scheduling solution; Based on the global optimal scheduling scheme, a hierarchical feedback correction mechanism is constructed to further dynamically correct the scheduling results to improve the stability and robustness of the scheduling; Based on the hierarchical feedback correction mechanism, an intelligent maintenance trigger signal is generated to identify potential faults and trigger preventive maintenance.

2. The intelligent scheduling method for micro switch assembly production line based on graph neural network according to claim 1 is characterized in that: The adaptive node weight aggregation algorithm assigns a dynamic weight w to each state parameter in the device state set. k , get the comprehensive status value of the node where w k Indicates the dynamic importance of state parameters, s k Indicates the current state of the node, and n is the number of states.

3. The intelligent scheduling method for micro switch assembly production line based on graph neural network according to claim 2 is characterized in that: The multi-scale state prediction model is based on the node aggregation weight matrix W agg And the dependency function f(p ij (t), s j ), generate the state prediction matrix P at time t t , where f(p ij (t), s j ) represents the state dependence of node i on node j, and further obtains the predicted state p ij (t+1)=α·W agg (v i )·f(p ij (t), s j ).

4. The intelligent scheduling method for micro switch assembly production line based on graph neural network according to claim 3 is characterized in that: The nonlinear global optimization scheduling strategy is based on the state prediction matrix P t and the global optimization function F opt , and further obtain the optimal scheduling solution, where the optimization function represents the task execution cost, m is the number of tasks, and n is the number of nodes.

5. The intelligent scheduling method for micro switch assembly production line based on graph neural network according to claim 4 is characterized in that: The hierarchical feedback correction mechanism adjusts the scheduling scheme by generating a feedback correction matrix H, so that the scheduling stability and robustness are further improved. β is the feedback coefficient.

6. The intelligent scheduling method for micro switch assembly production line based on graph neural network according to claim 5 is characterized in that: The intelligent maintenance module generates a maintenance trigger matrix based on the feedback correction matrix Used for preventive maintenance of equipment, where m i (t+1) = γ·h ii (t+1)·p ii (t+1), when m i When (t+1)>threshold, the maintenance of device i is triggered.

7. The method for intelligent scheduling of a micro switch production line based on a graph neural network according to any one of claims 1 to 6, characterized in that: The intelligent scheduling method further includes periodically updating the maintenance trigger threshold threshold through a scheduling feedback mechanism, so as to adjust the maintenance trigger condition according to the real-time status of the production line to implement an adaptive fault maintenance strategy.

8. A system using the micro switch production line intelligent scheduling method according to any one of claims 1 to 7, characterized in that: include: A graph structure building module, used to build the node graph structure based on static and dynamic properties of production line equipment; Adaptive node weight aggregation module, used to generate the aggregation weight matrix W based on the real-time status of the node agg ; A multi-scale state prediction module for generating a state prediction matrix P based on the aggregation weight matrix t ; A nonlinear global optimization scheduling module is used to generate an optimal scheduling solution based on the state prediction matrix. A hierarchical feedback correction module, used to generate a feedback correction matrix H based on the scheduling scheme; An intelligent maintenance module is used to generate a maintenance trigger matrix based on the feedback correction matrix And start performing preventive equipment maintenance.

9. The micro switch production line intelligent scheduling system according to claim 8 is characterized in that: The adaptive node weight aggregation module also includes a dynamic update mechanism for automatically adjusting the node weight coefficient w based on the change of production line status. k , to ensure the dynamic adaptability of the aggregation weight matrix.

10. The micro switch production line intelligent scheduling system according to claim 8 or 9, characterized in that: The intelligent maintenance module also includes a maintenance update module, which regularly updates the maintenance trigger threshold and trigger coefficient to ensure the real-time and flexibility of fault preventive maintenance of the production line under different conditions.

Images