Control method and system based on intelligent collaborative production

Abstract

The present invention relates to the field of intelligent control technology, and in particular to a control method and system based on intelligent collaborative production. The method comprises: traversing relevant nodes in a production network graph model according to the impact range of task entry, detecting conflicts with equipment resource occupation of existing tasks, and obtaining resource conflict detection results; starting from the time delay prediction value, recursively tracing the affected downstream processes along the dependency edges of the production network graph model to generate an impact propagation path; summarizing the weight values ​​of all affected nodes based on the dynamic quantization results of weights, generating a global perspective evaluation score, and determining the overall impact of the new task entry; if the global perspective evaluation score exceeds a preset threshold, adjusting the entry node and time of the new task, re-executing the resource conflict detection, and obtaining an optimized task entry plan; extracting the process execution sequence and resource allocation plan from the optimized task entry plan, updating the production network graph model, and generating a final collaborative production scheduling plan.

Description

Technical Field

[0001] The present invention relates to the field of intelligent control technology, and in particular to a control method and system based on intelligent collaborative production. Background Art

[0002] In complex production scenarios like discrete manufacturing and process industries, production tasks often involve strong dependencies between multiple processes and the sharing and scheduling of limited resources (such as equipment, manpower, and materials). Traditional production control methods, typically based on static plans and linear scheduling mechanisms, struggle to adapt to real-time changes in task demands. This is especially true when faced with unexpected order insertions, process delays, and resource conflicts. These methods can be slow to respond and inflexible, leading to production bottlenecks, wasted resources, and reduced overall efficiency.

[0003] In recent years, with the development of the Industrial Internet and intelligent manufacturing, the real-time collection and system modeling capabilities of production data have been continuously improved, and data-driven scheduling optimization methods have become a research hotspot. However, existing technologies still have the following shortcomings: (1) There is a lack of graph structure modeling of production process dependencies, which cannot dynamically reflect the path dependencies and influence propagation characteristics between processes; (2) The task insertion process is only based on local resource status judgment, and it is impossible to evaluate the chain effect of the introduced tasks from a global perspective; (3) The conflict detection and scheduling optimization links are separated from each other, and there is a lack of a unified dynamic feedback mechanism, which leads to lag or failure of the optimization solution.

[0004] In addition, although some current methods introduce intelligent algorithms such as linear programming, neural networks or clustering models, they are often limited to a single link and cannot form a closed-loop linkage with key processes such as task entry, conflict adjustment, and scheduling execution in actual production scenarios. It is difficult to achieve real-time, stable and efficient scheduling control in an environment where multi-task collaboration and resource constraints coexist.

[0005] Therefore, there is an urgent need for a collaborative production control method that can integrate graphical model analysis, influence propagation quantification, intelligent algorithm optimization and dynamic scheduling execution to improve the production system's responsiveness to dynamic task changes and overall collaborative scheduling efficiency. The above-mentioned problem is precisely the technical difficulty that the present invention aims to solve. Summary of the Invention

[0006] The present invention provides a control method based on intelligent collaborative production, which mainly includes:

[0007] Step S1: Obtain dependency data between processes in the production system and construct a production network graph model containing process nodes and dependency edges;

[0008] Step S2: according to the entry request of the new task, determine the process and resource requirements required for the new task, identify the nodes occupied by the new task in the production network graph model, and analyze the impact range of the task entry;

[0009] Step S3: Based on the impact range, detect resource conflicts between the new task and the existing tasks, and predict the time delay that may be caused by the introduction of the new task based on the process time and priority of the conflicting nodes;

[0010] Step S4: Based on the time delay, recursively analyze the affected downstream processes in the production network graph model, determine the impact propagation path, and quantify the impact weight of each node to obtain a global impact assessment result;

[0011] Step S5: Based on the global impact assessment results, adjust the entry node or time of the new task, re-perform resource conflict detection, and obtain the optimization results;

[0012] Step S6: Based on the optimization results, a collaborative production scheduling solution including a process execution sequence and a resource allocation plan is generated, and the production network graph model and the scheduling execution plan are updated.

[0013] As a preferred technical solution of the present invention, in step S1, the production network graph model is constructed, including:

[0014] Dependency data between processes are extracted from the production system database. Directed graph modeling is used to map process identifiers into nodes and dependency relationships into directed edges to form an initial production network diagram. Circular dependencies are eliminated through topological sorting to obtain an acyclic network structure. Edge weights are recorded using an adjacency matrix, and a depth-first search is performed to determine the critical path and bottleneck processes. Matrix accumulation operations are performed on the dependency strength of each process within the critical path to obtain a dependency strength distribution. Based on the dependency strength distribution, clustering is used to group process nodes to determine the modular structure of the production network.

[0015] As a preferred technical solution of the present invention, in step S2, obtaining the task entry impact range includes:

[0016] Parse the new task description, use the process parsing algorithm to decompose the new task into sub-processes, and clarify the resource requirements of each sub-process; identify the nodes occupied by the new task in the production network graph model, and analyze the overlap and resource conflicts between the nodes and existing processes; for overlapping or conflicting nodes, adjust the resource allocation plan through the resource allocation algorithm, and update the node occupancy status on the network graph; combine the network analysis algorithm to determine the node conflict situation. If a conflict is detected, dynamically adjust the task entry point and update the production network model; based on the updated network model, calculate and analyze the impact propagation path to determine the impact range of the task entry on the production network.

[0017] As a preferred technical solution of the present invention, in step S4, obtaining the influencing propagation path includes:

[0018] The initial affected processes are determined from the time delay prediction value, and the depth-first search algorithm is used to traverse the dependency edges of the production network graph model. The downstream affected processes are recursively tracked and summarized, and the process execution order is obtained based on the process dependency relationship. If a circular dependency is found, it is adjusted through the topological sorting algorithm to ensure a loop-free order. Under the loop-free process sequence, the propagation path weight is calculated based on the weighted directed graph model, the cumulative effect of the delay is derived, and the degree of impact of each downstream process is determined based on the process priority. Finally, the shortest path algorithm is used to optimize and determine the complete impact propagation path.

[0019] As a preferred technical solution of the present invention, in step S4, quantifying the influence weight of each node includes:

[0020] For each node in the influencing propagation path, the propagation distance from the starting node to the current node is obtained, and the corresponding influence weight is calculated using the exponential decay function to obtain a weight set; if the weight is lower than the preset threshold, the node is eliminated and the propagation path is updated, and the weight of the streamlined path is recalculated to form an optimized weight set; the nodes are grouped through the clustering algorithm, the mean weight of each group is calculated, the dynamic quantization result of the weight is generated, and the final weight quantization result is output.

[0021] As a preferred technical solution of the present invention, obtaining the optimization result includes:

[0022] When the global perspective evaluation score in the global impact assessment result exceeds the preset threshold, the current entry node and time of the task are obtained, and the task priority and resource occupancy are analyzed to determine the feasible adjustment node and time. Then, based on the adjustment plan, the existing resource allocation data is obtained, and the degree of resource conflict is evaluated using the linear programming algorithm to obtain the conflict detection result; if resource conflict still exists, the entry node and time are adjusted iteratively and the resource allocation is recalculated until a conflict-free task entry plan is determined, the execution efficiency index is collected, and its relationship with resource utilization is analyzed through the support vector machine algorithm to optimize the task allocation parameters; if the parameter meets the preset execution efficiency threshold, the simulation scheduling algorithm is used to generate the task scheduling execution sequence, and its change trend is analyzed in combination with real-time resource utilization data. Finally, based on time series analysis, future resource demand is predicted to obtain a dynamically adjustable task allocation plan.

[0023] As a preferred technical solution of the present invention, the production network graph model and the scheduling execution plan are updated and obtained, including: extracting the process execution sequence and resource allocation plan based on the optimized task entry plan, extracting the priority and resource utilization of each process through data analysis technology, and constructing a structured task data set; if the priority of any process exceeds a preset threshold, optimizing the resource allocation plan using a linear programming algorithm to obtain a more optimal resource configuration; then, based on the optimization results, updating the production network graph model to form a network topology structure that reflects the process execution sequence;

[0024] The scheduling sequence is extracted from the topological structure and combined with the simulated annealing algorithm to generate a preliminary collaborative production scheduling plan; if the production efficiency index corresponding to the plan is lower than the set threshold, the process priority is adjusted iteratively and the scheduling execution plan is updated.

[0025] In a second aspect, the present invention further provides a control system based on intelligent collaborative production, for implementing the above method, the system comprising:

[0026] A construction unit is used to obtain dependency data between processes in the production system and construct a production network graph model containing process nodes and dependency edges;

[0027] An analysis unit, configured to determine the process and resource requirements of the new task according to the entry request of the new task, identify the nodes occupied by the new task in the production network graph model, and analyze the impact range of the task entry;

[0028] a prediction unit, configured to detect conflicts between new tasks and existing tasks in terms of resource usage based on the impact range, and to predict possible time delays caused by the introduction of the new tasks based on the process times and priorities of the conflicting nodes;

[0029] a computing unit configured to recursively analyze the affected downstream processes in the production network graph model based on the time delay, determine the impact propagation path, and quantify the impact weight of each node to obtain a global impact assessment result;

[0030] The acquisition unit is used to adjust the entry node or time of the new task based on the global impact assessment results, re-detect resource conflicts, and obtain optimization results;

[0031] An updating unit is used to generate a collaborative production scheduling solution including a process execution sequence and a resource allocation plan based on the optimization result, and to update the production network graph model and the scheduling execution plan.

[0032] In a third aspect, the present invention further provides a computer-readable storage medium having instructions stored thereon, and the instructions implement the above method when executed by a processor.

[0033] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0034] The present invention discloses an intelligent production scheduling optimization method based on a production network graph model. This method addresses the resource conflicts and time delays caused by the entry of new tasks in complex production systems. By constructing a production network graph containing process nodes and dependency edges, the process and resource requirements of the new task are analyzed, the occupied nodes are accurately identified, and the impact range is determined. Upon resource conflict detection, if a conflict is found, the method calculates the time offset based on the process time and priority of the conflicting node, predicts the delay value, and recursively tracks the downstream process along the dependency edge to generate an impact propagation path. An exponential decay function is used to quantify the impact weight corresponding to the propagation distance, and the weight values ​​are aggregated to generate a global perspective evaluation score. If the score exceeds a threshold, the method adjusts the task entry node and time, reoptimizes the scheduling plan, and finally updates the production network graph to generate a collaborative production scheduling plan. Through dynamic quantification of impact weights and global evaluation, the present invention significantly improves production scheduling efficiency, reduces the risk of resource conflicts and time delays, and achieves intelligent management of complex production systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 This is a flow chart of a control method based on intelligent collaborative production in an embodiment of the present invention.

[0036] Figure 2 This is a structural diagram of a control system based on intelligent collaborative production in an embodiment of the present invention. DETAILED DESCRIPTION

[0037] In order to make the objectives, technical solutions and advantages of the present invention more clear, the present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0038] like Figure 1 In this embodiment, a control method based on intelligent collaborative production may specifically include:

[0039] Step S1: Obtain dependency data between processes in the production system and construct a production network graph model containing process nodes and dependency edges;

[0040] The construction of the production network graph model includes:

[0041] Dependency data between processes is extracted from the production system database. Directed graph modeling is used to map process identifiers into nodes and dependency relationships into directed edges to form an initial production network diagram. Circular dependencies are eliminated through topological sorting to obtain an acyclic network structure. The adjacency matrix is ​​used to record edge weights, and combined with a depth-first search algorithm, the critical path and bottleneck processes are determined. Matrix accumulation operations are performed on the dependency strength of each process within the critical path to obtain a dependency strength distribution. Based on the dependency strength distribution, clustering is used to group process nodes to determine the modular structure of the production network.

[0042] Specifically, in intelligent collaborative production scenarios, to address the scheduling challenges of multiple tasks and multiple processes, a production network graph model is constructed to structure the process dependencies. Specifically, inter-process dependency data is extracted from the production system database, and each process identifier is used as a node in the graph. The process sequence is mapped as directed edges to construct an initial production network graph model in the form of a directed graph. To avoid scheduling logic failures caused by cyclic dependencies, a topological sort is used to transform the graph structure into a directed acyclic structure, removing or marking loop paths to ensure the parsability of the scheduling process. After obtaining the acyclic structure, the dependency strength and path weights between nodes are expressed in the form of an adjacency matrix. A depth-first search algorithm is then used to identify the longest path group, from which the critical path and bottleneck process with the greatest impact on the overall production cycle are extracted. Matrix accumulation operations are then performed on the dependencies between processes in the critical path to quantify the distribution of the dependency strength of each process on subsequent links. Finally, based on this dependency strength data, a clustering algorithm is applied to automatically group nodes, thereby achieving a modular structure for the production network, providing a structural foundation and algorithmic support for the rapid entry of new tasks, resource conflict detection, and scheduling optimization.

[0043] Step S2: according to the entry request of the new task, determine the process and resource requirements required for the new task, identify the nodes occupied by the new task in the production network graph model, and analyze the impact range of the task entry;

[0044] Among them, the acquisition of the task entry impact range includes:

[0045] Decompose the new task into sub-processes and clarify the resource requirements of each sub-process; identify the nodes occupied by the new task in the production network graph model, and analyze the overlap and resource conflicts between the nodes and existing processes; for overlapping or conflicting nodes, adjust the resource allocation plan through the resource allocation algorithm and update the node occupancy status on the network graph; combine the network analysis algorithm to determine the node conflict situation. If a conflict is detected, dynamically adjust the task entry point and update the production network model; based on the updated network model, calculate and analyze the impact propagation path to determine the impact range of the task entry on the production network.

[0046] Specifically, in intelligent collaborative production scenarios, the process of obtaining the impact scope of task entry uses a process parsing algorithm to structurally decompose the task to be put into production, transforming the task into multiple sub-processes with clear resource requirements. The sub-processes are then mapped to corresponding nodes in the production network graph model, establishing structural associations with existing nodes. By comparing the overlap between the new task node and the existing node in the process identification and time resource dimensions, resource occupation conflicts are identified. The resource allocation algorithm is used to evaluate the utilization and allocation weight of the conflicting resources, thereby dynamically adjusting the resource allocation plan and updating the node occupation status. Based on resource allocation, the updated production network graph structure is used to determine whether bottlenecks and blockages have formed in the node hierarchy. If conflicts exist, resource overlap is avoided by reselecting the entry node and adjusting the task start time, and the production network graph model is structurally repaired. Based on the dependency paths between nodes in the network, the impact propagation path calculation model is used to simulate and predict the downstream propagation range of delays caused by task entry, thereby quantifying the transmission effect of task entry in the entire production process. For example, in an electronic assembly line, when a high-priority maintenance task needs to be inserted into the production system, this method can identify its impact on the delay of the existing process and complete the task insertion without disrupting the overall schedule, thereby improving the overall resource responsiveness and scheduling flexibility of the system; the above technical solution realizes the structural identification and feedback correction of the chain effects caused by the dynamic task insertion in the production system.

[0047] Step S3: Based on the impact range, detect resource conflicts between the new task and the existing tasks, and predict the time delay that may be caused by the introduction of the new task based on the process time and priority of the conflicting nodes;

[0048] Among them, resource conflict detection includes: traversing the relevant nodes in the task impact range in the production network graph model, and determining the resource occupancy status of each node; for nodes with equipment or resource occupancy, comparing the task allocation requirements with the actual resource utilization, and judging and summarizing the conflicting nodes based on the preset threshold; analyzing the network topology structure according to the set of conflicting nodes to determine a feasible resource scheduling path; if the resource scheduling path is feasible, adjusting the task allocation based on the network topology, using a heuristic algorithm to verify the optimized resource configuration, and outputting the final resource conflict detection result.

[0049] Specifically, by traversing the relevant nodes in the production network graph model that are affected by the new task one by one, dynamic identification and response to resource conflict risks are achieved. This process is based on the resource occupancy status of the task node, and quantitatively compares the task allocation demand with the resource utilization rate of the current node. It determines whether the resource is overloaded based on the threshold standard set by the system, thereby forming a conflict judgment mechanism; all nodes that meet the conflict conditions are collected to form a conflict node set, which has topological connection characteristics in the graph structure. By analyzing the structural distribution of the set in the network and combining the dependency relationship between nodes, possible resource readjustment paths are identified; if the path maintains accessibility and execution continuity, the task allocation direction is guided according to the network structure information to form a preliminary optimized configuration; in order to make the configuration result operational and system stable, the current resource allocation plan is verified according to the heuristic algorithm to determine whether it meets the task process execution constraints and equipment operation boundaries, and then outputs the feasible The resource allocation plan of the execution is taken as the final result of resource conflict detection; wherein, the task process execution constraints include at least the order of task execution, priority and execution time, and the above-mentioned equipment operation boundaries include at least the execution boundary, i.e., the equipment operation time period, the maintenance boundary, i.e., the equipment cannot be scheduled during the scheduled time period due to maintenance, and the energy consumption boundary, i.e., high-power consumption equipment cannot be operated during a specific time period; for example: in a certain production system, there are three tasks A, B, and C, all of which need to use the same equipment D. The system finds that the new task E is planned to be inserted in the time overlapping with task B. The heuristic algorithm can determine whether B can be postponed according to the priority rule, or determine whether it can be transferred to another idle equipment according to the resource utilization rate, and adjust the time window without destroying the dependent path, so that E can be smoothly inserted into the system to avoid conflicts; the above-mentioned technical solution realizes the accurate identification and dynamic mitigation of resource conflicts through the coordinated optimization of scheduling strategies, and solves the problems of untimely task conflict processing and strong scheduling rigidity in the existing system.

[0050] Furthermore, obtaining the time delay prediction value includes:

[0051] When a resource conflict is detected, the process time and priority data are obtained from the conflicting node, the resource occupancy is analyzed, and the task execution order is adjusted using priority sorting based on the degree of resource competition and priority between the new task and the existing task. The time offset introduced by the new task is calculated to obtain a preliminary time delay prediction value; if the prediction value exceeds the preset threshold, the offset is corrected through linear regression based on the historical scheduling data to obtain the final time delay prediction value.

[0052] Specifically, based on the process resource conflicts that may be caused by task entry, the system extracts process time and priority data of conflicting nodes to identify the current resource occupancy status, analyzes the degree of resource competition between new and existing tasks, and adjusts the task execution order based on a priority sorting algorithm. This calculates the time offset of the new task in the existing scheduling plan and forms a preliminary prediction of time delay. When this prediction exceeds the time tolerance threshold set by the scheduling system, historical scheduling data is further introduced to construct a regression model based on time offset and historical execution status. A linear regression algorithm is used to correct the preliminary prediction offset and output a time delay prediction that better reflects the actual system status. This time delay prediction serves as an input parameter for scheduling optimization, guiding task reordering and resource reallocation in the resource allocation module. It also works in conjunction with the current production network graph model to update the scheduling plan and achieve dynamic balance of resource allocation between tasks. In actual production processes, such as in continuous processing and manufacturing scenarios, if a new task requires the use of key equipment, this mechanism can identify its transmission effect on downstream bottleneck processes and proactively adjust the scheduling order to prevent local blockages from spreading to the entire process, thereby improving production line throughput and plan stability.

[0053] Step S4: Based on the time delay, recursively analyze the affected downstream processes in the production network graph model, determine the impact propagation path, and use a preset function to quantify the impact weight of each node to obtain a global impact assessment result;

[0054] Among them, factors that affect the acquisition of propagation paths include:

[0055] The initial affected processes are determined from the time delay prediction value, and the depth-first search algorithm is used to traverse the dependency edges of the production network graph model. The downstream affected processes are recursively tracked and summarized, and the process execution order is obtained based on the process dependency relationship. If there is a circular dependency, it is adjusted through a topological sorting algorithm. Under the acyclic process sequence, the propagation path weight is calculated based on the weighted directed graph model, the cumulative effect of the delay is derived, and the degree of impact of each downstream process is judged based on the process priority. Finally, the shortest path algorithm is used to optimize and determine the complete impact propagation path.

[0056] Specifically, based on the process dependency edges between nodes, a depth-first search algorithm is used to recursively track the downstream process nodes affected by time delays, and an ordered process execution sequence is automatically generated based on the topological relationship between processes. When a circular dependency between nodes is identified, a topological sorting algorithm is applied to acyclically process the graph structure to avoid path dead loops caused by dependency conflicts. To further quantify the hierarchical transmission effect of delays, a weighted directed graph model is constructed, and edge weights are constructed based on the dependency strength and distance information between processes. The cumulative calculation of path propagation impact is performed on the graph structure. In this process, a process priority index is introduced as a weighted adjustment factor to improve the ability to identify key nodes. With the help of the shortest path algorithm, the influence range and propagation depth of multiple feasible paths are comprehensively optimized, the scheduling main line is accurately located, and the impact propagation path with timeliness and operability is output to provide data support for subsequent dynamic scheduling and resource reallocation.

[0057] Furthermore, the influence weight of each node is quantified, including:

[0058] For each node in the impact propagation path, the propagation distance from the starting node to the current node is obtained, and the corresponding influence weight is calculated using the exponential decay function to obtain a weight set; if the weight is lower than the preset threshold, the node is eliminated and the propagation path is updated, and the weight of the streamlined path is recalculated to form an optimized weight set; the nodes are grouped by the clustering algorithm, the mean weight of each group is calculated, and the dynamic quantization result of the weight is generated. The groups with a mean value lower than the threshold are marked as low-impact groups, and the final weight quantization result and low-impact group label are output.

[0059] Specifically, for each node in the impact propagation path, the system first obtains the propagation distance from the starting node to the current node, and uses the exponential decay function The influence weight of the node is calculated, where d is the propagation distance and k is the set attenuation coefficient, and an initial weight set is formed based on the weight of each node; to improve the evaluation accuracy, nodes with weights lower than the preset threshold are eliminated, thereby updating the propagation path, and on this basis, the propagation distances and influence weights of the remaining nodes are recalculated to obtain the optimized weight set; based on the set, the nodes are grouped by clustering, and the weight mean of each group is calculated separately to form a dynamic quantization result of the weight; if the weight mean of a cluster group is lower than the influence threshold, it is marked as a "low-impact group"; otherwise, it is marked as a "high-impact group". The above technical solution realizes the quantitative identification and structured expression of key influence paths in the production network.

[0060] Furthermore, the acquisition of global impact assessment results includes:

[0061] The dynamically quantized weight value is obtained from each node, and a global perspective evaluation score is formed through weighted summation. If the score exceeds the preset threshold, the nodes are prioritized and high-priority nodes are screened. The support vector machine algorithm is used to classify the node impact. Combined with the operating status of key influencing nodes, a linear regression model is used to predict the changing trend of system performance after the introduction of tasks, and the task scheduling strategy is adjusted accordingly. After optimization, the node weight and global evaluation score are recalculated until the score is lower than the threshold.

[0062] Specifically, in a multi-task parallel production scheduling system, in order to solve the problem of downstream process chain delays and overall resource allocation efficiency decline caused by the entry of new tasks, on the basis of the constructed production network diagram, the influence weight of each node is obtained through dynamic quantification based on the path propagation distance, and then the global perspective evaluation score is generated through weighted summation, that is, ,in, Represents the influence weight of the i-th node. This score characterizes the comprehensive impact of the current task entry on the scheduling balance and resource utilization of the entire system, and serves as the basis for judging the optimization of the task entry strategy. When the evaluation score exceeds the threshold, all nodes are automatically sorted according to the process priority, and high-priority nodes are screened out. Combined with their state characteristics such as resource occupancy and scheduling tension, secondary classification processing is performed based on the degree of node influence to identify the core influencing node set that constitutes the critical path. The operating parameters corresponding to the node set are further input into the linear regression model, and combined with historical resource load and scheduling response data, the impact of the task entry on the entire system is predicted. The changing trends of performance indicators, such as average completion time, equipment utilization rate, and waiting time of bottleneck processes; when the prediction results show that the scheduling target cannot be achieved, the task scheduling strategy is adjusted based on the regression analysis results, and the weight and evaluation score of each node are recalculated to achieve iterative convergence of the scheduling plan until the overall scheduling cost is lower than the acceptable threshold; in discrete manufacturing scenarios, if a high-priority task needs to be inserted before the resource-constrained welding process, its potential coupling effect on subsequent painting, assembly and other nodes can be identified in advance, and then through scheduling strategy adjustment and machine replacement configuration optimization, the task can be cut in without causing overall efficiency reduction.

[0063] Step S5: Based on the global impact assessment results, if the impact exceeds a preset threshold, the entry node or time of the new task is adjusted, and resource conflict detection and optimization are re-performed until the preset conditions are met and the optimization result is obtained;

[0064] The acquisition of the optimization results includes: when the global perspective evaluation score in the global impact evaluation result exceeds a preset threshold, obtaining the current entry node and time of the task, analyzing it in combination with the task priority and resource occupancy, determining feasible adjustment nodes and times, obtaining existing resource allocation data based on the adjustment plan, and using a linear programming algorithm to evaluate the degree of resource conflict to obtain a conflict detection result; if resource conflicts still exist, adjusting the entry node and time in an iterative manner and recalculating resource allocation until a conflict-free task entry plan is determined; on this basis, collecting execution efficiency indicators, analyzing their relationship with resource utilization, and optimizing task allocation parameters; if the parameters meet the preset execution efficiency threshold, using simulation scheduling to generate a task scheduling execution sequence, and analyzing its changing trend in combination with real-time resource utilization data, and finally predicting future resource requirements based on time series analysis to obtain a dynamically adjustable task allocation plan as the optimization result.

[0065] Specifically, when the global perspective evaluation score of the global impact assessment result exceeds the preset threshold, the system obtains the entry node and entry time of the current task, and performs data fusion processing on it with the task priority and the resource occupancy corresponding to the node to form a set of constraint conditions; this set is used to establish a linear optimization model for resource allocation, solve the conflict status under the current resource configuration through a linear programming algorithm, and output the conflict detection result. If a conflict exists, the task entry position and time parameters are adjusted, the resource allocation model is rebuilt, and the solution process is repeated until the model constraints meet the feasible domain of the conflict-free solution, thereby forming a new task entry plan that meets the system resource boundary conditions. Based on this plan, real-time scheduling execution efficiency indicators are collected during task execution, multi-dimensional data association is performed with known resource utilization data to form a multi-dimensional vector. Based on this multi-dimensional vector, a mapping relationship between scheduling efficiency and resource allocation is established, and the allocation parameters with the smallest error are extracted and fed back to the resource allocation model to further optimize resource utilization. After meeting the preset scheduling efficiency requirements, a simulation scheduling algorithm is used to construct the task execution sequence based on the task dependency order, node priority, and dynamic resource utilization data, generating a scheduling plan with timing constraints and resource coordination capabilities. During the execution of the plan, the resource utilization data collected on site are combined with time series analysis methods to predict resource consumption trends, realizing the dynamic adjustment capability of the task allocation plan in the time dimension. The above technical solution collaboratively solves the conflict detection and scheduling optimization problems caused by the dynamic entry of multiple tasks.

[0066] Step S6: Based on the final optimization result, a collaborative production scheduling solution including a process execution sequence and a resource allocation plan is generated, and the production network graph model and the scheduling execution plan are updated;

[0067] Specifically, it includes: extracting the process execution sequence and resource allocation plan based on the optimized task entry plan, extracting the priority and resource utilization of each process through data analysis technology, and constructing a structured task data set; if the priority of any process exceeds the preset threshold, the linear programming algorithm is used to optimize the resource allocation plan to obtain a better resource configuration; then, based on the optimization results, the production network graph model is updated using the graph theory algorithm to form a network topology structure that reflects the process execution sequence; further extracting the scheduling sequence from the topology structure, and generating a preliminary collaborative production scheduling plan based on the simulated annealing algorithm; if the production efficiency index in the scheduling plan is lower than the set threshold, the process priority is adjusted iteratively and the scheduling execution plan is updated; finally, the task allocation rules are determined according to the adjusted scheduling plan, a collaborative production scheduling plan is generated and a production execution sequence is formed, and the production efficiency index in the execution sequence is written into the database as the performance output basis for the final scheduling result.

[0068] Specifically, this technical solution targets the problem of intelligent collaborative production scheduling with multiple processes and intensive resources. Based on the optimized task entry plan, it extracts the execution order and resource allocation plan of each process in the production task, and converts it into a structured task data set containing process priority and resource utilization through data analysis technology. If there are processes in the task data set whose priority exceeds the preset threshold, the resource allocation is re-optimized to achieve resource load balancing and priority execution of key processes. After the linear programming algorithm obtains the optimized resource allocation result, the production network graph model is dynamically updated with the help of graph theory algorithm, so that the process nodes and dependency edges jointly reflect the new process execution order, forming a topological structure and serving as the basic data for subsequent scheduling generation. During the scheduling generation process, the simulated annealing algorithm is applied to the process sequence in the topological structure. The system searches in sequence to find a scheduling path with better production efficiency indicators in the multi-solution space; if the production efficiency indicator in the generated scheduling plan is lower than the preset threshold, it iterative adjustment is performed based on the process priority in the structured task data set, the scheduling path is updated and the network topology is corrected to improve the overall executable and responsiveness of the system; after the scheduling path is stabilized, a collaborative production scheduling plan is generated according to the updated process sequence and resource allocation, and a clear task execution sequence is formed. At the same time, the production efficiency data in the execution sequence is collected and stored in the database as a structured output for performance evaluation, ensuring that the system scheduling results are verifiable and have continuous optimization capabilities; the above technical solution realizes the balance between precise task insertion and production rhythm, and solves the problem that traditional static scheduling cannot cope with the overall efficiency decline caused by dynamic task changes.

[0069] The present invention also provides a control system based on intelligent collaborative production, which is used to implement the above method, such as Figure 2 As shown, the system includes:

[0070] A construction unit is used to obtain dependency data between processes in the production system and construct a production network graph model containing process nodes and dependency edges;

[0071] An analysis unit, configured to determine the process and resource requirements of the new task according to the entry request of the new task, identify the nodes occupied by the new task in the production network graph model, and analyze the impact range of the task entry;

[0072] a prediction unit, configured to detect conflicts between new tasks and existing tasks in terms of resource usage based on the impact range, and to predict possible time delays caused by the introduction of the new tasks based on the process times and priorities of the conflicting nodes;

[0073] a computing unit configured to recursively analyze the affected downstream processes in the production network graph model based on the time delay, determine the impact propagation path, and quantify the impact weight of each node to obtain a global impact assessment result;

[0074] The acquisition unit is used to adjust the entry node or time of the new task based on the global impact assessment results, re-detect resource conflicts, and obtain optimization results;

[0075] An updating unit is used to generate a collaborative production scheduling solution including a process execution sequence and a resource allocation plan based on the optimization result, and to update the production network graph model and the scheduling execution plan.

[0076] The present invention also provides a computer-readable storage medium having instructions stored thereon, and the above-mentioned method is implemented when the instructions are executed by a processor.

[0077] In summary, the present invention achieves efficient production scheduling in a multi-task dynamic insertion environment by organically coordinating multiple steps such as process dependency modeling, task entry analysis, resource conflict detection, delay prediction and propagation evaluation, and scheduling optimization. Its technical effects are reflected in the following aspects: first, the dependency modeling step maps the process and dependency into a directed graph structure, providing a structured basis for subsequent task entry analysis; the task entry step locates the occupied nodes and defines the impact range by analyzing the task process and resource requirements; the resource conflict detection step identifies resource competition conflicts between tasks and guides them into the time delay prediction link to accurately evaluate potential delay risks; the delay propagation analysis combines the topological structure and propagation function to quantify the degree of impact on downstream processes and form a global evaluation index; when the index exceeds the preset threshold, the system iteratively adjusts the entry node and time, and uses linear programming and simulation scheduling algorithms to optimize resource conflicts and execution efficiency until a scheduling solution with stability and resource coordination is generated; finally, dynamic resource prediction and real-time scheduling adjustment are achieved through time series analysis to ensure the continuity and robustness of the system. The above steps are closely coupled based on data transmission and algorithm logic, forming a closed-loop control mechanism with dynamic linkage between tasks, resources and scheduling, achieving high-precision, scalable and adjustable collaborative production control goals under multi-task concurrency in complex manufacturing systems.

[0078] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A control method based on intelligent collaborative production, characterized in that: The method comprises: Step S1: Obtain dependency data between processes in the production system and construct a production network graph model containing process nodes and dependency edges; Step S2: according to the entry request of the new task, determine the process and resource requirements required for the new task, identify the nodes occupied by the new task in the production network graph model, and analyze the impact range of the task entry; Step S3: Detect resource conflicts between the new task and existing tasks based on the impact range, and predict the time delay caused by the introduction of the new task based on the process time and priority of the conflicting nodes; Step S4: Based on the time delay, recursively analyze the affected downstream processes in the production network graph model, determine the impact propagation path, and quantify the impact weight of each node to obtain a global impact assessment result; obtaining the global impact assessment result includes forming a global perspective assessment score through weighted summation; Step S5: Based on the global impact assessment results, adjust the entry node or time of the new task, re-perform resource conflict detection, and obtain the optimization results; Step S6: Based on the optimization results, a collaborative production scheduling solution including a process execution sequence and a resource allocation plan is generated, and the production network graph model and the scheduling execution plan are updated; The acquisition of the time delay prediction value includes: based on the process resource conflict caused by task entry, by extracting the process time and priority data of the conflicting node, identifying the current resource occupancy status, and analyzing the degree of resource competition between the new task and the existing tasks, adjusting the task execution order based on the priority sorting algorithm, thereby calculating the time offset of the new task in the existing scheduling plan and forming a preliminary prediction value of the time delay; when the prediction value exceeds the time tolerance threshold set by the scheduling system, further introducing historical scheduling data, constructing a regression model based on time offset and historical execution status, and using the linear regression algorithm to correct the preliminary prediction offset value, outputting a time delay prediction value that is more in line with the actual state of the system; In step S4, the acquisition of the influencing propagation path includes: The initial affected processes are determined from the predicted time delay values. A depth-first search algorithm is used to traverse the dependency edges of the production network graph model. The affected downstream processes are recursively tracked and summarized, and the process execution order is determined based on the process dependencies. If a circular dependency is found, a topological sorting algorithm is used to adjust it to ensure an acyclic order. In the acyclic process sequence, the propagation path weights are calculated based on a weighted directed graph model to derive the cumulative effect of delays. The degree of impact of each downstream process is determined based on the process priority. Finally, the shortest path algorithm is used to optimize and determine the complete impact propagation path. In step S5, the optimization result is obtained, including: when the global perspective evaluation score in the global impact evaluation result exceeds the preset threshold, the current entry node and time of the task are obtained, and the task priority and resource occupancy are analyzed to determine the feasible adjustment node and time, and then based on the adjustment plan, the existing resource allocation data is obtained, and the degree of resource conflict is evaluated using a linear programming algorithm to obtain a conflict detection result; if resource conflict still exists, the entry node and time are adjusted in an iterative manner and the resource allocation is recalculated until a conflict-free task entry plan is determined, the execution efficiency index is collected, and its relationship with resource utilization is analyzed by a support vector machine algorithm to optimize the task allocation parameters; if the parameter meets the preset execution efficiency threshold, the task scheduling execution sequence is generated by a simulation scheduling algorithm, and its change trend is analyzed in combination with real-time resource utilization data, and finally future resource demand is predicted based on time series analysis to obtain a dynamically adjustable task allocation plan; In step S4, the influence weight of each node is quantified, including: For each node in the influencing propagation path, the propagation distance from the starting node to the current node is obtained, and the corresponding influence weight is calculated using the exponential decay function to obtain a weight set; if the weight is lower than the preset threshold, the node is eliminated and the propagation path is updated, and the weight of the streamlined path is recalculated to form an optimized weight set; the nodes are grouped through the clustering algorithm, the mean weight of each group is calculated, the dynamic quantization result of the weight is generated, and the final weight quantization result is output.

2. The control method based on intelligent collaborative production according to claim 1, characterized in that: In step S1, the production network graph model is constructed, including: Dependency data between processes are extracted from the production system database. Directed graph modeling is used to map process identifiers into nodes and dependency relationships into directed edges to form an initial production network diagram. Circular dependencies are eliminated through topological sorting to obtain an acyclic network structure. Edge weights are recorded using an adjacency matrix, and a depth-first search is performed to determine the critical path and bottleneck processes. Matrix accumulation operations are performed on the dependency strength of each process within the critical path to obtain a dependency strength distribution. Based on the dependency strength distribution, clustering is used to group process nodes to determine the modular structure of the production network.

3. The method according to claim 1, characterized in that In step S2, the task entry impact range is obtained, including: Parse the new task description, use the process parsing algorithm to decompose the new task into sub-processes, and clarify the resource requirements of each sub-process; identify the nodes occupied by the new task in the production network graph model, and analyze the overlap and resource conflicts between the nodes and existing processes; for overlapping or conflicting nodes, adjust the resource allocation plan through the resource allocation algorithm, and update the node occupancy status on the network graph; combine the network analysis algorithm to determine the node conflict situation. If a conflict is detected, dynamically adjust the task entry point and update the production network model; based on the updated network model, calculate and analyze the impact propagation path to determine the impact range of the task entry on the production network.

4. The control method based on intelligent collaborative production according to claim 1, characterized in that: Updating the production network graph model and scheduling execution plan and obtaining the plan includes: extracting the process execution sequence and resource allocation plan based on the optimized task entry plan, extracting the priority and resource utilization of each process through data analysis technology, and constructing a structured task data set; if the priority of any process exceeds a preset threshold, optimizing the resource allocation plan using a linear programming algorithm to obtain a more optimal resource configuration; then, based on the optimization results, updating the production network graph model to form a network topology structure that reflects the process execution sequence; The scheduling sequence is extracted from the topological structure and combined with the simulated annealing algorithm to generate a preliminary collaborative production scheduling plan; if the production efficiency index corresponding to the plan is lower than the set threshold, the process priority is adjusted iteratively and the scheduling execution plan is updated.

5. A control system based on intelligent collaborative production, used to implement the method according to any one of claims 1 to 4, characterized in that: The system comprises: A construction unit is used to obtain dependency data between processes in the production system and construct a production network graph model containing process nodes and dependency edges; An analysis unit, configured to determine the process and resource requirements of the new task according to the entry request of the new task, identify the nodes occupied by the new task in the production network graph model, and analyze the impact range of the task entry; a prediction unit, configured to detect conflicts between new tasks and existing tasks in terms of resource usage based on the impact range, and to predict time delays caused by the introduction of the new tasks based on process times and priorities of conflicting nodes; a computing unit configured to recursively analyze the affected downstream processes in the production network graph model based on the time delay, determine the impact propagation path, and quantify the impact weight of each node to obtain a global impact assessment result; The acquisition unit is used to adjust the entry node or time of the new task based on the global impact assessment results, re-detect resource conflicts, and obtain optimization results; An updating unit is used to generate a collaborative production scheduling solution including a process execution sequence and a resource allocation plan based on the optimization result, and to update the production network graph model and the scheduling execution plan.

6. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by a processor, the method according to any one of claims 1 to 4 is implemented.

Images