Intelligent production scheduling process connection optimization method based on product structure tree

Abstract

The invention provides an intelligent production scheduling process connection optimization method based on a product structure tree, relates to the field of intelligent manufacturing, and solves the technical problem that process conflicts and resource contention are easy to exist in the prior art. The method comprises the following steps: constructing a dynamic evolution product structure tree; performing process decomposition based on the product structure tree, generating a process sequence and identifying a key path; performing multi-dimensional conflict detection on the process sequence, and performing conflict resolution according to a detection result; the multi-dimensional conflicts comprise space conflicts, resource conflicts and time sequence conflicts; and based on the process sequence after conflict resolution, generating an optimal production scheduling plan through a multi-objective optimization algorithm, collecting data in real time in the production process, and adjusting the optimal production scheduling plan based on the multi-index deviation between the plan and the reality.

Description

Technical Field

[0001] This invention belongs to the field of intelligent manufacturing, specifically a method for optimizing intelligent scheduling and process connection based on a product structure tree. Background Technology

[0002] In complex engineering fields such as manufacturing and construction, optimizing production scheduling processes has always been a key step in improving efficiency and reducing costs. Traditional scheduling methods typically rely on manual experience or isolated information systems, such as Enterprise Resource Planning (ERP) and Manufacturing Execution Systems (MES). These systems often operate independently, leading to fragmented design, process, and manufacturing data, creating information silos and resulting in data inconsistencies, transmission delays, and decision-making errors. This data fragmentation problem not only increases coordination costs but also reduces the accuracy and reliability of production scheduling plans.

[0003] Furthermore, traditional scheduling methods often rely on manual intervention to identify and adjust conflicts during the production process. Moreover, in terms of production planning optimization, traditional manual scheduling is highly subjective, easily overlooking complex constraints, leading to low planning efficiency, uneven resource allocation, or insufficient risk management. Summary of the Invention

[0004] This application provides an intelligent scheduling process connection optimization method based on a product structure tree, which solves the problems of existing methods. Technical issues.

[0005] To achieve the above objectives, this application adopts the following technical solution: Firstly, a method for optimizing intelligent scheduling processes based on a product structure tree is provided, including: Construct a dynamic evolution product structure tree; the product structure tree is a multi-level tree data structure, and the root node of the structure tree corresponds to the whole product, and the child nodes are organized hierarchically according to functional systems, manufacturing areas and resource units. Each node stores design attributes, process parameters and real-time status data. Based on the product structure tree, process decomposition is performed to generate process sequences and identify critical paths; The process sequence is subjected to multi-dimensional conflict detection, and conflict resolution is performed based on the detection results; the multi-dimensional conflicts include spatial conflicts, resource conflicts, and temporal conflicts. Based on the process sequence after conflict resolution, an optimal production schedule is generated through a multi-objective optimization algorithm. Data is collected in real time during the production process, and the optimal production schedule is adjusted based on the deviation of multiple indicators between the plan and the actual situation.

[0006] Based on the above technical solutions, the intelligent scheduling process connection optimization method based on the product structure tree provided in this application achieves full-process integration of design, process, and manufacturing data by constructing a dynamically evolving product structure tree, effectively solving the problem of information silos. The process decomposition and multi-dimensional conflict detection mechanism based on the product structure tree can identify and resolve spatial, resource, and temporal conflicts in advance, reducing uncertainty and rework risks in the production process. The optimal production schedule is generated by using a multi-objective optimization algorithm, achieving a comprehensive balance of schedule, cost, and risk, overcoming the subjectivity and limitations of traditional manual scheduling. Finally, the dynamic adjustment mechanism based on real-time data enables the system to have strong adaptive capabilities, quickly responding to changes in the production site, ensuring the efficiency and stability of the scheduling plan, thereby improving overall scheduling efficiency, plan feasibility, and system robustness.

[0007] Furthermore, the process of constructing the dynamic evolution product structure tree includes: Define each node in the product structure tree as a unique identifier (PSN), in the format PSN-{Product Code}-{Region Code}-{Process Type}; where the Product Code is the unique code of the product, the Region Code is the region identifier, and the Process Type is the process category to which the node belongs. Each node in the product structure tree is defined to contain a set of static attributes, a set of dynamic attributes, and a set of evolution rules; among them, The static attribute set includes design specifications, geometric constraints, and quality requirements; the dynamic attribute set includes current state, estimated duration, and resource requirements; and the evolution rule set includes preconditions, triggering timing, and evolution logic. Preconditions are used to define the evolutionary dependencies of nodes, triggering timing is based on event-driven settings, and evolution logic is used to specify node classification, merging, or attribute update rules. The product structure tree is automatically evolved based on the product's production stage to obtain a dynamically evolving product structure tree.

[0008] Furthermore, the automatic evolution means that nodes are organized according to functional systems during the design phase, reorganized according to manufacturing regions during the process planning phase, and mapped to resource units during the production execution phase. The evolution rules for node evolution are as follows: ;in, E represents the triggering action, E represents the event set, including process completion, resource change or order adjustment, and A represents the action set, including node splitting, node merging or attribute updating.

[0009] Furthermore, the design attributes include design specifications, geometric constraints, and quality requirements; wherein, design specifications include geometric dimensions, material type, tolerance range, and surface treatment requirements; geometric constraints include assembly relationships, spatial boundaries, and motion trajectories; and quality requirements include inspection standards, allowable defect rates, and testing methods. The process parameters include machining parameters, time parameters, and resource parameters; wherein, machining parameters include cutting speed, feed rate, temperature control, and pressure setting; time parameters include standard working time, buffer time, and setup time; and resource parameters include equipment capacity, tool specifications, and personnel skills. The real-time status data includes production status, resource status, and quality status; wherein, production status includes process progress, start time, end time, and actual time consumed; resource status includes equipment utilization rate, number of failures, maintenance records, and personnel attendance; and quality status includes inspection results, number of defects, and number of rework attempts.

[0010] Furthermore, the process decomposition based on the product structure tree includes: Traverse the nodes of the product structure tree, and use a depth-first search algorithm to visit the child nodes sequentially starting from the root node of the product structure tree, extract the process steps of each node, wherein the parent node process is processed before the child node process, and generate an initial process sequence. According to the formula Calculate the estimated time for each process. ;in, Let i be the base time unit for the i-th action. Let be the number of repetitions of the i-th action. For environmental adjustment coefficients, This is the process complexity correction value, where n is the total number of actions included in the process. Based on the estimated duration and logical relationship of each process, calculate the earliest start time ES and the latest start time LS for each process; where ES is calculated by adding the duration of the preceding processes from the project start time, and LS is calculated in reverse from the project deadline. The sequence of operations with the earliest start time ES and the latest start time LS are marked as the critical path, and the operations on the critical path are marked as high priority; The time window for processes on non-critical paths is calculated using the following formula: ;in, This is the amount of elastic buffer. Output a sequence of processes, which includes process order, duration, time window, and critical path identifier.

[0011] Furthermore, the multi-dimensional conflict detection of the process sequence includes: The process iterates through each process in the process sequence, loads the 3D model associated with each process, and calculates the volume overlap between the 3D models through geometric interferometry analysis; wherein, the geometric interferometry analysis includes detecting the intersection of the model bounding boxes and calculating the actual overlapping volume; When the volume overlap is greater than a preset safety distance threshold, it is determined to be a spatial conflict; Based on the time arrangement of the process sequence, the resource requirements of all concurrent processes are summarized at each time point; When the total demand for resources exceeds the maximum capacity of resources, it is considered a resource conflict. Construct a directed graph model of process logic relationships, wherein nodes in the directed graph model represent processes and edges represent dependencies between processes; In a directed graph model, if the time interval between each pair of adjacent processes is greater than the preset minimum interval, it is determined to be a timing conflict.

[0012] Furthermore, the conflict resolution based on the detection results includes: To address spatial conflicts, the principle of minimum displacement is applied, moving the process model involved in the interference along the shortest path until it is no longer in contact, and adjusting the spatial position attributes of the process. To address resource conflicts, a priority preemption rule is applied to automatically delay the start time of low-priority processes and allocate the resources of low-priority processes to high-priority processes. To address timing conflicts, a buffer pool borrowing rule is applied to borrow time from the buffer time of the conflicting process or adjacent processes until the minimum interval requirement is met.

[0013] Furthermore, the step of generating the optimal production schedule using a multi-objective optimization algorithm includes: Define a multi-objective optimization function to minimize the total project duration, resource costs, and risks. The objective function is expressed as: Where α, β, and γ are the weighting coefficients for each item. Indicates the total construction period. Indicates resource cost, Indicates risk indicators, Let be the probability of risk event s occurring. Let w be the influencing factor of risk event s, and w represent the number of risk events; The multi-objective optimization function is solved by a hybrid strategy combining mixed integer programming and genetic algorithm, and the constraints include resource capacity constraints, process timing constraints and space constraints. The optimal production schedule is output based on the solution results, including the process timetable, resource allocation table, risk assessment, and real-time status data in the product structure tree is updated.

[0014] Furthermore, the method of using a hybrid strategy combining mixed integer programming and genetic algorithms to solve the multi-objective optimization function includes: Initialize the genetic algorithm population, set the population size to N, and define each chromosome as an encoding of the permutation of the process sequence, the chromosome length as the total number of processes, and the gene position as the process number; Define the fitness function as F = 1 / Z, where Z is the value of the multi-objective optimization function; Mixed-integer programming is used to handle the linear constraints of multi-objective optimization functions, including resource capacity constraints, process timing constraints, and space constraints. The chromosomes are iteratively optimized using genetic algorithms involving selection, crossover, and mutation, and after each generation of purification, the offspring chromosomes are repaired using mixed integer programming to ensure that all linear constraints are met. If the change in fitness value after K consecutive iterations is less than the threshold or the maximum number of iterations is reached, then the iteration stops and the final population is obtained. Select the chromosome with the highest fitness value from the final population, decode it into process sequence and time arrangement, and generate the optimal production schedule.

[0015] Furthermore, the multi-objective optimization algorithm also includes resource coupling constraint handling: Define the resource coupling matrix ,in, The resource dependency intensity of processes i and j is represented by the formula: ;in, Let i be the set of resource requirements for process i. Let j be the set of resource requirements for process j. Let i be the spatial distance between process i and process j. The distance attenuation constant; The resource coupling constraint is defined as follows: when the resource dependency strength of process i and process j is greater than the preset strength threshold θ, process i and process j are forced to preferentially allocate the same resources or adjacent time windows.

[0016] Furthermore, the adjustment of the optimal production schedule based on the deviation between planned and actual multi-indicator data includes: Real-time collection of production data to obtain equipment status, process progress, resource usage, and quality inspection results; Calculate the multi-indicator deviation index based on production data. The calculation formula is: ;in, For the planned time vector, For the actual time vector, Assign vectors to planned resources. Assign vectors to actual resources. This is the quality deviation value. , , These are the weight coefficients for each item, and their sum is 1; Multi-level adjustment strategies are triggered based on multi-indicator deviation indices: When the multi-indicator deviation index is less than the first threshold, the fine-tuning strategy is activated, and the time window or resource allocation of non-critical processes is automatically adjusted through the IF-THEN rule. When the multi-indicator deviation index is greater than or equal to the first threshold and less than the second threshold, a local rearrangement strategy is initiated, and a heuristic algorithm is used to reallocate resources or adjust the timing of the affected processes. When the multi-indicator deviation index is greater than or equal to the second threshold, the global reconstruction strategy is initiated, and the multi-objective optimization algorithm is called again to generate a new production schedule. The adjustment results are pushed to the production site in real time, and the real-time status data in the product structure tree is updated.

[0017] Compared with the prior art, the beneficial effects of this application are: This application integrates design attributes, process parameters, and real-time status data at each node to form a complete digital twin of the product, effectively solving the problem of data fragmentation in traditional manufacturing processes. The product structure tree can automatically evolve according to the production stage, from the functional system organization in the design stage to the resource unit mapping in the manufacturing stage, ensuring data consistency and continuity and providing a reliable data foundation for subsequent process optimization.

[0018] In terms of process optimization, this application achieves precise control of the production process through a systematic process decomposition, conflict detection, and resolution mechanism. The process decomposition method based on depth-first search, combined with critical path identification technology, can quickly generate optimized process sequences. Multi-dimensional conflict detection covers spatial, resource, and temporal conflicts, and combined with intelligent resolution strategies, significantly improves the feasibility of production scheduling. In particular, the optimization method combining mixed integer programming and genetic algorithms achieves a multi-objective balance of schedule, cost, and risk while satisfying complex constraints, greatly improving scheduling efficiency and quality.

[0019] This application utilizes real-time data acquisition and multi-indicator deviation analysis to enable the system to dynamically adjust production scheduling plans, demonstrating strong adaptability. From conflict prediction and resolution to real-time dynamic adjustment, a complete control loop is formed, encompassing planning and execution monitoring. This intelligent scheduling method not only improves production efficiency and reduces operating costs but also enhances the manufacturing system's ability to cope with uncertainties, providing reliable technical support for intelligent manufacturing. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1A flowchart illustrating the intelligent scheduling process connection optimization method based on the product structure tree provided in this application embodiment; Figure 2 A flowchart illustrating another intelligent scheduling process connection optimization method based on a product structure tree provided in this application embodiment; Figure 3 A flowchart illustrating another intelligent scheduling process connection optimization method based on a product structure tree provided in this application embodiment; Figure 4 This is a flowchart illustrating another intelligent scheduling process connection optimization method based on a product structure tree, provided as an embodiment of this application. Detailed Implementation

[0022] In the description of this application, unless otherwise stated, " / " means "or," for example, A / B can mean A or B. The "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, and B alone. Furthermore, "at least one" means one or more, and "multiple" means two or more. The terms "first," "second," etc., do not limit the quantity or order of execution, and "first," "second," etc., do not necessarily imply differences.

[0023] It should be noted that, in this application, the terms "exemplary" or "for example" are used to indicate that something is being described as an example, illustration, or illustration. Any embodiment or design described as "exemplary" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or design solutions. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0024] To address the technical problems of fragmented production scheduling data, delayed conflict identification, and poor plan adaptability in existing technologies, embodiments of this application provide an intelligent production scheduling process connection optimization method based on a product structure tree. This method includes: Construct a dynamic evolution product structure tree; wherein, the product structure tree is a multi-level tree data structure, and the root node of the structure tree corresponds to the product as a whole, and the child nodes are organized hierarchically according to functional systems, manufacturing areas and resource units, and each node stores design attributes, process parameters and real-time status data; Based on the product structure tree, process decomposition is performed to generate process sequences and identify critical paths. Multi-dimensional conflict detection is performed on the process sequence, and conflict resolution is carried out based on the detection results; among which, multi-dimensional conflicts include spatial conflicts, resource conflicts, and temporal conflicts; Based on the process sequence after conflict resolution, an optimal production schedule is generated through a multi-objective optimization algorithm. Data is collected in real time during the production process, and the optimal production schedule is adjusted based on the deviation of multiple indicators between the plan and the actual situation.

[0025] Based on this, the method enables the integration of design, process and manufacturing data, resolves production conflicts in advance, dynamically adapts to changes on site, and balances multiple objectives to improve scheduling feasibility and production efficiency.

[0026] like Figure 1 As shown in the embodiments of this application, the intelligent scheduling process connection optimization method based on the product structure tree includes: S1. Construct a dynamic evolution product structure tree.

[0027] The product structure tree is a multi-level tree data structure. The root node corresponds to the product as a whole, and the child nodes are organized hierarchically according to functional systems, manufacturing areas and resource units. Each node stores design attributes, process parameters and real-time status data.

[0028] In some implementations, when constructing a product structure tree, the core functional modules of the product can be identified first, thus defining the first-level sub-nodes; then, manufacturing areas can be determined according to the physical division of the production site, serving as second-level sub-nodes; finally, specific processing equipment, tool sets, or operating teams can be used as resource units, forming third-level and lower-level sub-nodes. The integration of node data can be achieved by linking product design documents, process specifications, and production ledgers, enabling unified storage of design attributes, process parameters, and real-time status data.

[0029] It should be noted that the core of dynamic evolution is that the structure tree can be iteratively adjusted with the production stage. The hierarchical division of nodes and data content can be dynamically updated according to events such as design finalization, process changes, and resource scheduling, to ensure that the data is consistent with the actual production.

[0030] For example, the root node is a complete machine of a certain type of mechanical equipment, the first-level sub-nodes are divided into functional modules such as power system, transmission system, and control system; the second-level sub-nodes are divided according to manufacturing areas such as stamping workshop, welding workshop, and assembly workshop; the third-level sub-nodes correspond to resource units such as stamping machine tools, welding robots, and assembly teams in the workshop, and each node stores data such as the size specifications, processing speed, and equipment operating status of the corresponding module.

[0031] S2. Based on the product structure tree, decompose the process, generate the process sequence, and identify the critical path.

[0032] The critical path is the identification of the core processes that have a decisive impact on the overall project duration, providing a basis for prioritizing resource allocation and controlling key aspects of the schedule, and preventing overall production delays due to delays in critical processes.

[0033] In some implementations, generating the process sequence typically involves using a breadth-first search algorithm to traverse the product structure tree nodes, extracting the processing steps, assembly processes, and other process details corresponding to each node in a hierarchical order of "root node - parent node - child node," and then sorting them according to production logic to form the initial process sequence. Identifying the critical path can be achieved through a combination of forward and backward methods. The forward method starts from the project start time and accumulates the estimated execution time of each process to obtain the earliest start and earliest finish time for each process. The backward method works by subtracting the estimated execution time of each process from the project deadline to obtain the latest start and latest finish time for each process. Connecting the processes with the same earliest and latest start times together forms the critical path.

[0034] S3. Perform multi-dimensional conflict detection on the process sequence and resolve conflicts based on the detection results.

[0035] Among the multi-dimensional conflicts, spatial conflicts refer to the overlapping and interference of physical spaces such as operating space, material stacking area, and equipment movement range during the execution of a process, which prevents the process from being carried out normally; resource conflicts refer to the total demand of multiple concurrent processes for resources such as equipment, manpower, materials, and tools exceeding the current available capacity within the same time dimension, resulting in resource competition; and temporal conflicts refer to the execution order of processes not conforming to the process logic, or the time interval between adjacent processes being less than the safety threshold, resulting in the start of the subsequent process before the previous process is completed, or the excessive gap between processes causing inefficiency.

[0036] In some implementations, multi-dimensional conflict detection methods may include: spatial conflicts can be analyzed by drawing a process operation space layout diagram or constructing a space occupancy matrix to determine whether there is an overlap in the space requirements of different processes; resource conflicts can be determined by summarizing the resource requirement list of each process and comparing it with the real-time updated resource availability ledger to determine whether there is an imbalance between resource supply and demand; and temporal conflicts can be determined by drawing a Gantt chart to visually present the time arrangement of processes or constructing a process dependency matrix to check whether the logical relationships and time intervals between processes are compliant.

[0037] In some implementation methods, conflict resolution methods may include: spatial conflicts can be resolved by adjusting the spatial location of the process execution, staggering the operation time windows of concurrent processes, or optimizing the material stacking layout; resource conflicts can be resolved by using the resource balancing allocation method, delaying the resource occupation time of non-critical processes, or calling on spare resources or alternative resources to meet the demand; timing conflicts can be resolved by replanning the start and end times of processes, compressing the buffer time of non-critical processes, or adjusting the execution order of processes to ensure compliance with process dependencies.

[0038] It should be noted that conflict detection should be carried out throughout the entire process of process execution, rather than just during the production scheduling stage, in order to avoid new conflicts caused by real-time status changes; conflict resolution should prioritize ensuring the normal progress of critical path processes.

[0039] For example, during spatial conflict detection, by simulating the machine tool movement trajectory of the stamping process and the operating area of ​​the adjacent welding process, it was found that the spatial overlap rate between the two reached 30%, which was determined to be a spatial conflict. This was resolved by staggering the operation times, adjusting the stamping process to be executed in the morning and the welding process to be executed in the afternoon. During resource conflict detection, it was found that three processes simultaneously requested to use a certain precision machine tool during the same time period, but there was only one machine tool available, which was determined to be a resource conflict. Following the principle of prioritizing critical path processes, the two non-critical path processes were delayed until the machine tool's idle time.

[0040] S4. Based on the process sequence after conflict resolution, the optimal production schedule is generated through a multi-objective optimization algorithm. Data is collected in real time during the production process, and the optimal production schedule is adjusted based on the deviation of multiple indicators between the plan and the actual situation.

[0041] In some implementations, common forms of multi-objective optimization algorithms include the Non-dominated Sorting Genetic Algorithm (NSGA), Particle Swarm Optimization (PSO), and Simulated Annealing (SA). These algorithms aim to minimize the total project duration, resource costs, and quality risks, and find the optimal solution that balances multiple objectives through iterative search.

[0042] In some implementations, adjusting the optimal production schedule may include: setting multi-level deviation thresholds, collecting data such as equipment operating status, process completion progress, resource consumption, and quality inspection results in real time, and calculating the degree of deviation between the plan and the actual situation; when the deviation is less than the first threshold, only the time window or resource allocation ratio of non-critical processes is fine-tuned; when the deviation is between the first threshold and the second threshold, the affected local processes are rescheduled; when the deviation is greater than the second threshold, a new production schedule is generated by calling the multi-objective optimization algorithm again.

[0043] It should be noted that the weights of each objective in multi-objective optimization can be dynamically configured according to the company's production strategy. For example, when orders are urgent, the weight of the chief engineer option can be increased, and when cost control is needed, the weight of resource cost can be increased. Real-time data collection must ensure timeliness and accuracy to avoid improper adjustments due to data lag.

[0044] Based on the above technical solutions, the intelligent scheduling process connection optimization method based on product structure tree provided in this application realizes the full-process connection of design, process and manufacturing data by constructing a dynamic evolution product structure tree, breaking down information silos; process decomposition and critical path identification based on the structure tree ensure the rationality and priority distinction of the process sequence; multi-dimensional conflict detection and resolution mechanism resolves potential production problems in advance, reducing rework and interruptions; multi-objective optimization algorithm achieves a comprehensive balance of objectives such as schedule, cost and quality, and combined with the dynamic adjustment mechanism of real-time data, makes the scheduling plan highly adaptable, ultimately improving scheduling efficiency, planning feasibility and the robustness of the production system.

[0045] In one possible implementation of the embodiments of this application, combined with Figure 1 ,like Figure 2 As shown, the above-mentioned S1 constructing of the dynamic evolution product structure tree can be specifically implemented through the following S101, S102 and S103, which are explained in detail below: S101, Defines a unique identifier for a product structure tree node.

[0046] The unique identifier of each node is used to accurately distinguish and associate data between nodes, ensuring that node data in different production stages and different systems is traceable and matchable.

[0047] In some implementations, the unique identifier for a node uses the PSN format, fully expressed as PSN-{Product Code}-{Region Code}-{Process Type}. Here, the Product Code is the unique code for the product, used to distinguish different models or batches; the Region Code is a unique identifier for the manufacturing region, used to locate the physical location where the process is performed; and the Process Type is the process category to which the node belongs, used to specify the processing or assembly type corresponding to the node.

[0048] It should be noted that the coding rules of PSN need to be uniformly formulated within the enterprise to ensure that the identifier of each node is unique. At the same time, the coding structure needs to be readable and extensible to facilitate flexible adaptation when adding new products, regions or process types in the future.

[0049] For example, if the product code of a certain engine model is "ENG-2024-001", the manufacturing area code for its cylinder block machining process is "ASM-03", and the code corresponding to the process type "machining" is "MACH-02", then the unique identifier of this node is "PSN-ENG-2024-001-ASM-03-MACH-02".

[0050] S102. Define the attribute set of the product structure tree node.

[0051] Among them, the node attribute set is the carrier for storing product design, process and production related data, providing data support for process decomposition, conflict detection and production scheduling optimization.

[0052] In some implementations, each node contains three types of data: a static attribute set, a dynamic attribute set, and an evolution rule set; among them, The static property set includes design specifications, geometric constraints, and quality requirements: design specifications cover geometric dimensions, material types, tolerance ranges, and surface treatment requirements; geometric constraints include assembly relationships, spatial boundaries, and motion trajectories; and quality requirements include inspection standards, allowable defect rates, and testing methods.

[0053] The dynamic attribute set includes the current status, estimated duration, and resource requirements: the current status reflects the real-time operating condition of the process or resource corresponding to the node; the estimated duration is the expected time for the process to be executed; and the resource requirements specify the types and quantities of resources such as equipment, tools, and personnel required for the process.

[0054] The evolution rule set includes preconditions, triggering timing, and evolution logic: Preconditions define the dependencies of node evolution, and evolution can only be initiated if specific conditions are met; Triggering timing is based on event-driven settings and is associated with events such as process completion, resource changes, or order adjustments; Evolution logic is used to specify the specific rules for node classification, merging, or attribute updates.

[0055] It should be noted that the static attribute set is the basic data, the dynamic attribute set is the real-time dynamic data, and the evolution rule set is the core rule for the structure tree to adapt to production changes. The three together ensure the integrity and usability of the node data.

[0056] For example, in the static attribute set of a gear machining node, the design specifications are "20 teeth, module 2.5mm, material 45 steel, tolerance ±0.02mm", the geometric constraints are "coaxial assembly with the drive shaft, radial runout not exceeding 0.01mm", and the quality requirements are "magnetic particle testing standard GB / T 15822.1-2005, allowable defect rate ≤0.3%"; in the dynamic attribute set, the current state is "awaiting processing", the estimated time is "2 hours", and the resource requirements are "CNC lathe CK6140, cutting tool TCMT16T304, 1 intermediate lathe operator"; in the evolution rule set, the precondition is "gear blank forging completed and inspected and qualified", the triggering time is "blank arrives at the warehouse", and the evolution logic is "after triggering, the node status is updated to 'processing', and if processing is completed, it splits into two sub-nodes: 'gear finishing' and 'gear inspection'".

[0057] S103. Automatically evolve the product structure tree based on the product production stage.

[0058] Among them, automatic evolution is the core manifestation of the dynamic characteristics of the product structure tree, ensuring that the hierarchical organization and node configuration of the structure tree are synchronously adapted to the production process.

[0059] In some implementations, automated evolution proceeds in stages according to production phases, mainly including the design phase, process planning phase, and production execution phase, among which: During the design phase, nodes are organized according to functional systems, and the product is broken down into core functional modules such as power system and transmission system, with each module serving as a first-level sub-node. During the process planning phase, nodes are reorganized according to manufacturing areas, and based on functional modules, second-level sub-nodes are divided according to production sites such as stamping workshop and welding workshop. During the production execution phase, nodes are mapped to resource units, and specific processing equipment, tool groups or work teams are designated as third-level and lower-level sub-nodes.

[0060] In some implementations, node evolution follows a unified rule: Trigger(E) → Action(A), where Trigger represents the triggering action, E is a set of events, including key events in the production scenario such as process completion, resource changes, or order adjustments, and A is a set of actions, including structural adjustment methods such as node splitting, node merging, or attribute updates.

[0061] It should be noted that the core of automatic evolution is event-driven. The adjustment of the tree structure does not require manual intervention and is automatically triggered by preset events. At the same time, historical data of nodes must be retained during the evolution process to facilitate the tracking of the structure adjustment trajectory.

[0062] For example, the production evolution process of a certain mechanical equipment is shown below: During the design phase, the first-level sub-nodes under the root node "Complete Mechanical Equipment" are "Power System," "Transmission System," and "Control System." During the process planning phase, the "Power System" sub-node is reorganized into "Stamping Workshop - Power System Components," "Welding Workshop - Power System Components," and "Assembly Workshop - Power System Components." During the production execution phase, "Stamping Workshop - Power System Components" is further mapped to resource unit nodes such as "Stamping Machine A - Cylinder Block Stamping" and "Stamping Machine B - End Cap Stamping." When the "Cylinder Block Stamping" process is completed (Event E), a node splitting action (Action A) is triggered, generating two new nodes: "Cylinder Block Rough Machining" and "Cylinder Block Flaw Detection," while simultaneously updating the original node's attribute to "Completed."

[0063] Based on the above technical solutions, the dynamically evolving product structure tree constructed by S1 solves the technical problems of fragmented data, chaotic node identification, and inability of the structure to adapt to changes in production stages in traditional product structures. Its beneficial effects are reflected in three aspects: First, the unique identifier PSN enables global and precise node positioning, avoiding data confusion and laying the foundation for cross-system data integration; second, the complete attribute set definition integrates data from the entire design, process, and production process, breaking down information silos and providing comprehensive and reliable data support for subsequent process optimization; third, through automatic evolution driven by production stages, it ensures that the product structure tree is updated synchronously with actual production, guaranteeing data consistency and continuity, and improving the adaptability and feasibility of production scheduling.

[0064] In one possible implementation of the embodiments of this application, combined with Figure 1 ,like Figure 3 As shown, S2 above decomposes the process based on the product structure tree, generates the process sequence, and identifies the critical path. Specifically, it can be implemented through the following S201, S202, and S203, which are explained in detail below: S201. Traverse the product structure tree nodes to generate the initial process sequence.

[0065] In some implementations, a depth-first search (DFS) algorithm can be used to traverse the product structure tree. Starting from the root node of the structure tree, the algorithm first visits all child nodes of each parent node, extracts the process steps corresponding to each node one by one, and strictly follows the rule of "parent node processes take precedence over child node processes" to sort them, ultimately forming an initial process sequence.

[0066] It should be noted that the depth-first search algorithm can ensure that the complete process chain of each functional module or component is extracted continuously, avoiding the disorderly interleaving of processes across modules. At the same time, the parent node priority principle is in line with the production and manufacturing logic of the product's "whole-part" and "assembly-component" structure, ensuring the basic rationality of the process sequence.

[0067] For example, taking a new energy vehicle battery pack as the root node, its first-level parent nodes are "cell module", "battery management system" and "casing structure". When traversing the parent node of "cell module", first extract its "cell screening", "cell series and parallel connection" and "module packaging" processes, and then extract the processes of the child nodes "cell testing" and "tab welding". According to the parent node priority rule, the initial process sequence is "cell screening - cell testing - tab welding - cell series and parallel connection - module packaging - battery management system assembly - casing structure processing - battery pack assembly".

[0068] S202. Calculate the core parameters of the process, including the estimated duration, earliest / latest time and time window.

[0069] In some implementations, the calculation of the core parameters includes the following steps: First, use the formula Calculate the estimated time for each process step, taking into account process actions, environmental impacts, and technological complexity, to accurately predict the execution time of a single process step and avoid production scheduling imbalances caused by estimation errors. Estimate the time required for each process; This serves as the basic time unit for the i-th action, such as the standard time for tightening a single screw. This represents the number of repetitions of the i-th action, such as the number of times a screw is tightened. This is an environmental adjustment factor, such as a correction factor for efficiency decline under high temperature and high humidity conditions, and its value range is usually 1.0~1.5; This is a correction value for process complexity, such as the extra time required for high-precision machining and complex assembly, which is set to a positive number based on the difficulty of the process; n is the total number of actions included in the process.

[0070] Then, calculate the earliest start time (ES) and the latest start time (LS): ES is obtained by adding the estimated duration of all preceding processes from the project start time; LS is obtained by subtracting the estimated duration of subsequent processes from the project end time.

[0071] Next, according to the formula Calculating the time window for non-critical path processes provides a flexible execution range for these processes, clearly defining the latest execution boundary while reserving adjustment space for optimized resource allocation. The time window for non-critical processes; ES is the earliest start time; LS is the latest start time; δ is the flexible buffer amount, which is set according to production experience and is usually 10% to 20% of the difference between ES and LS.

[0072] For example, a certain "cell series-parallel connection" process includes three actions: cell positioning (B1=20s, M1=10), wire welding (B2=30s, M2=20), and insulation wrapping (B3=15s, M3=10), with an environmental adjustment coefficient K. env =1.1, Process complexity correction value ΔT adj =120s. Substitute into the formula to calculate: , If the project starts at 8:00 AM on the first day, and the estimated time for the preceding process "cell screening" is 30 minutes, then the ES for this process is 8:30. If the project ends at 12:00 PM on the first day, and the estimated time for the subsequent process "module packaging" is 25 minutes, then LS = 12:00 - 25 min - 19.4 min ≈ 11:15:36. For non-critical paths, δ = 10 minutes, and the time window Window = [8:30, 11:15:36] ∪ [9:52:48 - 10 min, 9:52:48 + 10 min] = [8:30, 11:15:36] ∪ [9:42:48, 10:02:48].

[0073] S203. Identify the critical path and output the complete process sequence.

[0074] Identifying the critical path is a core step in focusing on key production priorities. A complete sequence of processes can provide a clear basis for subsequent conflict detection and scheduling optimization.

[0075] In some implementations, the critical path is determined by: selecting processes with equal ES and LS values, connecting them according to production logic to form the critical path, and marking all processes on the critical path as high priority to ensure resource supply and schedule control. The output of the complete process sequence should include the process order, the estimated duration of each process, the time window for non-critical processes, and the critical path identifier to ensure that subsequent processes can directly access the core information.

[0076] It should be noted that the critical path is the core chain that affects the overall project duration. If any critical process is delayed, it will inevitably lead to an overall production delay. The output process sequence should be presented in a structured form (such as tables or JSON format) to facilitate computer system parsing and subsequent module calls. The critical path is not static. If the actual time of the process in subsequent production deviates significantly from the estimate, the parameters need to be recalculated and the critical path updated.

[0077] For example, combining the initial process sequence of S201 and the parameter calculation results of S202, processes with ES and LS equal, such as "Cell Screening (ES=8:00, LS=8:00)", "Cell Series-Parallel Connection (ES=8:30, LS=8:30)", "Module Packaging (ES=8:50, LS=8:50)", and "Battery Pack Assembly (ES=10:30, LS=10:30)", are selected, connected in series to form a critical path and marked as high priority; the output complete process sequence contains structured information such as "Process 1: Cell Screening, duration 30min, no time window, critical path; Process 2: Cell Inspection, duration 25min, time window [8:30, 10:45] ∪ [9:37:30, 9:57:30], non-critical path...".

[0078] Based on the above technical solutions, S2 quickly generates logically coherent and production-realistic process sequences through a depth-first search algorithm and a parent node priority principle, avoiding production chaos caused by disordered processes. Combined with quantitative formulas, it accurately calculates core parameters such as process estimated duration and time windows, providing data support for production scheduling and reducing subjective errors from manual estimation. Finally, it identifies critical paths and marks their priorities, clarifying key points of production control and providing a basis for priority resource allocation and progress risk warning. At the same time, the time window design for non-critical processes reserves flexibility for subsequent resource optimization and conflict resolution, thus improving the scientific nature and executability of the production scheduling plan as a whole.

[0079] In one possible implementation of this application embodiment, the above-mentioned S3 performs multi-dimensional conflict detection on the process sequence and resolves conflicts based on the detection results. Specifically, this can be achieved through the following S301, S302, and S303, which are described in detail below: S301, Spatial Conflict Detection and Resolution.

[0080] Spatial conflict refers to the overlapping and interference of physical spaces such as operating areas, equipment movement ranges, and material stacking spaces during the execution of a process, which can prevent the process from being carried out normally or cause safety hazards.

[0081] In some implementations, the specific operations for spatial conflict detection may include: The process iterates through each process in the process sequence, loads the 3D model associated with each process, and completes the conflict determination through geometric interference analysis. The geometric interference analysis includes two core operations: the first step is to detect the intersection of the model bounding boxes and quickly filter out process combinations that may have spatial overlap; the second step is to calculate the actual overlap volume of the filtered process models to obtain the volume overlap degree. When the volume overlap exceeds the preset safety distance threshold, it is determined to be a spatial conflict.

[0082] In some implementations, spatial conflict resolution methods may include: adopting the principle of minimum displacement, i.e., calculating the shortest separation path between the interfering process models, moving one or more models along this path until there is no overlap between the models and the safety distance requirement is met, while simultaneously adjusting the spatial position attributes of the corresponding processes. For processes that cannot be moved, the operation time windows of concurrent processes can be staggered to ensure that each process has sufficient space within that time.

[0083] S302, Resource Conflict Detection and Resolution.

[0084] Resource conflict refers to a situation where, within the same time frame, the total demand for production resources such as equipment, manpower, materials, and tools from multiple concurrent processes exceeds the maximum available capacity of those resources, leading to resource contention, process waiting, or inability to execute.

[0085] In some implementations, the specific operations for resource conflict detection may include: Based on the time arrangement of the process sequence, the time axis is split according to the time granularity, and the resource requirements of all concurrent processes are summarized at each time point, including resource type, quantity, and usage duration; Then compare the total demand for each type of resource with the maximum capacity of that resource (such as the total number of equipment units, the total number of personnel, and the upper limit of material inventory); When the total demand exceeds the maximum capacity, it is considered a resource conflict.

[0086] In some implementations, the resource conflict resolution method is as follows: adopt priority preemption rules, first clarify the priority of all processes; among them, the priority of critical path processes is higher than that of non-critical path processes, and the priority of urgent order processes is higher than that of ordinary order processes; Automatically delay the start time of low-priority processes, allocate the released resources to high-priority processes first, and update the time windows of low-priority processes.

[0087] S303, Timing Conflict Detection and Resolution.

[0088] Among them, timing conflict refers to the execution order of processes violating the process logic dependency relationship, or the time interval between adjacent processes being less than the preset minimum interval, resulting in the start of the subsequent process before the previous process is completed, or the chaotic process connection affecting production efficiency.

[0089] In some implementations, the specific operations for timing conflict detection include: First, construct a directed graph model of the logical relationship between processes. In this model, the nodes of the directed graph represent processes, and the directed edges between nodes represent the dependencies between processes, such as “part processing” → “part inspection” → “part assembly”. Based on this model, all adjacent process pairs are traversed, and the difference between the earliest completion time of the preceding process and the earliest start time of the following process is calculated, i.e., the time interval. When the time interval is greater than the preset minimum interval, it is determined to be a timing conflict.

[0090] In some implementations, the method for resolving timing conflicts is as follows: a buffer pool borrowing rule is adopted. The buffer pool consists of the elastic buffer time of each process, including the redundant time within the time window of non-critical processes. The time is borrowed from the buffer time of the process that has a timing conflict or its adjacent processes to supplement the time interval between the two processes until the time interval meets the preset minimum interval requirement.

[0091] For example, after constructing the directed graph model, it was found that the time interval between the preceding process "part heat treatment" and the following process "part finishing" is 3 minutes, while the preset minimum interval is 8 minutes, which is determined to be a timing conflict. A query of the buffer pool revealed that the flexible buffer time for "part heat treatment" is 10 minutes. 5 minutes were borrowed from this buffer to supplement the time interval. After adjustment, the end time of "part heat treatment" is delayed by 5 minutes, and the time interval between the two processes becomes 8 minutes, satisfying the minimum interval requirement.

[0092] Based on the above technical solutions, S3 achieves early identification, quantitative positioning, and efficient resolution of spatial overlap interference, resource contention conflicts, and temporal logic disorder before production by using 3D model geometric interference analysis for spatial conflict detection, time-point resource demand aggregation and comparison for resource conflict detection, and process logic directed graph model construction for temporal conflict detection. It also employs dedicated resolution strategies for the three types of conflicts, including minimum displacement, priority preemption, and buffer pool borrowing. This minimizes work stoppages, material shortages, rework, and safety hazards caused by conflicts, ensuring the continuity of process connections and the stability of production rhythm. It significantly improves the feasibility of production scheduling and solves the problems of traditional production scheduling relying on manual experience to judge conflicts, delayed conflict identification, blind and disordered resolution strategies, and low production efficiency, increased cost losses, and schedule delays caused by sudden spatial, resource, and temporal conflicts during production.

[0093] In one possible implementation of the embodiments of this application, combined with Figure 1 ,like Figure 4 As shown, the above S4 generates the optimal production schedule based on the process sequence after conflict resolution and adjusts it dynamically, specifically including the following S401 to S403: S401. Define the multi-objective optimization function and constraints.

[0094] In some implementations, multi-objective optimization focuses on minimizing the total project duration, resource costs, and risks. The complete form of the objective function is: ,in, To comprehensively optimize the target value, the smaller the value, the better the production scheduling plan; These are the weight coefficients for each item, with values ​​ranging from 0 to 1, and satisfying α+β+γ=1. The weights can be adjusted according to the company's production strategy. The total construction period is the total time from the start of the first process to the completion of the last process. Resource costs include the total input of all production resources, such as equipment depreciation, labor costs, material consumption, and tool wear and tear. This is a risk indicator used to quantify the impact of various risks in the production process. The calculation formula is as follows: ,in Let be the probability of the s-th risk event occurring, such as the probability of equipment failure, the probability of delayed material delivery, etc. Let w be the influencing factor of the s-th risk event, such as the duration of project delays caused by equipment failure, the proportion of cost losses, etc., and w be the total number of risk events.

[0095] In some implementations, the constraints of the multi-objective optimization function include: 1. Resource capacity constraints: The actual usage of various resources must not exceed the maximum available capacity, such as the total number of equipment, the total number of personnel, and the upper limit of material inventory; 2. Process timing constraints: The execution order of processes must follow the logical dependencies determined by S2, and the time interval between adjacent processes must meet the requirements after resolution by S3; 3. Spatial Constraints: The spatial location of the process must conform to the adjustment results after conflict resolution in S3, with no spatial overlap or interference; 4. Resource Coupling Constraints: Define the resource coupling degree matrix. ,in The resource dependency intensity of processes i and j is represented by the formula: ;in, Let i be the set of resource requirements for process i. Let j be the set of resource requirements for process j. Let i be the spatial distance between process i and process j. This is the distance attenuation constant, set according to the layout of the production area, typically taking a value of 5–10m. When When the intensity exceeds the preset threshold θ (usually 0.6~0.8), process i and process j are forced to prioritize the allocation of the same resources or adjacent time windows to reduce resource switching costs.

[0096] It should be noted that resource coupling constraints limit the intensity of resource dependency ( The resource allocation logic for processes with a distance greater than the preset threshold θ and a relatively short spatial distance clarifies the priority rules for such processes in terms of resource use and time scheduling. This is to minimize the time loss and cost of resource switching processes such as equipment changeover, tool adjustment, and material transfer by prioritizing the integration of resource needs of highly related processes. At the same time, it improves the utilization rate of core production resources, shortens the gap between processes, and avoids problems such as production rhythm fluctuations and process accuracy deviations caused by frequent resource switching. This further enhances the rationality and feasibility of production scheduling, so that resource allocation not only conforms to the process dependence characteristics but also adapts to the workshop space layout, achieving dual optimization of resource utilization efficiency and production stability.

[0097] S402. Use a hybrid strategy to solve the optimization function and generate the optimal production schedule.

[0098] The core of the hybrid strategy is to combine the global search capability of genetic algorithms with the constraint handling precision of mixed integer programming to efficiently find the optimal production scheduling scheme that satisfies all constraints.

[0099] In some implementations, the process of solving the optimization function includes the following steps: 1. Initialize the Genetic Algorithm (GA) population: Set the population size to N, which is determined by the total number of processes. Typically, N = 50~200. Each chromosome encodes the sequence of processes. The length of the chromosome is equal to the total number of processes. Each gene position corresponds to a process number. For example, the chromosome "3-1-4-2" represents the execution order of process 3 → process 1 → process 4 → process 2.

[0100] 2. Define the fitness function: The fitness function is used to evaluate the quality of the spawning scheme corresponding to the chromosome. The formula is F=1 / Z, where Z is the multi-objective optimization function value. The larger the F value, the better the overall optimization effect of the spawning scheme.

[0101] 3. Constraint Handling: Mixed Integer Programming (MIP) is used to handle the linear constraints of the multi-objective optimization function, including resource capacity constraints, process timing constraints, and space constraints. By establishing a linear programming model, the process timing and resource allocation schemes corresponding to the chromosomes are verified, and schemes that violate the constraints are eliminated.

[0102] 4. Genetic Algorithm Iterative Optimization: Chromosomes are iteratively optimized following the selection, crossover, and mutation operation process, specifically including: Selection operation: The roulette wheel selection method is used to select high-quality chromosomes to enter the next generation according to the proportion of fitness function values; Crossover operation: Using the single-point crossover method, a crossover point is randomly selected, and gene segments of two parent chromosomes are exchanged to generate offspring chromosomes; Mutation operation: Randomly select gene loci and change process numbers to avoid the population getting trapped in local optima.

[0103] 5. Offspring chromosome repair: After each generation iteration, mixed integer programming is used to repair the offspring chromosomes, adjusting the process time or resource allocation that violates the constraints, to ensure that all offspring chromosomes satisfy the linear constraints.

[0104] 6. Iteration Termination Criterion: When the change in the fitness function value is less than a preset threshold after K consecutive iterations (usually K=10~20), for example, less than 10... -4 If the maximum number of iterations is reached, the iteration stops, and the final population is obtained.

[0105] 7. Optimal Solution Decoding: Select the chromosome with the highest fitness function value from the final population, decode it into a specific process schedule, resource allocation table, and risk assessment report, and update the real-time status data in the product structure tree to form the optimal production schedule. The process schedule includes the start and end times of each process, the resource allocation table includes the equipment, personnel, tools, etc. corresponding to each process, and the risk assessment report includes the main risk events and corresponding measures.

[0106] S403: Real-time data collection and dynamic adjustment of production scheduling based on deviations of multiple indicators.

[0107] The core of dynamic adjustment is to correct the deviation between the plan and the actual situation in a timely manner through real-time data feedback, so as to ensure that the production schedule always adapts to changes in the production site.

[0108] In some implementations, dynamically adjusting the production schedule may include the following steps: First, four types of data are collected in real time through devices such as Manufacturing Execution System (MES), Internet of Things (IoT) sensors, and quality inspection terminals: Equipment status: running / faulty, utilization rate, maintenance records; Process progress: actual start / end time and duration; Resource usage: equipment occupancy, personnel attendance, and material consumption; Quality inspection results: number of defects, pass rate, and rework status.

[0109] The data collection frequency is set according to the complexity of the process: data is collected every 5 minutes for precision machining processes and every 15 minutes for ordinary assembly processes.

[0110] Next, based on the collected actual data and production schedule data, a multi-indicator deviation index is calculated. The formula is: This provides a quantitative basis for triggering strategy adjustments by quantifying the deviations between the plan and reality in three dimensions: time, resources, and quality. , , This is the deviation weighting coefficient, with a value range of 0 to 1 and a sum of 1, set according to the degree of impact of the deviation on production; This is a planned time vector, including the planned start / end times for each process. This is the actual time vector, including the actual start / end time of each process. This refers to the time deviation rate; Assign a vector to the planned resources, including the quantity / type of planned resources for each process. Assign a vector to the actual resources, including the actual quantity / type of resources for each process. ΔQ represents the resource deviation rate, and the vector norm adopts the L1 norm; ΔQ is the quality deviation value, that is, the difference between the actual defect rate and the planned allowable defect rate. If the actual defect rate is lower than the planned value, ΔQ=0.

[0111] Multi-level adjustment strategies are triggered based on multi-indicator deviation indices: when <First threshold (usually 0.1): Initiate fine-tuning strategy, automatically adjust the time window or resource allocation of non-critical processes through IF-THEN rules, such as extending the start time of a non-critical process by 30 minutes to avoid resource occupation by critical processes; When the first threshold ≤ <Second threshold (usually 0.3): Initiate a local reordering strategy, using heuristic algorithms to reallocate resources or adjust timing of affected processes, such as migrating the process corresponding to a faulty device to a standby device, or adjusting the start and end times of related processes; when ≥Second threshold: Initiate a global refactoring strategy, re-invoke the multi-objective optimization algorithm to generate a new production schedule, covering all unexecuted processes.

[0112] Finally, the adjusted production schedule is pushed to the production site terminal in real time, and the real-time status data in the product structure tree is updated to ensure that the production site is consistent with the plan.

[0113] For example, in a certain machining production scheduling, the deviation weighting coefficient is set to... =0.5、 =0.3、 =0.2, first threshold =0.1, second threshold =0.3. In actual production, process A is planned to take 2 hours, and actually takes 2.2 hours, so the time deviation rate is |2-2.2| / 2 = 0.1; the planned resource is 1 CNC lathe, and actually 1 is used, so the resource deviation rate is 0; the actual defect rate is 0.2%, the planned allowable defect rate is 0.3%, ΔQ = 0, then D dev =0.5×0.1+0.3×0+0.2×0=0.05<0.1, so the fine-tuning strategy is activated, delaying the start time of subsequent non-critical process B by 0.2 hours.

[0114] Based on the above technical solutions, S4 integrates the three major objectives of schedule, cost, and risk through quantitative integration, combined with resource coupling constraints, to make the generated production schedule more closely match the actual production needs of enterprises. By combining genetic algorithms and mixed integer programming methods to find the optimal solution, the solution efficiency can be significantly improved. Based on real-time data, multi-index deviation calculation and multi-level adjustment strategies enable the production schedule to quickly respond to changes in the field such as equipment failure, schedule delays, and quality fluctuations, avoiding production chaos caused by the disconnect between the plan and reality, and improving the robustness of the production schedule and the stability of the production process.

[0115] Although this application has been described herein in conjunction with various embodiments, those skilled in the art, by reviewing the accompanying drawings, disclosure, and appended claims, will understand and implement other variations of the disclosed embodiments in carrying out the claimed application. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude multiple instances. A single processor or other unit can implement several functions listed in the claims. While different dependent claims may recite certain measures, this does not mean that these measures cannot be combined to produce good results.

[0116] Although this application has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made thereto without departing from the spirit and scope of this application. Accordingly, this specification and drawings are merely exemplary illustrations of this application as defined by the appended claims, and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from the spirit and scope of this application. Thus, if such modifications and modifications of this application fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and modifications.

Claims

1. A method for optimizing intelligent production scheduling processes based on a product structure tree, characterized in that, include: Construct a dynamic evolution product structure tree; The product structure tree is a multi-level tree data structure, with the root node of the structure tree corresponding to the product as a whole, and the child nodes organized hierarchically according to functional systems, manufacturing areas and resource units. Each node stores design attributes, process parameters and real-time status data. Based on the product structure tree, process decomposition is performed to generate process sequences and identify critical paths; The process sequence is subjected to multi-dimensional conflict detection, and conflict resolution is performed based on the detection results; the multi-dimensional conflicts include spatial conflicts, resource conflicts, and temporal conflicts. Based on the process sequence after conflict resolution, an optimal production schedule is generated through a multi-objective optimization algorithm. Data is collected in real time during the production process, and the optimal production schedule is adjusted based on the deviation of multiple indicators between the plan and the actual situation.

2. The intelligent scheduling process connection optimization method based on the product structure tree according to claim 1, characterized in that, The process of constructing the dynamic evolution product structure tree includes: Define each node in the product structure tree as a unique identifier (PSN), in the format PSN-{Product Code}-{Region Code}-{Process Type}; where the Product Code is the unique code of the product, the Region Code is the region identifier, and the Process Type is the process category to which the node belongs. Each node in the product structure tree is defined to contain a set of static attributes, a set of dynamic attributes, and a set of evolution rules; among them, The static attribute set includes design specifications, geometric constraints, and quality requirements; the dynamic attribute set includes current state, estimated duration, and resource requirements; and the evolution rule set includes preconditions, triggering timing, and evolution logic. Preconditions are used to define the evolutionary dependencies of nodes, triggering timing is based on event-driven settings, and evolution logic is used to specify node classification, merging, or attribute update rules. The product structure tree is automatically evolved based on the product's production stage to obtain a dynamically evolving product structure tree.

3. The intelligent scheduling process connection optimization method based on the product structure tree according to claim 2, characterized in that, The automatic evolution means that nodes are organized according to functional systems during the design phase, reorganized according to manufacturing regions during the process planning phase, and mapped to resource units during the production execution phase. The evolution rules for node evolution are as follows: ;in, E represents the triggering action, E represents the event set, including process completion, resource change or order adjustment, and A represents the action set, including node splitting, node merging or attribute updating.

4. The intelligent scheduling process connection optimization method based on product structure tree according to claim 1, characterized in that, The process decomposition based on the product structure tree includes: Traverse the nodes of the product structure tree, and use a depth-first search algorithm to visit the child nodes sequentially starting from the root node of the product structure tree, extract the process steps of each node, wherein the parent node process is processed before the child node process, and generate an initial process sequence. According to the formula Calculate the estimated time for each process. ;in, Let i be the base time unit for the i-th action. Let be the number of repetitions of the i-th action. For environmental adjustment coefficients, This is the process complexity correction value, where n is the total number of actions included in the process. Based on the estimated duration and logical relationship of each process, calculate the earliest start time ES and the latest start time LS for each process; The sequence of operations with the earliest start time ES and the latest start time LS are marked as the critical path, and the operations on the critical path are marked as high priority; The time window for processes on non-critical paths is calculated using the following formula: ;in, This is the amount of elastic buffer. Output a sequence of processes, which includes process order, duration, time window, and critical path identifier.

5. The intelligent scheduling process connection optimization method based on product structure tree according to claim 1, characterized in that, The multi-dimensional conflict detection of the process sequence includes: The process iterates through each process in the process sequence, loads the 3D model associated with each process, and calculates the volume overlap between the 3D models through geometric interference analysis; wherein, the geometric interference analysis includes detecting the intersection of the model bounding boxes and calculating the actual overlapping volume; When the volume overlap is greater than a preset safety distance threshold, it is determined to be a spatial conflict; Based on the time arrangement of the process sequence, the resource requirements of all concurrent processes are summarized at each time point; When the total demand for resources exceeds the maximum capacity of resources, it is considered a resource conflict. Construct a directed graph model of process logic relationships, wherein nodes in the directed graph model represent processes and edges represent dependencies between processes; In a directed graph model, if the time interval between each pair of adjacent processes is greater than the preset minimum interval, it is determined to be a timing conflict.

6. The intelligent scheduling process connection optimization method based on product structure tree according to claim 1, characterized in that, The conflict resolution based on the detection results includes: To address spatial conflicts, the principle of minimum displacement is applied, moving the process model involved in the interference along the shortest path until it is no longer in contact, and adjusting the spatial position attributes of the process. To address resource conflicts, a priority preemption rule is applied to automatically delay the start time of low-priority processes and allocate the resources of low-priority processes to high-priority processes. To address timing conflicts, a buffer pool borrowing rule is applied to borrow time from the buffer time of the conflicting process or adjacent processes until the minimum interval requirement is met.

7. The intelligent scheduling process connection optimization method based on product structure tree according to claim 1, characterized in that, The process of generating the optimal production schedule using a multi-objective optimization algorithm includes: Define a multi-objective optimization function to minimize the total project duration, resource costs, and risks. The objective function is expressed as: Where α, β, and γ are the weighting coefficients for each item. Indicates the total construction period. Indicates resource cost, Indicates risk indicators, Let be the probability of risk event s occurring. Let w be the influencing factor of risk event s, and w represent the number of risk events; The multi-objective optimization function is solved by a hybrid strategy combining mixed integer programming and genetic algorithm, and the constraints include resource capacity constraints, process timing constraints and space constraints. The optimal production schedule is output based on the solution results, including the process timetable, resource allocation table, risk assessment, and real-time status data in the product structure tree is updated.

8. The intelligent scheduling process connection optimization method based on the product structure tree according to claim 7, characterized in that, The method of solving the multi-objective optimization function using a hybrid strategy combining mixed integer programming and genetic algorithms includes: Initialize the genetic algorithm population, set the population size to N, and define each chromosome as an encoding of the permutation of the process sequence, the chromosome length as the total number of processes, and the gene position as the process number; Define the fitness function as F = 1 / Z, where Z is the value of the multi-objective optimization function; Mixed-integer programming is used to handle the linear constraints of multi-objective optimization functions, including resource capacity constraints, process timing constraints, and space constraints. The chromosomes are iteratively optimized using genetic algorithms involving selection, crossover, and mutation, and after each generation of purification, the offspring chromosomes are repaired using mixed integer programming to ensure that all linear constraints are met. If the change in fitness value after K consecutive iterations is less than the threshold or the maximum number of iterations is reached, then the iteration stops and the final population is obtained. Select the chromosome with the highest fitness value from the final population, decode it into process sequence and time arrangement, and generate the optimal production schedule.

9. The intelligent scheduling process connection optimization method based on product structure tree according to claim 1, characterized in that, The multi-objective optimization algorithm also includes resource coupling constraint handling: Define the resource coupling matrix ,in, The resource dependency intensity of processes i and j is represented by the formula: ;in, Let i be the set of resource requirements for process i. Let j be the set of resource requirements for process j. Let i be the spatial distance between process i and process j. The distance attenuation constant; The resource coupling constraint is defined as follows: when the resource dependency strength of process i and process j is greater than the preset strength threshold θ, process i and process j are forced to preferentially allocate the same resources or adjacent time windows.

10. The intelligent scheduling process connection optimization method based on product structure tree according to claim 1, characterized in that, The method for adjusting the optimal production schedule based on the deviation between planned and actual multi-indicator data includes: Real-time collection of production data to obtain equipment status, process progress, resource usage, and quality inspection results; Calculate the multi-indicator deviation index based on production data. The calculation formula is: ;in, For the planned time vector, For the actual time vector, Assign vectors to planned resources. Assign vectors to actual resources. This is the quality deviation value. , , These are the weight coefficients for each item, and their sum is 1; Multi-level adjustment strategies are triggered based on multi-indicator deviation indices: When the multi-indicator deviation index is less than the first threshold, the fine-tuning strategy is activated, and the time window or resource allocation of non-critical processes is automatically adjusted through the IF-THEN rule. When the multi-indicator deviation index is greater than or equal to the first threshold and less than the second threshold, a local rearrangement strategy is initiated, and a heuristic algorithm is used to reallocate resources or adjust the timing of the affected processes. When the multi-indicator deviation index is greater than or equal to the second threshold, the global reconstruction strategy is initiated, and the multi-objective optimization algorithm is called again to generate a new production schedule. The adjustment results are pushed to the production site in real time, and the real-time status data in the product structure tree is updated.

Images