A dynamic material requirement planning and multi-constraint production scheduling collaborative optimization method

Abstract

This invention belongs to the field of enterprise production and operation optimization technology. Specifically, it discloses a method for collaborative optimization of dynamic material requirements planning and multi-constraint production scheduling. The method includes: constructing a unified information field to perform dual semantic mapping on production data; generating a flexible planning blueprint with potential conflict markers based on global data; conducting refined scheduling simulations for conflicting resources; generating scheduling adjustment instructions through intelligent arbitration decisions; and outputting a fused plan after iterative feedback optimization until convergence conditions are met. Simultaneously, a hierarchical response mechanism is set up to trigger local rescheduling or global replanning during production execution based on the degree of deviation. This invention achieves dynamic adaptation between global material flow and local capacity constraints, improving planning feasibility, dynamic adaptability of the production system, and decision-making efficiency, effectively optimizing operating costs and on-time order delivery rates.

Description

Technical Field

[0001] This invention belongs to the field of enterprise production and operation optimization technology, specifically involving a method for collaborative optimization of dynamic material requirements planning and multi-constraint production scheduling. Background Technology

[0002] In modern manufacturing business management and operations, production planning is a core element guiding enterprise resource allocation and production activities. It mainly consists of two parts: material requirements planning (MRP) and production scheduling. MRP is responsible for calculating the types, quantities, and timing of required materials based on the master production schedule and product structure. Production scheduling, on the other hand, at the specific production workshop level, assigns production tasks to specific machines and personnel and determines their processing sequence, aiming to efficiently utilize limited production resources.

[0003] In existing technologies, material requirements planning (MRP) and production scheduling are typically handled independently by different information system modules, with one-way or limited information exchange between them via data interfaces. Typically, the MRP system calculates based on idealized lead times and the assumption of unlimited capacity, generating material procurement and production instructions. Subsequently, the production scheduling system receives these instructions and, considering complex constraints such as the actual equipment status, personnel skills, and tooling conditions in the workshop, performs a detailed task sequencing.

[0004] However, this serial and decoupled planning mode has certain technical defects: (1) Due to the lack of awareness of the underlying execution details, the output plan of the material requirements planning often reveals infeasibility in actual scheduling, resulting in frequent scheduling conflicts.

[0005] (2) When equipment failure or emergency order insertion occurs on the production site, the information cannot be quickly and effectively fed back to the upper material plan for linkage adjustment, resulting in a serious disconnect between planning and execution, which ultimately leads to a series of operational and management problems such as low production efficiency, unbalanced inventory levels and delayed delivery of customer orders. Summary of the Invention

[0006] In view of this, in order to solve the problems mentioned in the background technology, a collaborative optimization method of dynamic material requirements planning and multi-constraint production scheduling is proposed.

[0007] The objective of this invention can be achieved through the following technical solution: This invention provides a method for collaborative optimization of dynamic material requirements planning and multi-constraint production scheduling, including: constructing a unified information field, performing dual semantic mapping on the collected production data of the target enterprise, and generating global semantic layer data and local semantic layer data.

[0008] Based on global semantic layer data and external input demand planning, with the primary goal of ensuring the continuity and completeness of material flow, a global planning simulation is run to generate a flexible planning blueprint that includes preliminary material demand time series and potential conflict markers.

[0009] Based on the flexible planning blueprint and local semantic layer data, a refined scheduling simulation is performed on the production resources corresponding to potential conflict markers, with the secondary objective of optimizing capacity utilization. During the simulation, intelligent arbitration decisions are made for scheduling conflicts, and scheduling adjustment instructions are generated.

[0010] The scheduling adjustment command is fed back to the global planning simulation as a dynamic constraint, triggering incremental adjustments to the flexible planning blueprint, and iteratively executing refined scheduling simulation and intelligent arbitration decision-making until the convergence condition with the function of determining the termination of iteration is met, and the fused plan is output.

[0011] The fusion plan is published to the production execution unit, and production status data is monitored during execution. When the production status data deviates from the fusion plan by more than the first functional threshold used to trigger a local response, a local rescheduling is triggered. When it exceeds the second functional threshold used to trigger a global response, the global planning simulation is retried.

[0012] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. The present invention solves the technical defect of material demand and capacity scheduling being disconnected in traditional production planning by constructing a dual-modal collaborative optimization closed loop of global planning and local scheduling. The fused plan is the global optimal solution achieved after considering micro-level execution constraints, which reduces production interruptions, material backlogs and resource idleness caused by discrepancies between the plan and reality, and improves the seriousness of the plan and execution efficiency.

[0013] 2. This invention transforms complex, multi-objective scheduling conflicts into clear, quantifiable decision options through an intelligent arbitration decision-making mechanism, thereby improving decision-making efficiency and quality. Furthermore, it helps to anticipate and avoid potential production bottlenecks and cascading risks, transforming traditional passive and delayed planning adjustments into proactive and forward-looking risk management.

[0014] 3. This invention employs a hierarchical response mechanism, namely, using local rescheduling to quickly resolve small-scale deviations and triggering global replanning under major disturbances. This enables the system to maintain the stability of the main plan while responding nimbly to uncertainties in the internal and external environment, ensuring the continuity and stability of production operations in complex and volatile environments. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments 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.

[0016] Figure 1 This is a flowchart illustrating the implementation steps of the method of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Please see Figure 1 As shown, this invention provides a method for collaborative optimization of dynamic material requirements planning and multi-constraint production scheduling. The specific steps are as follows: construct a unified information field, perform dual semantic mapping on the collected target enterprise production data, and generate global semantic layer data and local semantic layer data.

[0019] It should be noted that the unified information field achieves multi-source data access through standardized interface adapters, wherein: the interface with the Enterprise Resource Planning (ERP) unit adopts the JDBC protocol with a default polling interval of 20 minutes; the interface with the Manufacturing Execution Unit (MLU) adopts the OPC UA protocol with a data subscription update frequency of 1 second; the interface with the Internet of Things (IoT) platform adopts the MQTT v3.1 protocol with a QoS level of 2; and data storage adopts a time-series database, organized by a three-dimensional index of resource ID-timestamp-data type.

[0020] As a feasible embodiment of the present invention, the specific generation process of the global semantic layer data and the local semantic layer data includes: obtaining the target enterprise's production data from the enterprise resource planning unit, the manufacturing execution unit, and the Internet of Things platform.

[0021] Specifically, the target enterprise's production data is obtained from data sources such as the Enterprise Resource Planning Unit, Manufacturing Execution Unit, and Internet of Things platform through a preset interface adapter. The specific process includes: (1) using a polling mechanism with an interval of 10 to 30 minutes to extract order, bill of materials, and master production schedule data through the database connector of the Enterprise Resource Planning Unit; (2) using the Manufacturing Execution Unit to configure a real-time data subscription service to capture work order status, work-in-process flow, and equipment time records updated in seconds; and (3) using the subscription mode of the MQTT protocol of the Internet of Things platform to receive equipment status codes and operating parameters from sensors and PLCs in real time.

[0022] The order, bill of materials and master production schedule data, work order status, work-in-process turnover and equipment time records, equipment status codes and operating parameters are collectively referred to as the target enterprise's production data.

[0023] The data representing production resources in the production data are mapped to resource unit descriptions with time-material conversion capabilities, forming global semantic layer data.

[0024] Specifically, the generation of the global semantic layer data aims to abstract physical production resources into quantifiable and predictable logical production capacity units. The specific implementation process includes: identifying entities representing production resources from production data, including but not limited to machine tool numbers and production line identifiers, and associating their static attributes from manufacturing execution units, including but not limited to standard processing cycle time and rated utilization rate. This information is then transformed into a resource unit description with time-to-material conversion capabilities, i.e., establishing a dynamic capacity function for each resource unit. The specific formula for calculating this function is as follows: ,in It is at a future point in time. The expected output rate, measured in the number of items produced per unit time; It is the rated rate when a resource unit processes a specific material, obtained from the process database; It is a dynamic comprehensive efficiency coefficient based on historical data and real-time status prediction. It is a value between 0 and 1, which integrates factors such as equipment failure rate and personnel efficiency. It is the resource availability coefficient, which is also a value between 0 and 1. It is calculated based on the preset equipment maintenance plan and shift schedule. Its value is 0 during maintenance or non-working hours.

[0025] It should be added that the specific formula for calculating the dynamic comprehensive efficiency coefficient is as follows: ,in These are the weighting coefficients, and Historical average efficiency is calculated based on the ratio of the actual output rate of the resource unit to the rated rate over the past three months; real-time status correction coefficient is determined based on the current status code of the equipment, such as 1.0 for normal status, 0.7 for minor faults, and 0.9 for heavy load status.

[0026] It should be noted that by forming global semantic layer data, discrete equipment status data is transformed into continuous, future-oriented global semantic layer data, enabling the global planning simulation module to perform effective capacity assessment.

[0027] The data representing production tasks and process logic in the production data are mapped to task package descriptions with spatiotemporal constraint chains, forming local semantic layer data.

[0028] Specifically, the generation of the local semantic layer data aims to decompose production instructions into the smallest operational units with strict logical constraints that can be directly executed by the scheduling engine. The specific implementation process includes: parsing the work order information and bill of materials in the production data, and decomposing a complete production order into a series of sequentially performed operations based on the process route defined in the manufacturing execution unit. Each operation is encapsulated into a task package description, which specifically contains a data structure including the operation ID, required resource type, standard working hours, bill of materials, and quality inspection standards.

[0029] Furthermore, a spatiotemporal constraint chain connecting the descriptions of each task package is constructed. Specifically, the spatiotemporal constraint chain is a directed acyclic graph, where the nodes are task packages, and the directed edges define the mandatory sequential relationship between processes, the minimum and maximum time intervals such as cooling or transfer time, ranging from 5 minutes to 4 hours, and the resource transfer relationship, such as the semi-finished product that has completed process A must be transferred to the equipment where process B is located.

[0030] Based on global semantic layer data and external input demand planning, with the primary goal of ensuring the continuity and completeness of material flow, a global planning simulation is run to generate a flexible planning blueprint that includes preliminary material demand time series and potential conflict markers.

[0031] As a feasible embodiment of the present invention, the specific generation process of the flexible planning blueprint includes: based on the input demand plan, traversing the bill of materials, and calculating the gross and net requirements of materials at each level.

[0032] Specifically, based on the received demand plan and bill of materials, standard material requirements calculations are performed. The specific process includes: starting with the final product, recursively traversing the bill of materials layer by layer to calculate the gross requirements of each level of materials; combining current inventory data and in-transit material information to calculate the net requirements of each level of materials; and determining the latest demand date of each material based on its standard lead time.

[0033] Based on the nominal capacity described by the resource unit, the process of satisfying the net demand is simulated on the time axis to generate a preliminary material demand time series.

[0034] Specifically, based on net demand, the nominal capacity described by each resource unit in the unified information field is invoked to simulate the process of satisfying the net demand on the time axis. The specific process includes: for each net demand, identifying the corresponding resource unit required for production, and calculating the time required to complete the demand based on the dynamic capacity function of the resource unit. This process transforms discrete material demand points into a series of preliminary production tasks with start and end times. These tasks are arranged in reverse order on the time axis according to their latest demand date, thereby generating a preliminary material demand time series.

[0035] The analysis of the preliminary material demand time series reveals the load competition generated by the same resource unit within a set time period, identifies capacity bottleneck hotspots, marks these bottleneck hotspots as potential conflict indicators, and integrates them with the preliminary material demand time series to obtain a flexible planning blueprint.

[0036] Specifically, load analysis is performed on the generated preliminary material demand time series to identify capacity bottleneck characteristics. The process includes: dividing the future production cycle into continuous time intervals, denoted as set time periods (e.g., 8-hour shifts). Within each set time period, the load of each resource unit is quantitatively assessed. This quantitative assessment is achieved by calculating a capacity load factor, with the specific formula as follows: ,in It is the capacity load factor, which is a dimensionless proportional value; It is the total processing time required for all tasks allocated to resource units within a set time period based on the initial material demand time series, expressed in hours. This is the total available production time of a resource unit within a set time period, expressed in hours. This value is obtained from the scheduling and maintenance plan in the resource unit description.

[0037] When the calculated resource unit If the load exceeds a preset capacity load threshold, such as a configurable parameter between 0.9 and 1.1, the resource unit is determined to constitute a capacity bottleneck hotspot during the time period. This capacity bottleneck hotspot is then marked as a potential conflict and attached to the tasks involving the resource unit and the time period in the preliminary material demand time series.

[0038] This invention employs global planning simulations to generate a flexible planning blueprint that includes preliminary material demand time series and potential conflict markers. The aim is to transform the macro-level demand plan into a time-dimensional plan with resource conflict warnings, providing a clear optimization focus for subsequent refined scheduling. This process does not involve exhaustive scheduling, but rather uses rapid, nominal-capacity-based extrapolation to proactively identify medium- to long-term resource bottlenecks.

[0039] This invention solves the technical defect of material demand and capacity scheduling being disconnected in traditional production planning by constructing a dual-modal collaborative optimization closed loop of global planning and local scheduling. The integrated plan is a globally optimal solution achieved after considering micro-level execution constraints, which reduces production interruptions, material backlogs and resource idleness caused by discrepancies between plans and reality, and improves the seriousness and execution efficiency of the plan.

[0040] Based on the flexible planning blueprint and local semantic layer data, a refined scheduling simulation is performed on the production resources corresponding to potential conflict markers, with the secondary objective of optimizing capacity utilization. During the simulation, intelligent arbitration decisions are made for scheduling conflicts, and scheduling adjustment instructions are generated.

[0041] As a feasible embodiment of the present invention, the specific implementation process of the refined scheduling simulation includes: extracting the preliminary material demand time series and potential conflict markers from the flexible planning blueprint, and obtaining the material demand time window corresponding to each task in the preliminary material demand time series; the material demand time window is the production start and end time interval corresponding to the latest demand date of the task.

[0042] Obtain the task package and spatiotemporal constraint chain associated with the production resource corresponding to the potential conflict marker from the local semantic layer data.

[0043] Based on the spatiotemporal constraint chain, the tasks corresponding to the production resources with potential conflict markers are arranged on the time axis according to the sequence of processes and resource availability.

[0044] The execution process of each task is simulated one by one to verify whether each task can be completed within its corresponding material demand time window without violating the time and space constraints.

[0045] If the simulation determines that a task cannot simultaneously meet the material demand time window, resource availability, and time-space constraints, a scheduling conflict is triggered, and an intelligent arbitration decision is executed.

[0046] As a feasible embodiment of the present invention, the specific process of making intelligent arbitration decisions for scheduling conflicts includes: when the refined scheduling simulation determines that a certain material demand time window in the flexible planning blueprint cannot be met, based on the production process network, analyzing the chain impact of not meeting the demand on upstream material inventory, downstream order delivery and parallel production links, and generating an impact chain analysis report.

[0047] Specifically, in the refined scheduling simulation, when the scheduling engine attempts to place a task package description on its specified resource unit time axis, it finds that within its material demand time window, no set time period can simultaneously satisfy all constraints of resource availability and the completion time of preceding tasks. Once the conflict is determined, the impact chain analysis is immediately initiated. The specific process includes: using the production process network defined in the local semantic layer data, i.e. the directed acyclic graph between processes, to perform bidirectional graph traversal: (1) Traverse upstream to trace the input materials required by the task package, calculate the duration of the increase in upstream material work-in-process inventory if the task is delayed, and multiply by the unit inventory holding cost to obtain the inventory backlog cost; (2) Traverse downstream to track all subsequent processes to which the output materials of the task package will flow, until the final product, and calculate the impact of this delay on the final order delivery date, quantifying it as a delivery delay penalty according to the contract terms. At the same time, assess the potential waiting time of other parallel tasks caused by this conflict, quantifying it as resource idle cost.

[0048] Among them, the unit inventory holding cost is obtained from the cost management module of the enterprise resource planning unit and preset according to the material type; the late delivery penalty is extracted according to the order contract terms, and if there are no explicit terms, it is calculated based on the target enterprise's annual average default cost.

[0049] Based on the impact chain analysis report and the configuration parameters of the preset optimization target weight allocation function, at least two scheduling adjustment plans are generated; each scheduling adjustment plan includes adjustment schemes for material demand time windows, production resource allocation or production batch splitting and their corresponding global impact assessment parameters.

[0050] Specifically, the scheduling adjustment plan is a built-in conflict resolution strategy. For example, Plan 1 is to accept the delay and schedule the task package at the first feasible future time point; Plan 2 is to activate backup resources or work overtime to produce, which will generate additional operating costs; Plan 3 is to split production batches to accommodate fragmented available time, which will increase changeover setup costs.

[0051] For each contingency plan, a small-scale scheduling simulation is performed again, and the impact chain analysis module is called to calculate its corresponding global total cost. This global total cost is then used as a parameter for the global impact assessment of the contingency plan. The specific calculation formula is as follows: ,in It is the total cost of the overall plan; This refers to the cost of inventory backlog resulting from the implementation of the contingency plan; These are penalties for delivery delays incurred after the implementation of the contingency plan; This is due to an increase in the direct operating costs of the contingency plan itself, such as overtime pay or additional setup fees; These are preset, dimensionless optimization target weights, which sum to 1 and are configured by production managers based on the current business strategy.

[0052] It should be added that the optimization objective weights Configured by the user through the human-computer interaction interface, the default configuration is: order delivery priority scenario. In scenarios where cost optimization is the priority .

[0053] The optimal scheduling adjustment plan is selected from all scheduling adjustment plans based on the smallest global impact assessment parameters. The specific adjustment operations contained in the optimal plan are then used as scheduling adjustment instructions.

[0054] Specifically, specific execution actions are parsed from the optimal plan, such as adjusting the time window of task package A to a new start and end time or assigning task package B to standby device Z. These specific actions are then encapsulated into a standardized data object, which serves as the final output scheduling adjustment instruction and is fed back to the collaborative optimization iteration engine.

[0055] As a feasible embodiment of the present invention, after generating at least two scheduling adjustment plans, a decision complexity assessment step is also included. The specific process includes: pushing all generated scheduling adjustment plans and their corresponding global impact assessment parameters to the human-computer interaction interface.

[0056] Specifically, after the intelligent arbitration decision generates at least two scheduling adjustment plans, a second decision complexity assessment is performed. This assessment is based on preset rules. For example, if the difference in the global impact assessment parameters of all plans is less than a certain percentage, such as 5%, or if the conflict involves orders marked as strategic customers, then it is determined that manual intervention is required for this decision. At this point, the system does not automatically select the optimal plan directly, but instead pushes all scheduling adjustment plans and their corresponding global impact assessment parameters to the human-computer interaction interface through a standard API interface.

[0057] It should be noted that the human-computer interaction interface is a visual dashboard that clearly displays each scheduling adjustment plan in the form of side-by-side cards or Gantt charts. Each card not only lists the specific adjustment content of the plan, such as delaying task A by 8 hours or activating overtime production for order B, but also intuitively presents its corresponding global impact assessment parameters in the form of data charts, such as the specific amount of cost increase, the expected delivery delay days, and the impact on the load factor of relevant resources.

[0058] Receive the pre-selection command sent by the user from the human-computer interaction interface.

[0059] Specifically, the system waits to receive a plan selection instruction from the user via the human-computer interaction interface. After reviewing and comparing the various plans on the interface, the planner makes a decision by clicking the confirmation button. This operation generates a plan selection instruction containing a unique identifier for the selected plan.

[0060] The user-selected plan is taken as the final optimal plan, and scheduling adjustment instructions are generated accordingly.

[0061] Specifically, upon receiving a contingency plan selection instruction, the contingency plan specified in the instruction is confirmed as the final optimal contingency plan.

[0062] It should be noted that the priority of the optimal contingency plan obtained through manual selection is higher than that of the optimal contingency plan obtained through automatic selection based on the lowest global impact assessment parameter.

[0063] The human-computer interaction decision-making process introduced in this invention aims to address complex scheduling conflicts that have a significant impact or are difficult to judge optimally based solely on quantitative data. It combines the computational and analytical capabilities of machines with the experience, wisdom, and strategic insight of human planners to form a human-computer collaborative decision-making model, thereby making more robust and commercially sound decisions at critical junctures.

[0064] The scheduling adjustment command is fed back to the global planning simulation as a dynamic constraint, triggering incremental adjustments to the flexible planning blueprint, and iteratively executing refined scheduling simulation and intelligent arbitration decision-making until the convergence condition with the function of determining the termination of iteration is met, and the fused plan is output.

[0065] As a feasible embodiment of the present invention, the specific process of triggering the incremental adjustment of the flexible planning blueprint includes: parsing the scheduling adjustment instruction and extracting the key modification content therein.

[0066] Specifically, the scheduling adjustment instructions are parsed, their data structures are read, and key modifications are extracted. These typically include identifiers describing one or more task packages and updates to their associated attributes, such as the start and end points of material requirement time windows, identifiers of resource usage, or changes in the number of production batches.

[0067] The key modifications are transformed into new constraints, and the input to the global planning simulation is updated.

[0068] Specifically, key modifications are transformed into new constraints that can be understood by the global planning simulation module.

[0069] For example, if the scheduling adjustment instruction changes the time window of task package X from Adjusted to This will generate a new constraint that states that the material requirements for task package X must be met between T3 and T4.

[0070] If the instruction assigns task package Y to standby device Z, the new constraint is that the capacity consumption of task package Y should be transferred from the original device to device Z for calculation.

[0071] Based on the preliminary material demand time series that has already been generated, the material demand in the local preset time period affected by the new constraints is re-simulated and adjusted to form an updated flexible planning blueprint.

[0072] Specifically, incremental adjustments are intended to avoid unnecessary delays and schedule jitter caused by a complete recalculation. The specific process of forming the updated flexible planning blueprint includes: identifying the task packages directly affected by the new constraints and their upstream and downstream related materials on the bill of materials, thereby defining a local, affected preset time period, which is usually set to 1.5 to 2 times the standard production cycle of the relevant materials.

[0073] Within this specific, pre-defined time period, a demand fulfillment simulation based on a dynamic capacity function is rerun for the affected materials, strictly adhering to the newly added constraints. After the simulation is complete, this locally adjusted time series is replaced with the corresponding positions in the original preliminary material demand time series, forming an updated flexible planning blueprint.

[0074] It should be noted that the updated blueprint was then resubmitted to the local scheduling and arbitration module to verify whether the new global plan eliminated the original scheduling conflicts or introduced new problems, thereby driving the next iteration.

[0075] It should also be noted that after the incremental adjustment of the flexible planning blueprint, the capacity load factor needs to be recalculated. If the newly added resource unit exceeds the capacity load threshold, it will be marked as a new potential conflict marker and included in the next round of refined scheduling simulation.

[0076] The engineering objective of triggering incremental adjustments to the flexible planning blueprint in this invention is to efficiently integrate feedback information from local scheduling conflict solutions back into the global material plan, avoiding a disruptive recalculation of the entire plan. This process ensures the overall stability of the global plan through precise local corrections, while rapidly responding to changes in underlying execution constraints.

[0077] As a feasible embodiment of the present invention, the convergence condition includes at least one of the following conditions: (1) the number of iterations reaches the upper limit value.

[0078] Specifically, checking whether the number of iterations has reached the upper limit is a protective measure to prevent infinite loops. The system has an iteration counter, initially set to 0, which increments by 1 each time a global adjustment-local arbitration loop is completed. When the counter reaches a preset upper limit for the number of iterations, such as an integer between 10 and 50, the iteration will terminate regardless of whether other conditions are met. This upper limit is set based on historical computational performance and experience with plan complexity.

[0079] (2) The overall estimated cost change rate of the integration plan is lower than the cost change threshold.

[0080] Specifically, assess the rate of change of the overall projected cost of the convergence plan. At the end of each iteration, calculate the overall projected cost of the convergence plan based on the current flexible planning blueprint and refined scheduling simulation results.

[0081] The overall estimated cost change rate is a comprehensive indicator, which includes, but is not limited to, material holding costs, production and operating costs, and potential delay penalties. Specifically, the overall estimated cost change rate is the relative change rate between the estimated cost of the current iteration and the estimated cost of the previous iteration, and its specific calculation formula is as follows: ,in It is the overall estimated cost change rate. It is the total cost calculated in the current iteration. That is the total cost of the previous round. When If the value is less than the preset cost threshold, such as a decimal between 0.001 and 0.01, then it is considered that the cost optimization effect brought by further iterations is no longer significant, and the convergence condition is met.

[0082] (3) All high-priority material requirements in the flexible planning blueprint have been confirmed and satisfied by the scheduling adjustment instructions.

[0083] Specifically, in each iteration, each material requirement on the bill of materials is checked to see if it has been simulated through refined scheduling in the current flexible planning blueprint, and whether its final scheduling adjustment instruction is confirmed to be met, meaning no delay or batch reduction is required. When the status of all high-priority material requirements on the bill of materials is confirmed to be met, the core planning objective is considered to have been achieved, and the convergence condition is met.

[0084] It should be noted that once any convergence condition is triggered, the collaborative optimization iterative engine stops looping and packages the current optimal elastic plan blueprint and scheduling scheme as the final fusion plan output.

[0085] This invention transforms complex, multi-objective scheduling conflicts into clear, quantifiable decision options through an intelligent arbitration decision-making mechanism, thereby improving decision-making efficiency and quality. Furthermore, it helps to anticipate and mitigate potential production bottlenecks and cascading risks, transforming traditional passive and delayed planning adjustments into proactive and forward-looking risk management.

[0086] The fusion plan is published to the production execution unit, and production status data is monitored during execution. When the production status data deviates from the fusion plan by more than the first functional threshold used to trigger a local response, a local rescheduling is triggered. When it exceeds the second functional threshold used to trigger a global response, the global planning simulation is retried.

[0087] As a feasible embodiment of the present invention, the specific process of triggering local rescheduling includes: extracting the current production status data that deviates from the fusion plan from the production status data, thereby determining the affected production resources and time range.

[0088] Specifically, the planning, execution, and monitoring module continuously compares the actual work order completion time obtained from the manufacturing execution unit with the scheduled time in the integrated plan. When the time difference exceeds a first preset threshold, such as 15 to 60 minutes, a partial rescheduling is triggered.

[0089] The specific process for determining the affected production resources and time range includes: starting with the work order where the deviation occurred, tracing downstream along the production process network to identify all subsequent tasks that directly or indirectly depend on the output of that work order, and marking the production resources involved in these tasks as affected production resources. Simultaneously, starting from the current time point, a preset local rearrangement cycle, typically 8 to 24 hours in the future, is extended forward as the affected time range.

[0090] For unaffected production tasks, starting from the latest state of affected production resources within the stated time frame, re-run refined scheduling simulation and intelligent arbitration decision-making to generate local adjustment plans.

[0091] Specifically, the status of all production tasks outside the affected production resources and time range is locked, making them unadjustable in this calculation, thereby fixing the unaffected production tasks.

[0092] The rescheduling starts with the latest state of the production resource at the current moment where the deviation occurred. For example, if it is a equipment failure, its state is updated to unavailable with an estimated repair time. The module takes all incomplete and unstarted tasks within the affected area as input and re-invokes the refined scheduling simulation and intelligent arbitration decision-making module. Under fixed boundary conditions and updated resource states, this module generates an optimized and feasible local adjustment plan for the local task set.

[0093] The unaffected parts of the partial adjustment plan and the integration plan will be merged to form a revised plan, which will then be issued.

[0094] Specifically, the plan data corresponding to the affected time range and resource portion in the original integration plan is discarded, and the gap is filled with data from the newly generated partial adjustment plan, thus seamlessly connecting with the unaffected portion to form a complete, time-continuous revision plan. This revision plan is then distributed to the relevant manufacturing execution system terminals, equipment control consoles, or material distribution systems to update the execution instructions at the shop floor level.

[0095] The purpose of triggering local rescheduling in this invention is to quickly and accurately digest the deviation in the affected local area when the deviation in the production execution process does not exceed the upper limit of the system's adaptive capability, and generate a correction scheme that minimizes global disturbance, thereby restoring the executability of the plan without interrupting the overall production rhythm.

[0096] The specific process of re-triggering the global planning simulation includes: updating the unified information field using all production status data at the current moment as the new initial state.

[0097] Specifically, the second preset threshold is usually a composite indicator, such as a critical equipment failure time of more than 4 hours, a core material supply interruption of more than one shift, or a cumulative order delay rate of more than 5%.

[0098] If a global response is triggered, a complete update of the unified information field is performed, using all production status data at the current moment as the new initial state. This forcibly refreshes the location and quantity of all work-in-process inventory, the availability and load of all resource units, and the latest estimated arrival time of all materials in transit.

[0099] Based on the updated unified information field and the incomplete requirements plan, the entire process from running the global planning simulation to outputting the new fusion plan is restarted.

[0100] Specifically, identifying all unfinished demand plans involves: traversing all currently open sales and production orders, removing completed portions, and using the remaining demand as input for this replanning.

[0101] By invoking the top-level collaborative optimization process, starting with running a global planning simulation, the entire collaborative optimization process from running the global planning simulation to outputting a new fusion plan is re-executed. This means generating a completely new flexible planning blueprint based on a completely new initial state and remaining demand, and then iteratively optimizing it through multiple rounds of local scheduling simulations and intelligent arbitration decisions until convergence conditions are met. Finally, a completely new fusion plan adapted to the significant disturbance is generated and released, thereby replacing the old plan that has become invalid and guiding production activities back to an orderly state.

[0102] This invention employs a hierarchical response mechanism, namely, using local rescheduling to quickly resolve small-scale deviations and triggering global replanning under major disturbances. This enables the system to maintain the stability of the main plan while responding nimbly to uncertainties in the internal and external environment, ensuring the continuity and stability of production operations in complex and volatile environments.

[0103] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0104] Those skilled in the art will recognize that the algorithmic steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0105] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0106] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0107] Finally, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for collaborative optimization of dynamic material requirements planning and multi-constraint production scheduling, characterized in that: include: A unified information field is constructed, and dual semantic mapping is performed on the collected target enterprise production data to generate global semantic layer data and local semantic layer data. Based on global semantic layer data and external input demand planning, with the primary goal of ensuring the continuity and completeness of material flow, a global planning simulation is run to generate a flexible planning blueprint that includes preliminary material demand time series and potential conflict markers. Based on the flexible planning blueprint and local semantic layer data, for the production resources corresponding to potential conflict markers, with optimizing capacity utilization as the secondary objective, a refined scheduling simulation is performed, and intelligent arbitration decisions are made for scheduling conflicts during the simulation process to generate scheduling adjustment instructions. The scheduling adjustment command is fed back to the global planning simulation as a dynamic constraint, triggering incremental adjustments to the flexible planning blueprint, and iteratively executing refined scheduling simulation and intelligent arbitration decision-making until the convergence condition with the function of determining the termination of iteration is met, and the fused plan is output. Deploy the integration plan to the production execution unit and monitor production status data during execution; When the deviation of production status data from the fusion plan exceeds the first functional threshold used to trigger a local response, a local rescheduling is triggered; when it exceeds the second functional threshold used to trigger a global response, a global planning simulation is retried.

2. The method for collaborative optimization of dynamic material requirements planning and multi-constraint production scheduling according to claim 1, characterized in that: The specific generation process of the global semantic layer data and the local semantic layer data includes: Obtain target enterprise production data from enterprise resource planning units, manufacturing execution units, and IoT platforms; The data representing production resources in the production data are mapped to resource unit descriptions with time-material conversion capabilities, forming global semantic layer data; The data representing production tasks and process logic in the production data are mapped to task package descriptions with spatiotemporal constraint chains, forming local semantic layer data.

3. The method for collaborative optimization of dynamic material requirements planning and multi-constraint production scheduling according to claim 2, characterized in that: The specific process for generating the flexible planning blueprint includes: Based on the input demand plan, traverse the bill of materials and calculate the gross and net requirements for materials at each level; Based on the nominal capacity described by the resource unit, the process of satisfying the net demand is simulated on the time axis to generate a preliminary material demand time series. The analysis of the preliminary material demand time series reveals the load competition generated by the same resource unit within a set time period, identifies capacity bottleneck hotspots, marks these bottleneck hotspots as potential conflict indicators, and integrates them with the preliminary material demand time series to obtain a flexible planning blueprint.

4. The method for collaborative optimization of dynamic material requirements planning and multi-constraint production scheduling according to claim 3, characterized in that: The specific implementation process of the refined scheduling simulation includes: Extract the preliminary material requirements time series and potential conflict markers from the flexible planning blueprint, and obtain the material requirements time window corresponding to each task in the preliminary material requirements time series; the material requirements time window is the production start and end time interval corresponding to the latest requirement date of the task. Obtain the task package and spatiotemporal constraint chain associated with the production resources corresponding to the potential conflict markers from the local semantic layer data; Based on the spatiotemporal constraint chain, the tasks corresponding to the production resources with potential conflict markers are arranged on the time axis according to the sequence of processes and resource availability. The execution process of each task is simulated one by one to verify whether each task can be completed within its corresponding material demand time window without violating the time and space constraints. If the simulation determines that a task cannot simultaneously meet the material demand time window, resource availability, and time-space constraints, a scheduling conflict is triggered, and an intelligent arbitration decision is executed.

5. The method for collaborative optimization of dynamic material requirements planning and multi-constraint production scheduling according to claim 4, characterized in that: The specific process of making intelligent arbitration decisions for scheduling conflicts includes: When the refined scheduling simulation determines that a certain material demand time window in the flexible planning blueprint cannot be met, the chain reaction of not meeting the demand on upstream material inventory, downstream order delivery and parallel production links is analyzed based on the production process network, and an impact chain analysis report is generated. Based on the impact chain analysis report and the configuration parameters of the preset optimization target weight allocation function, at least two scheduling adjustment plans are generated; each scheduling adjustment plan includes adjustment schemes for material demand time windows, production resource allocation or production batch splitting and their corresponding global impact assessment parameters. The optimal scheduling adjustment plan is selected from all scheduling adjustment plans based on the smallest global impact assessment parameters. The specific adjustment operations contained in the optimal plan are then used as scheduling adjustment instructions.

6. The method for collaborative optimization of dynamic material requirements planning and multi-constraint production scheduling according to claim 5, characterized in that: After generating at least two scheduling adjustment plans, a decision complexity assessment step is also included, the specific process of which includes: Push all generated scheduling adjustment plans and their corresponding global impact assessment parameters to the human-computer interaction interface; Receive the user's pre-selection instruction sent from the human-computer interaction interface; The user-selected plan is taken as the final optimal plan, and scheduling adjustment instructions are generated accordingly.

7. The method for collaborative optimization of dynamic material requirements planning and multi-constraint production scheduling according to claim 1, characterized in that: The specific process for triggering incremental adjustments to the elastic planning blueprint includes: Analyze the scheduling adjustment instruction and extract the key modifications; Transform the key modifications into new constraints and update the input to the global planning simulation; Based on the preliminary material demand time series that has already been generated, the material demand in the local preset time period affected by the new constraints is re-simulated and adjusted to form an updated flexible planning blueprint.

8. The method for collaborative optimization of dynamic material requirements planning and multi-constraint production scheduling according to claim 1, characterized in that: The convergence condition includes at least one of the following conditions: (1) The number of iterations reaches the upper limit; (2) The overall estimated cost change rate of the integration plan is lower than the cost change threshold; (3) All high-priority material requirements in the flexible planning blueprint have been confirmed and satisfied by the scheduling adjustment instructions.

9. The method for collaborative optimization of dynamic material requirements planning and multi-constraint production scheduling according to claim 1, characterized in that: The specific process of triggering local rescheduling includes: Extract current production status data that deviates from the integration plan from the production status data to determine the affected production resources and time range; For unaffected production tasks, starting from the latest state of affected production resources within the stated time frame, rerun the refined scheduling simulation and intelligent arbitration decision-making to generate a localized adjustment plan; The unaffected parts of the partial adjustment plan and the integration plan will be merged to form a revised plan, which will then be issued.

10. The method for collaborative optimization of dynamic material requirements planning and multi-constraint production scheduling according to claim 1, characterized in that: The specific process of re-triggering the global planning simulation includes: Update the unified information field using all production status data at the current moment as the new initial state; Based on the updated unified information field and the incomplete requirements plan, the entire process from running the global planning simulation to outputting the new fusion plan is restarted.

Images