A Smart Question Answering and Decision Support Method for Production Management Systems

Abstract

This invention relates to the field of production management technology, specifically disclosing an intelligent question-and-answer and decision-making support method for production management systems. This method addresses the problem in existing production management systems where question-and-answer results and decision suggestions remain as independent information outputs, failing to correspond to the actual production execution chain. The method includes constructing a production object association topology, generating commitment execution units and commitment penetration chains, constructing forward execution mirrors and reverse compensation mirrors, filtering target execution paths, and generating question-and-answer text and production execution instructions with closed-loop updates. This invention achieves unified constraints on the same target execution path for both question-and-answer results and production execution instructions through commitment execution unit objectification, commitment penetration chain expansion, forward execution mirror and reverse compensation mirror deduction, and fallback window parameter filtering.

Description

Technical Field

[0001] This invention relates to the field of production management technology, and in particular to an intelligent question-answering and decision support method for production management systems. Background Technology

[0002] Existing production management systems typically include business modules such as order management, work order management, equipment management, inventory management, quality management, and personnel scheduling management. They can process needs such as production progress inquiry, abnormal status identification, material supply judgment, equipment operation judgment, and scheduling suggestion generation through database queries, knowledge base retrieval, rule matching, statistical analysis, or question-and-answer interaction. For production-related questions raised by users, existing technologies usually extract search conditions based on the question content, then obtain relevant results from corresponding business data, rule data, or historical records, and output text answers or suggestions based on the obtained results.

[0003] However, in actual production, there are continuous constraints between order execution, process connection, equipment occupancy, inventory supply, quality inspection, and shift scheduling. Moreover, the on-site status will continue to change with equipment depreciation, material shortages, quality re-inspection, and shift adjustments. Most existing methods generate output content for each single query or suggestion, lacking continuous analysis of the execution link corresponding to the output content. They also lack pre-verification of compensation paths and rollback boundaries under execution interruption conditions. As a result, although the answers or suggestions generated by the system can meet the requirements of local queries or local scheduling, they are prone to inconsistencies with the actual resource status, obstruction of subsequent link connections, and lack of rollback basis after anomalies once they enter the production site. Summary of the Invention

[0004] The technical problem to be solved by this invention is to provide an intelligent question-and-answer and decision-making support method for production management systems. By performing forward fulfillment analysis and reverse compensation analysis on the execution link before the question-and-answer text is output, and by filtering out unrecoverable paths with backoff window parameters before the production execution instructions are issued, the final output result can simultaneously have link fulfillment, anomaly backoffability, and subsequent closed-loop correction capability.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] An intelligent question-answering and decision support method for production management systems includes:

[0007] Step S1: Receive production problem instructions sent by the user terminal and extract the identifiers and time windows of orders, processes, equipment, and materials;

[0008] Step S2: Collect status parameters from the order execution unit, process equipment unit, inventory supply unit, quality inspection unit, and shift operation unit, and establish a production object association topology according to the relationship of order subordinate, process sequence, equipment occupation, inventory supply, quality inspection, and shift allocation.

[0009] Step S3: Based on the order identifier, process identifier, equipment identifier, material identifier, and time window, match the target production object node in the production object association topology, and encapsulate the matching result with the action type, execution time window, dependent resources, and verification conditions into a commitment execution unit;

[0010] Step S4: Starting from the starting node of the commitment execution unit, proceed chain by chain along the order constraint chain, process connection chain, equipment occupancy chain, inventory supply chain and shift capacity chain to generate the commitment penetration chain;

[0011] Step S5: Generate a forward execution image and a reverse compensation image based on the commitment penetration chain; calculate path perturbation parameters and backoff window parameters for the preset counterfactual scenario; and select the target execution path based on the execution constraints.

[0012] Step S6: Generate question and answer text and production execution instructions based on the target execution path, send them to the execution terminal, and receive the returned status parameters to update the production object association topology and commitment execution unit.

[0013] It should be noted that this invention does not add execution verification after the question-and-answer text is generated, nor does it add exception handling after the scheduling suggestion is directly generated. Instead, it binds the question-and-answer text and the production execution instruction to the same target execution path, so that the question-and-answer text and the production execution instruction correspond to the same production link that has undergone execution disturbance analysis and rollback boundary analysis at the source of generation.

[0014] As a further aspect of the present invention, in step S2, the status parameters of the order execution unit include order number, process number, planned start time, planned end time, and current completion rate; the status parameters of the process equipment unit include equipment number, equipment process type, current load rate, and available time period; the status parameters of the inventory supply unit include material number, available quantity, locked quantity, and replenishment time; the status parameters of the quality inspection unit include inspection batch number, inspection status, and re-inspection duration; and the status parameters of the shift operation unit include shift number, number of workers, shift start time, and shift end time.

[0015] It should be noted that by using order, equipment, inventory, quality, and shift status parameters as the input basis for the same execution link, this invention enables the subsequently generated target execution path to not only reflect the local state of a single business object, but also simultaneously reflect order progress, equipment capacity, material accessibility, quality inspection constraints, and shift capacity boundaries.

[0016] As a further aspect of the present invention, in step S2, the production object association topology includes order nodes, process nodes, equipment nodes, inventory nodes, quality nodes, and shift nodes; the order subordination relationship is established based on the order process route table, the process sequence relationship is established based on the process order table, the equipment occupancy relationship is established based on the equipment allocation table, the inventory supply relationship is established based on the material demand table and the inventory supply table, the quality inspection relationship is established based on the quality inspection process table, and the shift allocation relationship is established based on the shift scheduling table.

[0017] It should be noted that, unlike existing technologies that only use business relationships for query retrieval or graph recall, the production object association topology in this invention is directly used as the basis for expanding the commitment penetration chain, transforming multi-source business relationships from queryable relationships to executable propagation relationships. As a result, the subsequent analysis objects are no longer discrete candidate results, but rather a continuously propagating execution chain.

[0018] As a further embodiment of the present invention, in step S3, the commitment execution unit includes a commitment subject identifier, a commitment object identifier, an action type, an execution time window, a set of dependent devices, a set of dependent materials, a set of verification conditions, and a rollback identifier; the action type is one of progress query, production scheduling adjustment, material allocation, equipment switching, or abnormal handling; the commitment execution unit is encapsulated by the target production object node, the action type, the execution time window, the set of dependent devices, the set of dependent materials, and the set of verification conditions.

[0019] It should be noted that this invention transforms the originally scattered question-and-answer results and suggestion results into a unified execution object with time windows, resource dependencies, and verification conditions through a commitment execution unit. This makes the question-and-answer content no longer just text answers for users, and the suggestion content no longer just static recommendations for scheduling, but a unified technical object that can continue to unfold, deduce, and filter along the production chain.

[0020] As a further aspect of the present invention, in step S4, when generating the commitment penetration chain, the chain is expanded sequentially in the order of order constraint chain, process connection chain, equipment occupancy chain, inventory supply chain and shift capacity chain; each commitment penetration chain records the path start node, path end node, link node sequence, node entry time, node exit time and constraint relationship; the node entry time and node exit time are determined according to the planned time, equipment available time period, process duration, inventory replenishment time and shift start and end time.

[0021] It should be noted that this invention does not simply select the most relevant path in the graph structure, but rather constructs a commitment penetration chain with time and resource attributes along the preset production constraint sequence, so that a production problem can be continuously propagated from the initial production object to the subsequent objects affected by it, providing a continuous link foundation for subsequent dual-mirror inference.

[0022] As a further aspect of the present invention, in step S5, the forward execution mirror is the target state sequence of each link node when the commitment execution unit executes according to the commitment penetration chain. The target state sequence includes equipment occupancy status, inventory deduction status, process flow status, quality release status, and shift allocation status. The reverse compensation mirror is the compensation state sequence of each link node when the commitment execution unit is interrupted or fails to execute. The compensation state sequence includes equipment release status, inventory replenishment status, process rollback status, quality re-inspection status, and shift rescheduling status.

[0023] It should be noted that this invention does not merely predict what will happen after execution, but simultaneously constructs a compensatory state sequence of how to recover after execution failure, so that the production management system has both positive state evolution basis and reverse recovery evolution basis before output; the dual mirror structure is not a simple repetition of the existing unidirectional execution deduction, but establishes a coupling relationship between positive fulfillment analysis and reverse compensation analysis on the same commitment penetration chain.

[0024] As a further aspect of the present invention, in step S5, the counterfactual scenarios include normal execution scenarios, equipment de-rated scenarios, material shortage scenarios, quality re-inspection scenarios, and shift change scenarios; counterfactual deduction is performed on the forward execution mirror to calculate equipment load drift, inventory occupancy drift, process waiting drift, quality inspection insertion, and delivery node drift; counterfactual deduction is performed on the reverse compensation mirror to calculate equipment release time, inventory replenishment time, process rollback time, quality re-inspection time, and shift rescheduling time; wherein, each drift is determined by the difference between the deduced value under the counterfactual scenario and the benchmark value under the benchmark execution scenario.

[0025] It should be noted that by performing counterfactual deductions on the forward execution mirror and the reverse compensation mirror respectively, this invention places both execution disturbance analysis and compensation cost analysis in the pre-output screening stage. This enables the production management system to not only obtain how much a candidate execution path will deviate under different abnormal conditions, but also to obtain the time and resource costs required to restore the deviation to a controllable state. This effect cannot be naturally obtained by conventional retrieval, conventional sorting, or conventional one-way scheduling simulation.

[0026] As a further aspect of the present invention, in step S5, the rollback window parameters include the maximum rollback duration, the set of recoverable equipment, the set of recoverable materials, the rollback cutoff node, and the upper limit of secondary disturbances; the maximum rollback duration is determined based on the equipment release duration, inventory replenishment duration, process rollback duration, quality re-inspection duration, and shift rescheduling duration; the rollback cutoff node is determined based on the delivery constraint node in the commitment penetration chain; the upper limit of secondary disturbances is determined based on the threshold for the allowed delay duration of non-target orders, the threshold for the allowed additional load of non-target equipment, and the threshold for the allowed occupancy of non-target inventory.

[0027] It should be noted that this invention moves the recovery problem, which was originally only passively handled after an anomaly occurred, to the pre-output judgment stage by using the fallback window parameter. This allows the production management system to filter out candidate paths that are short-term executable but would exceed the fallback boundary if they fail, before generating question and answer text and executing instructions. Unlike the anomaly retries, anomaly alarms, or anomaly remedies in the prior art, this invention provides boundary constraints on the feasibility of fallback before the result is output.

[0028] As a further aspect of the present invention, in step S5, the execution constraints include equipment capacity threshold, inventory availability threshold, shift capacity threshold, quality release conditions, and delivery time limit conditions; when determining the target execution path, only candidate execution paths that meet the following conditions are retained:

[0029] The equipment load drift, inventory occupancy drift, process waiting drift, and delivery node drift during forward mirroring do not exceed the preset thresholds;

[0030] The maximum rollback duration, rollback cutoff node, and upper limit of secondary disturbance of the reverse compensation mirror meet the preset conditions;

[0031] When multiple candidate execution paths meet the conditions, the target execution path is determined in order of priority: delivery node drift amount, equipment load drift amount, and maximum rollback duration.

[0032] It should be noted that the screening of candidate execution paths in this invention no longer relies on a single correlation score or a single execution cost, but is simultaneously subject to the dual constraints of forward execution perturbation and reverse backoff boundary, so that the final retained target execution path satisfies both the perturbation constraint under normal execution conditions and the recovery constraint under abnormal conditions; this dual constraint screening mechanism makes the question-and-answer results and execution instructions obtained by this invention have execution consistency and abnormal recoverability.

[0033] As a further aspect of the present invention, in step S6, both the question-and-answer text and the production execution instruction are generated from the target execution path; the question-and-answer text includes the commitment subject identifier, the commitment object identifier, the action type, the execution time window, and the constraint conditions; the production execution instruction includes the execution terminal identifier, the execution node sequence, the resource allocation sequence, the verification conditions, and the rollback trigger conditions; the returned status parameters include the equipment status returned parameters, the inventory status returned parameters, the quality status returned parameters, and the order progress returned parameters; by overwriting the execution node status of the returned status parameters, switching the constraint edge on and off, recalibrating the edge weights, and truncating and reconnecting the failed links, the association relationship between the order node, equipment node, inventory node, quality node, and shift node in the production object association topology is corrected, and the execution time window, resource dependency sequence, verification condition sequence, and rollback trigger conditions of the commitment execution unit are updated based on the corrected association relationship.

[0034] It should be noted that since the question-and-answer text and the production execution instructions are generated from the same target execution path, this invention avoids the separation problem in the prior art where the answer is one set of logic and the execution is another set of logic. At the same time, by overwriting the node state after the execution terminal returns, switching the opening and closing of constraint edges, recalibrating edge weights, and truncating and reconnecting failed links, this invention updates not a single text result, but the execution link itself and its committed objects, so that subsequent question-and-answer and decision-making can be based on the corrected production object association topology. The closed-loop update function is not a normal feedback recalculation, but a structural layer correction of the production execution link.

[0035] Compared with the prior art, the present invention has at least the following technical effects:

[0036] Firstly, this invention unifies the question-and-answer results and suggestion results into a unified object through a commitment execution unit, transforming the output object from an information result into an execution object that propagates and is verified along the production chain;

[0037] Secondly, this invention transforms the discrete relationships between orders, processes, equipment, inventory, quality, and shifts into a continuous execution chain by committing to a penetration chain, thus changing the object of analysis for production problems from local hit results to cross-chain propagation paths;

[0038] Third, this invention enables simultaneous analysis of execution fulfillment conditions and compensation conditions after execution failure by simultaneously constructing a forward execution mirror and a reverse compensation mirror on the same commitment penetration chain.

[0039] Fourth, this invention enables the target execution path to simultaneously satisfy both normal execution constraints and abnormal rollback constraints by jointly filtering path perturbation parameters and rollback window parameters.

[0040] Fifth, this invention ensures that the question-and-answer text and execution instructions are generated synchronously from the same target execution path, so that the question-and-answer output and the execution output are consistent at the source.

[0041] Sixth, the present invention enables the production object association topology and commitment execution unit to be continuously updated as the field execution status changes by triggering node state overwriting, constraint edge opening and closing switching, edge weight recalibration, and failed link truncation and reconnection through the return of status parameters. Attached Figure Description

[0042] Figure 1 This is a technical roadmap of the method of the present invention. Detailed Implementation

[0043] 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.

[0044] Example 1

[0045] like Figure 1 As shown, this embodiment provides an intelligent question-answering and decision-making support method for a production management system. The method is deployed in the production management system, which is connected to the order management module, process equipment management module, inventory supply management module, quality inspection management module, shift operation management module, and execution terminal. The execution terminal includes at least one of the following: production scheduling terminal, equipment control terminal, warehousing terminal, and quality terminal.

[0046] In this embodiment, the production problem instruction input by the user terminal is: "Can order A be delivered 8 hours ahead of schedule under the current production conditions?" After receiving the production problem instruction, the production management system first performs field parsing on the problem text to extract the order identifier A, the target time window T, the action type identification field, and the constraint prompt field; among them, the action type is identified as progress query, and the target time window is to move the delivery node forward by 8 hours.

[0047] Subsequently, the production management system collects status parameters from each business unit: the order execution unit returns the order number, process number, planned start time, planned end time, and current completion rate corresponding to order A; the process equipment unit returns the equipment number, equipment process type, current load rate, and available time period corresponding to each process of order A; the inventory supply unit returns the material number, available quantity, locked quantity, and replenishment time corresponding to the materials required for order A; the quality inspection unit returns the current inspection batch number, inspection status, and re-inspection duration; and the shift operation unit returns the shift number, number of workers, shift start time, and shift end time.

[0048] After collecting the above status parameters, the production management system establishes a production object association topology. Specifically, it establishes order nodes based on order numbers, process nodes based on combinations of order numbers and process numbers, equipment nodes based on equipment numbers, inventory nodes based on material numbers, quality nodes based on inspection batch numbers or process inspection point identifiers, and shift nodes based on shift numbers. Then, it establishes order dependency edges based on the order process route table, process sequence edges based on the process sequence table, equipment occupancy edges based on the equipment allocation table, inventory supply edges based on the material demand table and inventory supply table, quality inspection edges based on the quality inspection process table, and shift allocation edges based on the shift scheduling table. Each edge records the relationship type, start node, end node, start available time, end available time, capacity threshold, and rollback flag.

[0049] In this embodiment, the production management system extracts the order identifier A and target time window T from the production problem instruction, locates the order node, process node, equipment node, inventory node, quality node, and shift node corresponding to order A in the production object association topology, and encapsulates the located target production object node with action type, execution time window, dependent equipment set, dependent material set, verification condition set, and rollback identifier to form a commitment execution unit. Specifically, the commitment subject identifier is the scheduling role identifier of the current user account, the commitment object identifier is order A, the action type is progress query, the execution time window is from the current time to 8 hours before the target delivery time, the dependent equipment set includes the candidate equipment set corresponding to all processes of order A, the dependent material set includes the raw materials, auxiliary materials, and spare parts set required for all processes of order A, and the verification condition set includes equipment capacity threshold, inventory availability threshold, quality release condition, and delivery time limit condition.

[0050] After forming a commitment execution unit, the production management system starts from the starting node corresponding to the commitment execution unit and expands along the order constraint chain, process connection chain, equipment occupancy chain, inventory supply chain, and shift capacity chain to generate a commitment penetration chain. Specifically, it first expands from the order node to all uncompleted process nodes, then maps the corresponding equipment nodes from each process node, maps the currently available shift nodes from the equipment nodes, and then maps the required inventory nodes from both the process nodes and equipment nodes, and inserts the quality nodes corresponding to each process node. For each link node, the node entry time, node exit time, and corresponding constraint relationship are recorded. The node entry time is determined comprehensively based on the upstream node exit time, equipment availability period, material replenishment time, and shift start time, while the node exit time is determined comprehensively based on the process duration, equipment processing time, and quality inspection time. If the same process node corresponds to multiple equipment nodes, multiple shift nodes, or multiple inventory supply nodes, different candidate execution paths are formed respectively.

[0051] Subsequently, the production management system constructs a forward execution mirror and a reverse compensation mirror based on the commitment penetration chain. The forward execution mirror is used to characterize the target state sequence of each link node when the commitment execution unit executes according to the commitment penetration chain. For example, the target state of the equipment node is "occupancy status + occupation duration + target load rate", the target state of the inventory node is "deduction status + remaining quantity after deduction", the target state of the process node is the flow status of "pending execution - in execution - completed", the target state of the quality node is "pending inspection - release", and the target state of the shift node is "personnel occupation + time period occupation". The reverse compensation mirror is used to characterize the compensation state sequence of each link node when the commitment execution unit is interrupted or fails to execute. For example, the compensation state of the equipment node is "release status", the compensation state of the inventory node is "replenishment status", the compensation state of the process node is "return status", the compensation state of the quality node is "re-inspection status", and the compensation state of the shift node is "rearrangement status".

[0052] In this embodiment, the production management system presets normal execution scenarios, equipment derating scenarios, material shortage scenarios, quality re-inspection scenarios, and shift change scenarios. First, under the normal execution scenario, the baseline equipment occupancy time, baseline inventory occupancy, baseline process waiting time, and baseline delivery completion time are calculated. Then, under each of the other counterfactual scenarios, the above parameters are recalculated, and the equipment load drift, inventory occupancy drift, process waiting drift, quality inspection insertion, and delivery node drift are obtained by the difference between the extrapolated values ​​and the baseline values. For the reverse compensation mirror, the equipment release time, inventory replenishment time, process rollback time, quality re-inspection time, and shift rescheduling time are calculated respectively, and a rollback window parameter consisting of the maximum rollback time, recoverable equipment set, recoverable material set, rollback cutoff node, and secondary disturbance upper limit value is formed accordingly.

[0053] During the path selection phase, the production management system retains only candidate execution paths that simultaneously meet the following conditions: First, the equipment load drift, inventory occupancy drift, process waiting drift, and delivery node drift corresponding to the forward execution mirror do not exceed the corresponding preset thresholds; Second, the maximum rollback duration, rollback cutoff node, and secondary disturbance upper limit corresponding to the reverse compensation mirror meet the preset conditions; If there are multiple candidate execution paths that meet the conditions, the target execution path is determined according to the priority order of delivery node drift, equipment load drift, and maximum rollback duration.

[0054] After determining the target execution path, the production management system synchronously generates a question-and-answer text and a production execution instruction based on the target execution path. The question-and-answer text includes the commitment subject identifier, the commitment object identifier, the action type, the execution time window, and the constraints, such as "Order A can be completed 4 hours ahead of schedule under the current equipment, inventory, shift, and quality constraints, but cannot meet the condition of delivery 8 hours ahead of schedule". The production execution instruction includes the execution terminal identifier, the execution node sequence, the resource allocation sequence, the verification conditions, and the rollback trigger conditions, and is sent to the corresponding execution terminal.

[0055] After execution by the terminal, the device status, inventory status, quality status, and order progress parameters are returned to the production management system. Upon receiving the returned status parameters, the production management system performs status overwriting, constraint edge opening / closing switching, edge weight recalibration, and failed link truncation and reconnection on the production object's associated topology. Based on this, it updates the execution time window, dependent device set, dependent material set, verification condition set, and rollback trigger condition in the commitment execution unit for subsequent processing of new production problem instructions.

[0056] In this embodiment, by objectifying the commitment execution unit, expanding the commitment penetration chain, deducing the forward execution mirror and the reverse compensation mirror, jointly filtering the path disturbance parameters and the backoff window parameters, and updating the execution terminal feedback loop, the unified constraint of the question-and-answer results and execution instructions on the same target execution path is realized, so that the output results not only have semantic answering capabilities, but also have execution fulfillment basis, anomaly compensation basis, and dynamic correction basis.

[0057] Example 2

[0058] The difference between Embodiment 2 and Embodiment 1 is that this embodiment illustrates how to construct a forward execution image and a reverse compensation image based on the same commitment penetration chain under different counterfactual scenarios, and form path perturbation parameters and backoff window parameters.

[0059] In this embodiment, taking the commitment execution unit corresponding to order B as an example, the action type of the commitment execution unit is production scheduling adjustment, and the execution time window is between the start time of the current shift and the time of the next delivery node. The production management system generates a commitment penetration chain based on order B, which sequentially passes through order node B1, process nodes B2 and B3, equipment nodes E1 and E2, inventory nodes M1 and M2, quality node Q1, and shift nodes S1 and S2.

[0060] (1) Normal execution scenario

[0061] Normal execution scenario refers to the baseline scenario in which the commitment execution unit executes along the commitment penetration chain when the order execution, process connection, equipment occupancy, inventory supply, quality inspection and shift arrangement corresponding to the current production plan are all operating according to the predetermined conditions, and there are no abnormal interference factors such as equipment capacity reduction, material shortage, re-inspection insertion, shift adjustment or other abnormal interference factors.

[0062] In a normal execution scenario, equipment nodes E1 and E2 are both at their rated capacity, the available quantities of inventory nodes M1 and M2 meet the requirements of their respective processes, quality node Q1 is in a direct release state, and shift nodes S1 and S2 both meet the required number of workers. The production management system generates a forward execution mirror under the normal execution scenario, resulting in the following target state sequence: equipment E1 is occupied in the first time period, equipment E2 is occupied in the second time period, materials M1 and M2 are deducted before entering their respective processes, process nodes B2 and B3 complete their flow according to the process sequence, quality node Q1 is released directly after the second process, and shift nodes S1 and S2 undertake their respective work tasks. Simultaneously, a reverse compensation mirror is generated, resulting in the following compensation state sequence: if execution is interrupted, equipment E1 and E2 are released sequentially, materials M1 and M2 are replenished sequentially, process nodes B2 and B3 revert to the previous recoverable state, quality node Q1 enters the re-inspection preparation state, and shift nodes S1 and S2 return to their original scheduling state. The normal execution scenario serves as the baseline execution scenario for calculating the drift amount under subsequent counterfactual scenarios.

[0063] (2) Equipment derating scenarios

[0064] Equipment derating scenarios refer to situations where one or more equipment nodes in the committed penetration chain, while still operational, have actual processing capacity, available load capacity, or processing capacity per unit time lower than the rated capacity in the corresponding baseline execution scenario, resulting in changes to the processing time, waiting time, or delivery time of related process nodes. The core focus of equipment derating scenarios is the equipment nodes and their corresponding equipment occupancy edges, highlighting the propagation impact of reduced equipment capacity on the process connection chain, shift capacity chain, and delivery nodes. The triggering condition for an equipment derating scenario is: the actual available processing capacity of any equipment node in the committed penetration chain is lower than the rated processing capacity recorded in the equipment ledger, and the derating amount reaches a preset derating threshold; the preset derating threshold can be set to any fixed value among 10%, 15%, or 20% of the rated capacity reduction.

[0065] In the equipment derating scenario, the available processing capacity of equipment node E1 decreases to 80% of its rated capacity, while its available time remains unchanged. The production management system recalculates the occupancy time of equipment E1 and the exit time of the corresponding process node B2 in the forward execution mirror. It finds that the exit time of process B2 is delayed, which causes the entry time of process B3 to be delayed synchronously, and further causes the delivery node completion time to be shifted. The equipment load drift, process waiting drift, and delivery node drift are obtained by subtracting the extrapolated value of the current scenario from the baseline value of the normal execution scenario.

[0066] Meanwhile, in the reverse compensation mirror, if device E1 is interrupted during derated operation, it is necessary to release device E1, relocate the replacement device, and rearrange the corresponding shift, forming the device release time, process rollback time, and shift rearrangement time. If the device load drift exceeds the device capacity threshold, or the maximum rollback time formed by the device release time and process rollback time exceeds the allowed rollback time, the candidate execution path corresponding to device E1 is eliminated.

[0067] (3) Material shortage scenario

[0068] A material shortage scenario refers to a situation where the available quantity of one or more inventory nodes in the commitment-through chain is lower than the required quantity of materials for the corresponding process node within the target execution time window, or where the current demand is met but the demand for subsequent processes cannot be covered, resulting in process waiting, replenishment insertion, or supply path switching. The triggering condition for a material shortage scenario is: the available quantity of any inventory node in the commitment-through chain is lower than the required quantity for the corresponding process node within the target execution time window, or the inventory node meets the current process demand but cannot cover the demand for subsequent processes, and the quantity gap reaches a preset shortage threshold.

[0069] In a material shortage scenario, the available quantity of inventory node M1 is insufficient to meet all material requirements corresponding to process node B2, only satisfying the first stage of demand, with the remainder awaiting replenishment. The production management system recalculates the inventory deduction status and process waiting time in the forward execution mirror to obtain the inventory occupancy drift, process waiting drift, and delivery node drift. Simultaneously, in the reverse compensation mirror, if execution is interrupted due to material shortage, the occupied materials need to be replenished, forming an inventory replenishment time, and the upper limit of secondary disturbance is determined based on the remaining available material quantity of non-target orders after replenishment. If the inventory occupancy drift exceeds the available inventory threshold, or the upper limit of secondary disturbance exceeds the preset boundary, the corresponding candidate execution path is not retained.

[0070] (4) Quality re-inspection scenario

[0071] The quality re-inspection scenario refers to a situation where quality nodes in the commitment-through-chain no longer operate according to the direct release process in the baseline execution scenario. Instead, a re-inspection process is inserted after the target process is completed, an additional inspection process is added, or an anomaly judgment is triggered, thereby changing the flow status, waiting time, or delivery node of subsequent processes. The triggering conditions for the quality re-inspection scenario are: the quality inspection rule corresponding to the quality node triggers an additional re-inspection process, or the test result reaches the preset re-inspection judgment threshold, causing the original direct release path to become a re-inspection path.

[0072] In the quality re-inspection scenario, quality node Q1 changes from direct release to adding a re-inspection process; in the forward execution mirror, the production management system changes the target status of quality node Q1 from "release status" to "pending re-inspection - re-inspection completed - release status", and recalculates the quality inspection insertion amount and delivery node drift amount; in the reverse compensation mirror, if the re-inspection result is abnormal, process node B3 needs to be rolled back to the previous recoverable node, forming process rollback time and quality re-inspection time; if the re-inspection insertion amount causes the delivery node drift amount to exceed the delivery time limit, or the quality re-inspection time causes the maximum rollback time to exceed the allowable rollback boundary, the candidate execution path is eliminated.

[0073] (5) Shift change scenario

[0074] A shift change scenario refers to a situation where one or more shift nodes in the commitment pass-through chain experience changes in shift start and end times, number of workers, cancellation of available shifts, or shift replacements within the target execution time window, resulting in changes to the executable time periods of associated equipment nodes, process nodes, or resource allocation nodes. The triggering conditions for a shift change scenario are: changes in the number of workers, shift start time, shift end time, number of shifts, or shift availability status of a shift node, and these changes cause a change in the executable time period of at least one equipment node or process node in the commitment pass-through chain.

[0075] In shift change scenarios, the number of workers at shift node S2 decreases or the shift duration is shortened. The production management system recalculates the capacity status of shift node S2, the entry and exit times of process B3, and the completion time of the delivery node in the forward execution mirror, forming the process waiting drift amount and the delivery node drift amount. In the reverse compensation mirror, if shift node S2 is insufficient to support the predetermined execution path, some tasks need to be migrated to the new shift node, forming the shift rescheduling time. If the shift rescheduling time causes the rollback deadline node to be moved after the delivery constraint node, the candidate path is eliminated.

[0076] In this embodiment, after the production management system performs simulations for different counterfactual scenarios, it writes the path disturbance parameters and backoff window parameters corresponding to each scenario into the candidate execution path evaluation table, and then selects candidate execution paths that simultaneously meet the forward execution constraints and reverse backoff constraints based on preset thresholds.

[0077] This embodiment illustrates that the present invention does not only perform unidirectional reasoning for the normal execution state, but also obtains execution offset information and recovery boundary information simultaneously through dual-mirror deduction under different counterfactual scenarios, thereby providing a dual constraint basis for target execution path selection.

[0078] Example 3

[0079] Unlike Examples 1 and 2, this embodiment differs in the field emphasis of the commitment execution unit, the expansion emphasis of the commitment penetration chain, and the final output question-and-answer text and execution instructions under different action types.

[0080] In this embodiment, the preset action types include progress query, production scheduling adjustment, material allocation, equipment switching, and abnormal handling.

[0081] (1) Progress query action type

[0082] When a user terminal inputs "Can order C be completed before the end of this shift?", the production management system identifies the action type as a progress query. At this time, the commitment execution unit focuses on the commitment object identifier, execution time window, delivery node, and verification condition set. The dependent equipment set and dependent material set adopt the resource set already allocated in the current production plan. The commitment penetration chain generated by the production management system focuses on expanding the order constraint chain, process connection chain, and equipment occupancy chain to determine whether the existing execution path can meet the target time window. The forward execution mirror focuses on analyzing the target completion status of each process node and equipment node, while the reverse compensation mirror focuses on analyzing whether there is a rollback path if a process is interrupted. The final generated question and answer text is the progress judgment result. If there is a target execution path that meets the conditions, a confirmation execution instruction to maintain the current execution order is generated at the same time.

[0083] (2) Types of production scheduling adjustment actions

[0084] When a user terminal inputs "Can order D be adjusted to a higher priority queue?", the production management system identifies the action type as production scheduling adjustment. At this time, the commitment execution unit, in addition to the commitment object identifier and execution time window, also focuses on multiple candidate shift nodes, candidate equipment nodes, and priority adjustment conditions. The commitment penetration chain focuses on expanding the process connection chain, equipment occupancy chain, and shift capacity chain to analyze the resource rearrangement after adjustment. The forward execution mirror analysis analyzes the target status of each process, equipment, and shift after the production scheduling adjustment, and the reverse compensation mirror analysis analyzes how to restore the original production schedule if the adjustment fails. The final generated question and answer text explains whether the adjustment conditions are met, and the production execution instruction includes the new execution node sequence, shift allocation sequence, and rollback trigger conditions.

[0085] (3) Types of material transfer actions

[0086] When a user terminal inputs "Can material M required for order E be transferred from warehouse X to warehouse Y?", the production management system identifies the action type as material transfer. The commitment execution unit focuses on the dependent material set, inventory node set, supply path, and inventory verification conditions. The commitment penetration chain focuses on expanding the inventory supply chain, order constraint chain, and process connection chain to analyze the material accessibility after transfer and its impact on subsequent processes. The forward execution mirror focuses on analyzing the inventory deduction status after transfer and the continuous execution status of the target order. The reverse compensation mirror focuses on analyzing how to perform replenishment of occupied materials if the transfer is interrupted and the impact of replenishment on non-target orders. The final generated question and answer text explains whether the transfer conditions are met, and the production execution instruction includes the target warehouse terminal identifier, transfer resource sequence, and rollback trigger conditions.

[0087] (4) Equipment switching action type

[0088] When a user terminal inputs "Can process F be switched from equipment G to equipment H for execution?", the production management system identifies the action type as equipment switching. The commitment execution unit focuses on the dependent equipment set, candidate switching equipment, equipment capacity verification conditions, and execution time window. The commitment penetration chain focuses on expanding the equipment occupancy chain, process connection chain, and shift capacity chain to analyze the equipment accessibility and shift capacity after switching. The forward execution mirror focuses on analyzing the occupancy status of equipment H after switching and the timing status of subsequent process nodes, while the reverse compensation mirror focuses on analyzing the recovery path of returning to the original equipment G or switching to the backup equipment when switching fails. The final generated question and answer text explains whether the equipment switching conditions are met, and the production execution instruction includes the new equipment terminal identifier, switching order, and verification conditions.

[0089] (5) Types of abnormal handling actions

[0090] When a user terminal inputs "How should order J continue to be executed after equipment I fails?", the production management system identifies the action type as an anomaly handling. The commitment execution unit focuses on rollback identifiers, rollback trigger conditions, alternative resource sets, and anomaly handling rule sets. The commitment penetration chain simultaneously expands the upstream and downstream process nodes, equipment nodes, inventory nodes, and shift nodes affected by the failure to analyze the scope of anomaly propagation. The forward execution mirror focuses on analyzing the target state sequence under the alternative execution plan, while the reverse compensation mirror focuses on analyzing the secondary compensation path if the alternative plan still fails. The final generated question and answer text explains the recommended handling path, and the production execution instruction includes the alternative execution node sequence, resource reallocation sequence, and anomaly escalation trigger conditions.

[0091] In this embodiment, although different action types share the production object association topology, commitment execution unit, commitment penetration chain, and double mirror structure, the commitment execution unit field focus, commitment penetration chain expansion focus, mirror attention state, and execution instruction content are different for each action type. This shows that the commitment execution unit in this invention is not an abstract description, but a unified technical object that can form different execution structures and different deduction paths for different production problems.

[0092] Example 4

[0093] Unlike Embodiments 1, 2, and 3, this embodiment describes how to generate multiple candidate execution paths for the same commitment execution unit, how to retain only the candidate execution paths that meet the conditions, and how to dynamically modify the production object association topology and commitment execution unit after the execution terminal returns the data.

[0094] In this embodiment, the production problem instruction input by the user terminal is "Which execution path should be used to complete the delivery of this batch of order K under the current equipment and inventory conditions?" After parsing the production problem instruction, the production management system forms a commitment execution unit and locates the process node, equipment node, inventory node and shift node corresponding to order K in the production object association topology. Since a process node can correspond to multiple equipment nodes, the same equipment node can correspond to multiple shift nodes, and the same material requirement can be supplied by multiple inventory nodes, the production management system generates multiple candidate execution paths, denoted as paths P1, P2, P3 and P4.

[0095] Among them, path P1 corresponds to equipment group E1+E3, inventory supply group M1+M2, and shift group S1; path P2 corresponds to equipment group E2+E3, inventory supply group M1+M3, and shift group S1+S2; path P3 corresponds to equipment group E4+E5, inventory supply group M2+M3, and shift group S2; and path P4 corresponds to equipment group E2+E4, inventory supply group M1+M2, and shift group S3. For each candidate execution path, the production management system generates a forward execution image and a reverse compensation image, and performs counterfactual deduction under normal execution scenarios, equipment de-rated scenarios, material shortage scenarios, quality re-inspection scenarios, and shift change scenarios to obtain the corresponding equipment load drift, inventory occupancy drift, process waiting drift, quality inspection insertion, delivery node drift, as well as equipment release time, inventory replenishment time, process rollback time, quality re-inspection time, and shift rescheduling time.

[0096] In the first-level screening, the production management system eliminates paths that do not meet the execution constraints based on the forward execution mirror parameters. Specifically, when the equipment load drift corresponding to path P1 exceeds the equipment capacity threshold, path P1 is eliminated; when the delivery node drift corresponding to path P3 exceeds the delivery time limit condition, path P3 is eliminated.

[0097] In the second layer of screening, the production management system eliminates paths that do not meet the rollback boundary conditions based on the reverse compensation mirror parameters; specifically, when the maximum rollback duration corresponding to path P2 exceeds the allowed rollback duration, or when the rollback deadline node of path P2 is later than the delivery constraint node, path P2 is eliminated.

[0098] After completing two layers of screening, only path P4 satisfies the forward execution disturbance constraint and the reverse backoff boundary constraint. Therefore, path P4 is determined as the target execution path. If multiple remaining paths simultaneously meet the conditions, they are sorted according to the priority order of delivery node drift amount, equipment load drift amount, and maximum backoff time. The path with the best sorting result is selected as the target execution path.

[0099] After determining the target execution path P4, the production management system synchronously generates a question-and-answer text and a production execution instruction based on path P4. The question-and-answer text includes the commitment subject identifier, the commitment object identifier, the action type, the execution time window, and the constraints, such as "Order K can be delivered under the current conditions according to path P4, and has the ability to roll back after an anomaly is triggered". The production execution instruction includes the execution terminal identifier, the execution node sequence, the resource allocation sequence, the verification conditions, and the rollback trigger conditions, and is issued to the production scheduling terminal, the equipment control terminal, and the warehouse terminal, respectively.

[0100] During execution, the execution terminals return status parameters; for example, the equipment control terminal returns "Equipment E2 actual load decreased by 10%", the warehousing terminal returns "Material M1 arrival delayed by 30 minutes", the quality terminal returns "Quality node Q2 enters re-inspection status", and the order management module returns "Process node K3 actual completion time delayed by 20 minutes compared to the plan"; the production management system performs the following update operations based on the above returned status parameters:

[0101] First, overwrite the node status of equipment node E2, inventory node M1, quality node Q2 and process node K3;

[0102] Second, implement constraint switching for equipment occupation edge, inventory supply edge and quality inspection edge, such as marking equipment occupation edge that has lost its feasibility as closed.

[0103] Third, perform edge weight recalibration on the edges that remain available. For example, recalibrate the priority of equipment occupancy edges based on changes in equipment load, and recalibrate the availability time of inventory supply edges based on delivery delays.

[0104] Fourth, when a local link in the original target execution path no longer meets the constraints, the current local link is truncated, and a new alternative link is searched in the production object's associated topology for reconnection.

[0105] After receiving the status parameters from the execution terminal, the production management system triggers different update actions based on different feedback events:

[0106] (1) Triggering rules for node state overwriting

[0107] When the returned status parameters indicate that the current state of a node is inconsistent with the state recorded in the topology associated with the production object, the node state overwriting is triggered.

[0108] For example, when the equipment status feedback parameters indicate that the equipment changes from "available" to "operating at reduced capacity" or "downtime due to malfunction", the status of the corresponding equipment node is overwritten; when the inventory status feedback parameters indicate that the inventory availability changes, the status of the corresponding inventory node is overwritten; when the quality status feedback parameters indicate that the quality node changes from "pending inspection" to "under re-inspection", the status of the corresponding quality node is overwritten; when the order progress feedback parameters indicate that the completion time of the process node changes, the status of the corresponding order node or process node is overwritten.

[0109] (2) Triggering rules for switching on and off of constraint edges

[0110] When the returned status parameters indicate that a certain associated edge can no longer be used as a valid execution relationship at the current moment, the constraint edge is triggered to close; when the returned status parameters indicate that the originally closed edge has regained its validity, the constraint edge is triggered to open.

[0111] For example, when equipment failure causes the equipment occupancy edge to become invalid, the equipment occupancy edge is switched to the closed state; when the inventory supply edge becomes valid again after material replenishment is completed, the current inventory supply edge is switched to the open state.

[0112] (3) Triggering rules for edge weight remarking

[0113] When the returned status parameters indicate that a certain associated edge has not failed, but its capacity, time, or priority attributes have changed, the edge weight recalibration is triggered.

[0114] For example, when equipment de-rated results in a decrease in processing capacity but still allows operation, the capacity and priority attributes of the equipment occupancy edge are recalibrated; when material arrival delays cause changes in replenishment time, the time attribute of the inventory supply edge is recalibrated; when the number of workers per shift decreases, the capacity attribute of the shift allocation edge is recalibrated.

[0115] (4) Triggering rules for reconnection after failure link is cut off

[0116] When the returned status parameters indicate that any critical link node or critical associated edge in the target execution path can no longer meet the execution requirements corresponding to the committed execution unit, and cannot be restored even after overwriting the node status, switching the constraint edge on / off, or recalibrating the edge weight, the failed link is truncated and reconnected.

[0117] Specifically, the production management system cuts off the subsequent links from the failed node to the end of the target execution path, and re-searches for alternative nodes and alternative edges that satisfy the current execution constraints and backoff boundaries in the production object association topology to form a new local alternative link, which is then connected to the valid link segment before the failed path.

[0118] After completing the topology correction, the production management system updates the original commitment execution units, specifically including: updating the execution time window, replacing the dependent equipment set and dependent material set, updating the verification condition set, and modifying the rollback trigger condition; the updated commitment execution units continue to serve as input objects for subsequent production problem handling and subsequent execution path selection.

[0119] in:

[0120] Update the execution time window when the entry or exit time of a node in the target execution path changes;

[0121] When the original dependent device node or dependent material node fails and is replaced, update the dependent device set and dependent material set.

[0122] When the threshold boundary, verification conditions, or release conditions change, update the verification condition set;

[0123] Update the rollback trigger conditions when the rollback cutoff node, allowed rollback duration, or secondary disturbance boundary changes.

[0124] This embodiment illustrates that the present invention does not merely select an optimal suggested path before output, but rather retains candidate execution paths that simultaneously satisfy both forward execution constraints and reverse recovery constraints through a two-layer path filtering process. After execution, it performs structural layer correction on the production object association topology and commitment execution units through node state overwriting, constraint edge opening and closing switching, edge weight recalibration, and failed link truncation and reconnection, forming a continuous closed-loop execution link update mechanism.

[0125] 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.

[0126] In conclusion, 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. An intelligent question-answering and decision support method for production management systems, characterized in that, include: Step S1: Receive production problem instructions sent by the user terminal and extract the identifiers and time windows of orders, processes, equipment, and materials; Step S2: Collect status parameters from the order execution unit, process equipment unit, inventory supply unit, quality inspection unit, and shift operation unit, and establish a production object association topology according to the relationship of order subordinate, process sequence, equipment occupation, inventory supply, quality inspection, and shift allocation. Step S3: Based on the order identifier, process identifier, equipment identifier, material identifier, and time window, match the target production object node in the production object association topology, and encapsulate the matching result with the action type, execution time window, dependent resources, and verification conditions into a commitment execution unit; Step S4: Starting from the starting node of the commitment execution unit, proceed chain by chain along the order constraint chain, process connection chain, equipment occupancy chain, inventory supply chain and shift capacity chain to generate the commitment penetration chain; Step S5: Generate a forward execution image and a reverse compensation image based on the commitment penetration chain; calculate path perturbation parameters and backoff window parameters for the preset counterfactual scenario; and select the target execution path based on the execution constraints. Step S6: Generate question and answer text and production execution instructions based on the target execution path, send them to the execution terminal, and receive the returned status parameters to update the production object association topology and commitment execution unit.

2. The intelligent question-answering and decision-making support method for production management systems according to claim 1, characterized in that, In step S2, the status parameters of the order execution unit include order number, process number, planned start time, planned end time, and current completion rate; the status parameters of the process equipment unit include equipment number, equipment process type, current load rate, and available time period; the status parameters of the inventory supply unit include material number, available quantity, locked quantity, and replenishment time. The status parameters of the quality inspection unit include the inspection batch number, inspection status, and re-inspection duration; the status parameters of the shift operation unit include the shift number, number of workers, shift start time, and shift end time.

3. The intelligent question-answering and decision support method for production management systems according to claim 1, characterized in that, In step S2, the production object association topology includes order nodes, process nodes, equipment nodes, inventory nodes, quality nodes, and shift nodes; Order hierarchy is established based on the order process route table; process sequence is established based on the process order table; equipment occupancy is established based on the equipment allocation table; inventory supply is established based on the material requirements table and inventory supply table; quality inspection is established based on the quality inspection process table; and shift allocation is established based on the shift scheduling table.

4. The intelligent question-answering and decision support method for production management systems according to claim 1, characterized in that, In step S3, the commitment execution unit includes a commitment subject identifier, a commitment object identifier, an action type, an execution time window, a set of dependent equipment, a set of dependent materials, a set of verification conditions, and a rollback identifier; the action type is one of progress query, production scheduling adjustment, material transfer, equipment switching, or abnormal handling; the commitment execution unit is encapsulated by the target production object node, action type, execution time window, set of dependent equipment, set of dependent materials, and set of verification conditions.

5. The intelligent question-answering and decision support method for production management systems according to claim 1, characterized in that, In step S4, when generating the commitment penetration chain, the chain is expanded one by one in the order of order constraint chain, process connection chain, equipment occupancy chain, inventory supply chain and shift capacity chain; each commitment penetration chain records the path start node, path end node, link node sequence, node entry time, node exit time and constraint relationship; the node entry time and node exit time are determined according to the planned time, equipment available time period, process duration, inventory replenishment time and shift start and end time.

6. The intelligent question-answering and decision support method for production management systems according to claim 1, characterized in that, In step S5, the forward execution mirror is the target state sequence of each link node when the commitment execution unit executes according to the commitment penetration chain. The target state sequence includes equipment occupancy status, inventory deduction status, process flow status, quality release status, and shift allocation status. The reverse compensation mirror is the compensation state sequence of each link node when the commitment execution unit is interrupted or fails to execute. The compensation state sequence includes equipment release status, inventory replenishment status, process rollback status, quality re-inspection status, and shift rescheduling status.

7. The intelligent question-answering and decision support method for production management systems according to claim 6, characterized in that, In step S5, the counterfactual scenarios include normal execution scenario, equipment depreciation scenario, material shortage scenario, quality re-inspection scenario, and shift change scenario; Counterfactual analysis is performed on the forward execution mirror to calculate equipment load drift, inventory occupancy drift, process waiting drift, quality inspection insertion, and delivery node drift. Counterfactual inference is performed on the reverse compensation mirror to calculate equipment release time, inventory replenishment time, process rollback time, quality re-inspection time, and shift rescheduling time; among which, each drift amount is determined by the difference between the inferred value under the counterfactual scenario and the benchmark value under the benchmark execution scenario.

8. The intelligent question-answering and decision support method for production management systems according to claim 7, characterized in that, In step S5, the rollback window parameters include the maximum rollback duration, the set of recoverable equipment, the set of recoverable materials, the rollback cutoff node, and the upper limit of secondary disturbances. The maximum rollback duration is determined based on the equipment release duration, inventory replenishment duration, process rollback duration, quality re-inspection duration, and shift rescheduling duration. The rollback cutoff node is determined based on the delivery constraint node in the commitment penetration chain. The upper limit of secondary disturbances is determined based on the threshold for the allowed delay duration of non-target orders, the threshold for the allowed additional load of non-target equipment, and the threshold for the allowed occupancy of non-target inventory.

9. The intelligent question-answering and decision support method for production management systems according to claim 8, characterized in that, In step S5, the execution constraints include equipment capacity threshold, inventory availability threshold, shift capacity threshold, quality release conditions, and delivery time limit conditions; when determining the target execution path, only candidate execution paths that meet the following conditions are retained: The equipment load drift, inventory occupancy drift, process waiting drift, and delivery node drift during forward mirroring do not exceed the preset thresholds; The maximum rollback duration, rollback cutoff node, and upper limit of secondary disturbance of the reverse compensation mirror meet the preset conditions; When multiple candidate execution paths meet the conditions, the target execution path is determined in order of priority: delivery node drift amount, equipment load drift amount, and maximum rollback duration.

10. The intelligent question-answering and decision support method for production management systems according to claim 1, characterized in that, In step S6, both the question-and-answer text and the production execution instruction are generated from the target execution path; the question-and-answer text includes the commitment subject identifier, the commitment object identifier, the action type, the execution time window, and the constraints; the production execution instruction includes the execution terminal identifier, the execution node sequence, the resource allocation sequence, the verification conditions, and the rollback trigger conditions. The returned status parameters include equipment status returned parameters, inventory status returned parameters, quality status returned parameters, and order progress returned parameters. By performing node status overwriting, constraint edge opening and closing switching, edge weight recalibration, and failed link truncation and reconnection on the returned status parameters, the association relationships of order nodes, equipment nodes, inventory nodes, quality nodes, and shift nodes in the production object association topology are corrected. Based on the corrected association relationships, the execution time window, resource dependency sequence, verification condition sequence, and rollback trigger condition of the commitment execution unit are updated.

Images