Five-axis numerical control machine tool whole life cycle management system and method

Abstract

The application discloses a five-axis numerical control machine tool full life cycle management system and method, and particularly relates to the technical field of numerical control machine tool full life cycle management, and comprises obtaining structural parameters, assembly deviation, compensation records, processing results and maintenance records formed in the design and research and development stage, the production and manufacturing stage, the installation and debugging stage and the operation and maintenance service stage of a target five-axis numerical control machine tool, and precision requirements, process requirements and delivery requirements corresponding to a to-be-executed processing task, performing time alignment, object alignment and stage attribution processing on the data of each stage, constructing a digital twin responsibility state diagram, and outputting a stage state node set, a stage inheritance edge set and a task responsibility constraint set; the digital twin responsibility state diagram is constructed by fusing full life cycle stage data and task responsibility constraints, and based on a current running state, responsibility offset is solved, a candidate management path is generated, a target management action is determined by optimization and feedback is executed and written back to update.

Description

Technical Field

[0001] This invention relates to the field of CNC machine tool lifecycle management technology, and more specifically, to a five-axis CNC machine tool lifecycle management system and method. Background Technology

[0002] In the existing five-axis CNC machine tool management technology, the mainstream practice in the industry mainly addresses how to uniformly manage the machine tool's operating status, maintenance needs, and usage efficiency. This typically involves collecting data such as spindle speed, feed rate, cutting force, thermal deformation, vibration, tool wear, and alarm records, and combining this data with digital twin models, operation and maintenance rules, or predictive analysis models to monitor, warn, arrange maintenance, and schedule resources for the machine tool's health status.

[0003] For example, in the high-end equipment manufacturing and aerospace parts processing scenarios in Fujian, the same five-axis CNC machine tool often needs to undertake virtual process verification in the design verification stage, high-speed processing in the batch production stage, and small-batch trial production or high-precision contour processing in the order switching stage. At the same time, it is subject to hard constraints such as continuous production without stopping, frequent switching of processing tasks, significant differences in the accuracy responsibility of different parts, and limited maintenance window.

[0004] Under this constraint, mainstream practices consistently reveal an observable and verifiable flaw: although the system can output unified health conclusions, maintenance recommendations, or production scheduling recommendations based on the same set of operational status data, the management meaning corresponding to the same physical deviation will change substantially when the processing responsibility undertaken by the machine tool changes. This leads to a situation where the machine tool is judged to be usable under ordinary processing tasks, but is actually no longer qualified for delivery under high-precision critical component tasks. The reason for this is that the existing solution only treats the machine tool status as an independent judgment object, without taking the current processing responsibility undertaken by the machine tool as a constraint premise for digital twin interpretation and management decisions.

[0005] Therefore, the technical problem to be solved by this application is: how to incorporate the current machining responsibility of the machine tool and the current physical state of the machine tool into the digital twin management judgment during the full life cycle management of a five-axis CNC machine tool, so as to avoid the same state being output as distorted management conclusions under different responsibility scenarios. Summary of the Invention

[0006] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a five-axis CNC machine tool full lifecycle management system and method. By constructing a digital twin responsibility state diagram that integrates data from all lifecycle stages and task responsibility constraints, and solving the responsibility offset based on the current operating state, generating candidate management paths, selecting the optimal target management action, and performing feedback write-back updates, the problems mentioned in the background art are solved.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for full lifecycle management of five-axis CNC machine tools, comprising:

[0008] S1. Obtain the structural parameters, assembly deviations, compensation records, machining results, and maintenance records of the target five-axis CNC machine tool during the design and development, production and manufacturing, installation and commissioning, and operation and maintenance service stages, as well as the accuracy requirements, process requirements, and delivery requirements corresponding to the machining tasks to be performed. Perform time alignment, object alignment, and stage attribution processing on the data of each stage, construct a digital twin responsibility state diagram, and output the stage state node set, stage inheritance edge set, and task responsibility constraint set.

[0009] S2. Read the current running data and the current processing context data, map the current running data to the corresponding stage state node in the digital twin responsibility state diagram, calculate the state difference between the current running data and each stage state node, and perform state propagation calculation and historical offset inversion calculation along the stage inheritance edge set, and output the responsibility offset result.

[0010] S3. Based on the responsibility offset results, map the task responsibility constraint set to the digital twin responsibility state diagram, generate a continuation path for each stage state node, generate a downgraded acceptance path and a maintenance acceptance path for nodes with excessive responsibility offset, generate a task transfer path for nodes that do not meet delivery requirements, and output a candidate management path set.

[0011] S4. Perform constraint-driven adversarial simulation on each candidate management path in the candidate management path set, calculate the corresponding task completion results, accuracy mismatch results and resource consumption results, and perform multi-objective constraint screening and path advantage ranking to solve for the candidate management path with the highest ranking as the target management path.

[0012] In a preferred embodiment, it further includes:

[0013] S5. When the target management path is solved, the target five-axis CNC machine tool is identified as a machine tool that can accept responsibility and the target management action is output; when the target management path is not solved, the target five-axis CNC machine tool is identified as a machine tool that cannot accept responsibility and the maintenance intervention action or task transfer action is output, generating the responsibility acceptance result.

[0014] S6. Read the processing feedback data, maintenance feedback data and task fulfillment data after the execution of the target management action, calculate the result deviation of the corresponding stage state node, the transmission deviation of the corresponding stage inherited edge and the adaptation deviation of the task responsibility constraint set in the digital twin responsibility state diagram, perform recursive correction and continuous convergence update on the digital twin responsibility state diagram, and output the updated digital twin responsibility state diagram and digital twin management baseline.

[0015] In a preferred embodiment, S1 includes:

[0016] S1-1: Read the equipment identification field, component identification field, process identification field, and time identification field from the structural parameters, assembly deviations, compensation records, processing results, and maintenance records. Perform primary key merging on fields with the same name across stages and mapping merging on fields with different names across stages. Rearrange the data according to the order of occurrence of the design and development stage, production and manufacturing stage, installation and commissioning stage, and operation and maintenance service stage, and output a unified data unit set for each stage.

[0017] S1-2. For the unified data unit set of the stage, extract the stable field representing the inherent capability, the offset field representing the stage change, and the constraint field representing the responsibility boundary. Generate stage state nodes from the stable field, generate stage inheritance edges from the offset fields of the same object in adjacent stages, and map the accuracy requirements, process requirements, and delivery requirements into task responsibility constraint edges, and output the initial structure of the digital twin responsibility state diagram.

[0018] S1-3. Perform closure verification on the initial structure of the digital twin responsibility state graph, identify missing stage state nodes, broken stage inheritance edges, and unmapped task responsibility constraint edges, and write the verified stage state nodes, stage inheritance edges, and task responsibility constraint edges into the stage state node set, stage inheritance edge set, and task responsibility constraint set, respectively, and output the digital twin responsibility state graph.

[0019] In a preferred embodiment, S2 includes:

[0020] S2-1. Read the current running data and the current processing context data, perform context segmentation on the current running data, extract state features from each segment, and write the current processing context data into the corresponding state features to form the current state feature set. Compare the field differences before and after writing and remove mismatched fields that do not correspond to the responsibility level.

[0021] S2-2. Map the current state feature set one by one to the stage state nodes in the digital twin responsibility state diagram, calculate the node difference between each stage state node and the current state feature set, perform node-by-node screening according to the node difference, process correspondence and responsibility level correspondence, retain the stage state nodes whose screening results have not changed, and generate the target stage state node and node difference set.

[0022] In a preferred embodiment, S2 further includes:

[0023] S2-3. Taking the target stage state node and node difference set as input, the node difference is passed along the stage inheritance edge set one by one. The change in difference between the two ends of each stage inheritance edge and the result of consistent direction are calculated. The edge passing result is updated according to the result of consistent change in difference and direction. Stop when the update results of the two rounds are consistent. Output the stage propagation offset set and convergence edge set.

[0024] S2-4. Taking the stage propagation offset set and convergence edge set as input, trace back the offset source stage by stage along the reverse path of the stage inheritance edge set, calculate the contribution value of each stage to the current offset, perform round-by-round resolution according to the contribution value and stage attribution conflict relationship, stop when the stage attribution results of the previous and next rounds are consistent, and output the stage offset contribution set.

[0025] S2-5. Perform consistency checks on the stage offset contribution set, stage propagation offset set, and convergence edge set. Identify conflict items where the stage offset contribution direction is inconsistent with the stage propagation direction. Perform conflict resolution according to the responsibility level correspondence and stage attribution order, and output the responsibility offset result.

[0026] In a preferred embodiment, S3 includes:

[0027] S3-1. Based on the responsibility offset results, task responsibility constraint set and digital twin responsibility state diagram, write the accuracy requirements, process requirements and delivery requirements in the task responsibility constraint set into the state nodes of each stage respectively, calculate the responsibility satisfaction value, delivery satisfaction value and offset occupancy value of each stage state node for the current processing task, and output the node constraint evaluation set.

[0028] S3-2. For the node constraint evaluation set, generate a continuing path for stage state nodes where both responsibility satisfaction value and delivery satisfaction value are met, generate a downgraded path and a maintenance path for stage state nodes where responsibility satisfaction value is not met but delivery satisfaction value is met, generate a task transfer path for stage state nodes where delivery satisfaction value is not met, and connect the corresponding stage state nodes according to the stage inheritance edge set, and output the path candidate set.

[0029] S3-3. Calculate the path responsibility closure value, path delivery closure value, and path offset transfer value for each path in the path candidate set, eliminate paths with phase breaks or responsibility reversals, and write the remaining paths into the candidate management path set.

[0030] In a preferred embodiment, S4 includes:

[0031] S4-1. Obtain the candidate management path set, digital twin responsibility state diagram and task responsibility constraint set. Write the current processing task into the stage state nodes in each candidate management path one by one, write the corresponding responsibility offset into each stage inheritance edge one by one, perform forward deduction along the nodes of each candidate management path in sequence, and perform reverse perturbation along the edges of each candidate management path in sequence to generate the adversarial deduction branch set corresponding to each candidate management path.

[0032] S4-2. Calculate the number of process completion segments, delivery closure segments, and responsibility maintenance segments after forward deduction to form the task completion result. Calculate the precision offset increment and responsibility offset increment of each stage state node before and after reverse disturbance to form the precision mismatch result. Calculate the maintenance insertion number, task switching number, and stage occupation time corresponding to each candidate management path to form the resource occupation result. Output the path deduction result set.

[0033] S4-3. Perform constraint screening on each candidate management path in the path deduction result set, and eliminate candidate management paths with delivery breaks in task completion results, candidate management paths with responsibility reversals in accuracy mismatch results, and candidate management paths with stage conflicts in resource usage results. Then, sort the remaining candidate management paths in the order of task completion results, accuracy mismatch results, and resource usage results in rounds. When the first position of the sorting is consistent in the previous and next rounds, the candidate management path with the first position is determined as the target management path.

[0034] In a preferred embodiment, S5 includes:

[0035] S5-1. Perform a closure check on the order of stage state nodes, the order of stage inheritance edges, and the order of path actions in the target management path. Mark the machine tool corresponding to the target management path that passes the check as a candidate machine tool that can take on responsibility, and mark the machine tool that has not formed a target management path as a candidate machine tool that cannot take on responsibility. Output the intermediate result of the acceptance judgment.

[0036] S5-2. For the intermediate results of the acceptance judgment, extract the machining execution nodes, maintenance insertion nodes and path switching nodes corresponding to the target management path for candidate machine tools whose responsibilities can be accepted, and generate target management actions in the order of node appearance; for candidate machine tools whose responsibilities cannot be accepted, extract the fracture position and fracture type after the last closed node in the candidate management path set, generate maintenance intervention actions or task transfer actions according to the fracture type, and output the action result set.

[0037] S5-3. Write the responsibility attribution for each action result in the action result set, write the target management action and the candidate machine tool for responsibility acceptance into the responsibility acceptance result, write the maintenance intervention action or task transfer action and the candidate machine tool for responsibility non-acceptance into the responsibility non-acceptance result, and generate the responsibility acceptance result.

[0038] In a preferred embodiment, S6 includes:

[0039] S6-1. Map the processing feedback data to the corresponding stage state node in the digital twin responsibility state diagram, map the maintenance feedback data to the corresponding stage inheritance edge, map the task fulfillment data to the task responsibility constraint set, calculate the result deviation of each stage state node, the transmission deviation of each stage inheritance edge and the adaptation deviation of the task responsibility constraint set, and output the deviation association set.

[0040] S6-2. Write back the node parameters for each stage state node according to the result deviation, write back the edge parameters for each stage inherited edge according to the transmission deviation, and write back the constraint boundary for the task responsibility constraint set according to the adaptation deviation. The stop condition is that the node deviation sorting, edge deviation sorting and constraint deviation sorting after the two rounds of writing back are consistent. Output the updated digital twin responsibility state diagram.

[0041] S6-3. Read the updated digital twin responsibility state diagram, perform responsibility closure verification on the updated state nodes of each stage, the inherited edges of each stage, and the task responsibility constraint set, write the node parameters, edge parameters, and constraint boundaries that pass the verification into the digital twin management baseline, and output the digital twin management baseline.

[0042] A five-axis CNC machine tool lifecycle management system, comprising a map construction module, a solution module, a path generation module, an optimization module, a judgment module, and a baseline update module.

[0043] The graph construction module is used to acquire the structural parameters, assembly deviations, compensation records, machining results, and maintenance records of the target five-axis CNC machine tool during the design and development, production and manufacturing, installation and commissioning, and operation and maintenance service stages, as well as the accuracy requirements, process requirements, and delivery requirements corresponding to the machining tasks to be performed. It performs time alignment, object alignment, and stage attribution processing on the data of each stage, constructs a digital twin responsibility state graph, and outputs the stage state node set, stage inheritance edge set, and task responsibility constraint set.

[0044] The solver module is used to read the current running data and the current processing context data, map the current running data to the corresponding stage state node in the digital twin responsibility state diagram, calculate the state difference between the current running data and each stage state node, and perform state propagation calculation and historical offset inversion calculation along the stage inheritance edge set, and output the responsibility offset result.

[0045] The path generation module is used to map the task responsibility constraint set to the digital twin responsibility status diagram based on the responsibility offset result, generate a continuation path for each stage status node, generate a downgraded succession path and a maintenance succession path for nodes with excessive responsibility offset, generate a task transfer path for nodes that do not meet the delivery requirements, and output a candidate management path set.

[0046] The optimization module is used to perform constraint-driven adversarial simulation on each candidate management path in the candidate management path set, calculate the corresponding task completion results, accuracy mismatch results and resource consumption results, and perform multi-objective constraint screening and path advantage ranking to solve the candidate management path with the first place in the ranking as the target management path.

[0047] The determination module is used to identify the target five-axis CNC machine tool as a machine tool that can assume responsibility and output target management actions when the target management path is solved; when the target management path is not solved, the target five-axis CNC machine tool is identified as a machine tool that cannot assume responsibility and output maintenance intervention actions or task transfer actions, generating responsibility assumption results.

[0048] The baseline update module is used to read the processing feedback data, maintenance feedback data and task fulfillment data after the execution of target management actions, calculate the result deviation of the corresponding stage state node, the transmission deviation of the corresponding stage inherited edge and the adaptation deviation of the task responsibility constraint set in the digital twin responsibility state diagram, perform recursive correction and continuous convergence update on the digital twin responsibility state diagram, and output the updated digital twin responsibility state diagram and digital twin management baseline.

[0049] The technical effects and advantages of this invention are as follows:

[0050] 1. This solution writes the current processing responsibility and the current physical state into the digital twin responsibility state diagram, and solves the responsibility offset result accordingly. This allows the same physical deviation to be interpreted according to the corresponding responsibility caliber in different responsibility scenarios, thereby relatively suppressing the management distortion caused by the unified health conclusion.

[0051] 2. Perform time alignment, object alignment, and stage attribution processing on data during the design and development, production and manufacturing, installation and commissioning, and operation and maintenance service stages to form stage status nodes, stage inheritance edges, and task responsibility constraint edges. This helps to improve the problem of data fragmentation across stages and enhance the continuity of the entire life cycle management chain.

[0052] 3. Performing state propagation calculation and historical offset inversion calculation along the stage inheritance edge can distinguish the source of the current offset in each life cycle stage, so that the responsibility offset result not only reflects the current state difference, but also reflects the offset attribution relationship, thereby improving the pertinence of subsequent management actions;

[0053] 4. Based on responsibility satisfaction value, delivery satisfaction value, and offset occupancy value, candidate management paths such as continued acceptance, downgraded acceptance, post-maintenance acceptance, and task transfer are generated. This enables management decisions to be based on the combined effect of responsibility constraints and state constraints, thereby relatively improving the adaptability of path generation.

[0054] 5. Perform constraint-driven adversarial simulations on candidate management paths, and filter and rank them based on task completion results, accuracy mismatch results, and resource usage results. This can form a unified comparison standard among delivery constraints, accuracy constraints, and resource constraints, thereby relatively improving the rationality of target management paths. Attached Figure Description

[0055] Figure 1 This is a flowchart outlining the method steps of the present invention;

[0056] Figure 2 This is a schematic diagram of the system module structure of the present invention. Detailed Implementation

[0057] 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.

[0058] Refer to the instruction manual appendix Figure 1-2 The five-axis CNC machine tool lifecycle management method of the present invention includes:

[0059] S1. Obtain the structural parameters, assembly deviations, compensation records, machining results, and maintenance records of the target five-axis CNC machine tool during the design and development, production and manufacturing, installation and commissioning, and operation and maintenance service stages, as well as the accuracy requirements, process requirements, and delivery requirements corresponding to the machining tasks to be performed. Perform time alignment, object alignment, and stage attribution processing on the data of each stage, construct a digital twin responsibility state diagram, and output the stage state node set, stage inheritance edge set, and task responsibility constraint set.

[0060] In this implementation, S1 is used to organize the scattered data generated during the design and development phase, the manufacturing phase, the installation and commissioning phase, and the operation and maintenance service phase into a unified data object, and based on this, construct a digital twin responsibility state diagram that can be called upon for subsequent responsibility offset calculation. The processing order is as follows: first, cross-phase field merging and time alignment are completed; then, stable fields, offset fields, and constraint fields are divided according to their functions; subsequently, phase state nodes, phase inheritance edges, and task responsibility constraint edges are generated respectively; finally, closure verification is performed, and the digital twin responsibility state diagram is output. This implementation process includes the following steps:

[0061] The purpose of S1-1 is to form a unified set of stage-specific data units that can be uniformly calculated. Inputs include equipment identifier, component identifier, process identifier, and time identifier fields from structural parameters, assembly deviations, compensation records, processing results, and maintenance records. Processing actions are as follows: Equipment identifier fields are used to lock the equipment object; component identifier fields are used to determine the component level under the equipment object; and process identifier fields are used to determine the processing stage. Fields with the same name across stages are merged using primary keys based on consistent field names and business meanings. Fields with different names across stages are merged using a field mapping table, which includes at least the field name, business meaning, value type, and stage. Time identifier fields are consolidated according to a unified time granularity, which is based on manufacturing processes. The data collection cycle and the operation and maintenance service cycle are relatively coarse-grained; then the records are rearranged in the order of design and development stage, production and manufacturing stage, installation and commissioning stage and operation and maintenance service stage; when the same record crosses the boundary of two stages, it belongs to the stage where the data result is formed first; the output is a unified data unit set for each stage, each unified data unit includes at least equipment object identifier, component level identifier, process identifier, unified time identifier and merged field group, and is written to the stage data cache for S1-2 to read; the abnormal or missing handling is as follows: records with missing equipment identifier fields are directly deleted, records with missing component identifier fields are attached to the equipment object level, and records with missing time identifier fields are filled in according to the median time of adjacent records in the same batch and written with a filling mark;

[0062] The purpose of S1-2 is to convert the unified data unit set of the stage into the initial skeleton of the graph structure. The input is the unified data unit set of the stage and the accuracy requirements, process requirements and delivery requirements corresponding to the processing task to be executed. The processing action is to perform field classification on each unified data unit of the stage. Fields with consistent field names, consistent business meanings and continuous values ​​within the allowable range of manufacturing deviations in the four stages are classified as stable fields. Fields that are added, deleted, drifted, compensated, covered or maintained between adjacent stages are classified as offset fields. Fields that directly limit the processing responsibility boundary, process adaptation boundary or delivery boundary are classified as constraint fields.

[0063] When the same field has both stable and offset attributes, it is split into a base value field and a variable field. Stage state nodes are generated using the stable fields of the same equipment object in the same stage. Stage inheritance edges are generated using the difference in offset fields of the same equipment object in adjacent stages. Task responsibility constraint edges are generated using accuracy requirements, process requirements, and delivery requirements. Accuracy requirements are mapped to capability boundary fields, process requirements are mapped to process adaptation fields, and delivery requirements are mapped to delivery closure fields. The output is the initial structure of the digital twin responsibility state graph and is written to the graph structure buffer for S1-3 to read. The handling of anomalies or missing fields is as follows: fields whose categories cannot be determined are written to the unclassified field area and do not participate in the generation of nodes and edges in this round. Task objects with missing task responsibility fields do not generate task responsibility constraint edges and are marked with unmapped constraints.

[0064] The purpose of S1-3 is to confirm that the initial structure of the digital twin responsibility state diagram satisfies the closure conditions for subsequent offset propagation and path deduction. The input is the initial structure of the digital twin responsibility state diagram. The processing actions are as follows: First, check whether each device object has a unique stage state node in the design and development stage, production and manufacturing stage, installation and commissioning stage, and operation and maintenance service stage. If it is missing, it is recorded as a missing stage state node. Then, check whether there is a stage inheritance edge connecting the same device object between adjacent stage nodes. If it is missing, it is recorded as a broken stage inheritance edge. Then, check whether each processing task to be executed has task responsibility constraint edges corresponding to accuracy requirements, process requirements, and delivery requirements. If they are missing, they are recorded as unmapped task responsibility constraint edges.

[0065] For the parts that pass the verification, the stage status node is written to the stage status node set, the stage inheritance edge is written to the stage inheritance edge set, and the task responsibility constraint edge is written to the task responsibility constraint set. For the parts that fail the verification, if a stage status node is missing and a complete record exists in an adjacent stage, a supplementary node is generated based on the inheritance of the stable field of the adjacent stage. If there is a broken stage inheritance edge, the inheritance edge is supplemented based on the difference in the corresponding offset field of the preceding and following nodes. If there is an unmapped task responsibility constraint edge, the constraint edge is supplemented by reading back from the task responsibility field source table. The consistency of the verification results of the previous and subsequent rounds is used as the stopping condition. The digital twin responsibility state diagram is output and written to the S2 call area. The handling of anomalies or missing parts is as follows: nodes, edges and constraint edges that still cannot be closed after supplementation are written to the broken object record table and removed from the digital twin responsibility state diagram of this round.

[0066] Through the above steps, data across stages, names, and timeframes are unified into a single digital twin responsibility state diagram. This ensures that subsequent responsibility offset calculations, candidate path generation, and target path deduction are all based on a structure that ensures consistency in objects, stages, and responsibilities. In practical applications: a five-axis CNC machine tool forms structural parameters and target accuracy boundaries during the design and development stage, assembly deviation records during the manufacturing stage, compensation records during the installation and commissioning stage, and maintenance records and processing results during the operation and maintenance service stage. This implementation first locks the same machine tool object according to the equipment identifier, then forms unified data units for each stage according to the process identifier and a unified time granularity. Subsequently, stable fields such as spindle geometric base values ​​and rotary table inherent error base values ​​are extracted to generate stage state nodes, and offset fields such as assembly deviation increments and thermal compensation corrections are extracted to generate stage inheritance edges. The accuracy requirements, process requirements, and delivery requirements of the current processing task are mapped to task responsibility constraint edges. Finally, a digital twin responsibility state diagram is obtained through closure verification.

[0067] S2. Read the current running data and the current processing context data, map the current running data to the corresponding stage state node in the digital twin responsibility state diagram, calculate the state difference between the current running data and each stage state node, and perform state propagation calculation and historical offset inversion calculation along the stage inheritance edge set, and output the responsibility offset result.

[0068] In this embodiment, S2 is used to map the current operating state to the digital twin responsibility state diagram and solve the responsibility offset result of the current state relative to the baseline of each lifecycle stage. The principle is as follows: first, the current operating data is segmented according to the current processing context and state features are extracted; then, the state features are mapped to the state nodes of each stage to form node differences; then, offset propagation is performed along the stage inheritance edge; then, offset source backtracking is performed along the reverse path; finally, conflicts are resolved through consistency verification to obtain a unique responsibility offset result. This implementation process includes the following steps:

[0069] The purpose of S2-1 is to form a current state feature set that can be used for node mapping, and to first remove mismatched fields that do not match the current responsibility level. The inputs are current running data and current machining context data. The current running data includes at least spindle load, axis position deviation, cutting force, thermal deformation, and vibration response. The current machining context data includes at least workpiece category, process type, tool combination, material properties, and responsibility level. The processing action is as follows: First, the current running data is segmented according to the process type change points, tool combination switching points, and responsibility level switching points in the current machining context data to form multiple running segments. Then, state features are extracted for each running segment. The state features consist of the segment mean, segment range, segment end value, and adjacent sample difference of each running field within the segment.

[0070] Subsequently, the workpiece category, process type, tool combination, material properties, and responsibility level from the current processing context data are written into the status features of the corresponding running segment, forming the current status feature set after writing. Then, the field differences before and after writing are compared. The field differences are generated based on whether the field value after writing is consistent with the field value of the same name before writing, or whether the field label after writing is consistent with the field label before writing. If the process type field and the responsibility level field in a certain status feature do not have a pairing relationship in the responsibility level correspondence table corresponding to the digital twin responsibility status diagram, then the field is identified as a mismatch field and removed from the current status feature set. The output is the current status feature set, which is written to the node mapping buffer for S2-2 to read. The abnormal or missing handling is as follows: when a running segment is missing a single running field, it is linearly filled in according to the adjacent sampling points of the segment. If the consecutive missing exceeds one-third of the sampling number of the segment, the running segment is directly removed. When the current processing context data is missing the responsibility level, the task responsibility constraint set corresponding to the current processing task is read back to fill in the responsibility level. If it is still missing, the current round of mapping is stopped and a context missing mark is written.

[0071] The purpose of S2-2 is to locate the target stage state node corresponding to the current state feature set in the digital twin responsibility state diagram and generate the node difference set required for subsequent propagation. The inputs are the current state feature set and the digital twin responsibility state diagram. The processing actions are as follows: map the current state feature set to each stage state node one by one, calculate the node difference for each stage state node, and obtain the node difference by comparing each field in the current state feature set with the corresponding field of the stage state node item by item. The numerical field uses the absolute value of the difference, the category field uses a consistent label, and the direction field uses a positive and negative consistent label. Then, perform node-by-node screening according to the node difference, the process correspondence, and the responsibility level correspondence. Among them, the node difference is used to determine the closeness of the current state to the stage baseline, the process correspondence is used to determine whether the current process type is consistent with the process carried by the stage node, and the responsibility level correspondence is used to determine whether the current responsibility level falls within the responsibility range carried by the stage node.

[0072] During screening, first, stage status nodes with invalid process correspondences are removed, then stage status nodes with invalid responsibility level correspondences are removed. The remaining stage status nodes are then sorted in ascending order of node differences. The condition for stopping is that the order of stage status nodes retained after the previous two rounds of screening remains consistent. Stage status nodes whose screening results have not changed are retained as target stage status nodes, and their corresponding differences are written into the node difference set. The output consists of the target stage status nodes and the node difference set, which are written to the propagation calculation buffer for S2-3 to read. Anomaly or missing information is handled as follows: if a stage status node is missing a corresponding field, that field is not included in the node difference calculation for this round, and a missing field marker is written into the node difference set. If multiple nodes are ranked in parallel, the later stage status nodes are retained first according to the stage order to ensure that subsequent offset propagation prioritizes the interpretation of the current responsibility status.

[0073] The purpose of S2-3 is to propagate node differences along the stage inheritance edge set, solve the propagation results of the current offset on each stage chain, and lock the converged edge set. The inputs are the target stage state node, the node difference set, and the stage inheritance edge set. The processing actions are as follows: taking the target stage state node as the propagation starting point, writing the differences of each field in the node difference set into the stage inheritance edge connected to the target stage state node, and performing edge-by-edge propagation in the stage order; for each stage inheritance edge, calculating the change in difference between the node difference at the front end of the edge and the node difference at the back end of the edge, and generating a direction consistency result based on whether the direction of the difference at the front end of the edge and the direction of the difference at the back end of the edge are consistent; when the change in difference of a stage inheritance edge is consistent with the direction of the offset field recorded by the edge, the edge is recorded as a valid propagation edge in this round, and the propagation result of the edge is written into the next stage node; when the directions are inconsistent, the edge is recorded as an edge to be verified and does not continue to propagate to the next stage.

[0074] Then, all edge propagation results are updated according to the consistency of the difference change and direction, and the update order is executed in the stage order. The stopping condition is that the effective propagation edge order and edge propagation results are consistent after the previous two rounds of updates. When stopping, the corresponding results of all effective propagation edges are written into the stage propagation offset set, and all effective propagation edges are written into the convergence edge set. The output is the stage propagation offset set and the convergence edge set, which are written into the offset inversion buffer for S2-4 to read. The exception or missing handling is as follows: when the offset field of the inherited edge is missing in a certain stage, it will not participate in the current round of propagation and an edge missing mark will be written. If all edges are edges to be verified when propagation reaches a certain stage, the propagation will stop and the effective propagation edges formed in the previous stage will be output as the convergence edge set.

[0075] The purpose of S2-4 is to trace back the current offset source based on the stage propagation offset set and the convergence edge set, and solve for the contribution value of each stage to the current offset. The inputs are the stage propagation offset set, the convergence edge set, and the digital twin responsibility state diagram. The processing action is as follows: trace back the offset source from the current target stage state node to the design and development stage, the production and manufacturing stage, the installation and commissioning stage, and the operation and maintenance service stage one by one along the reverse path of the convergence edge set. For each stage, read the propagation offset corresponding to the stage state node and the inherited edge of its successor stage, and calculate the stage contribution value of the current offset by subtracting the propagation offset of the successor stage from the current stage's propagation offset.

[0076] When multiple stages contribute to the same offset field simultaneously, a stage attribution conflict relationship is generated, and the conflict is resolved round by round according to the stage contribution value from largest to smallest. If the stage contribution values ​​are the same, the earlier stage is retained as the offset source in order of stage sequence to ensure that the cause of the preceding stage is explained first. The consistency of the stage attribution results in the previous and next rounds is used as the stopping condition. The output is the stage offset contribution set, which is written to the consistency check buffer for S2-5 to read. The exception or missing handling is as follows: when a reverse path is interrupted, the previous stage corresponding to the last converged edge before the interruption is used as the temporary offset source, and a path interruption mark is written. If the same offset field cannot form a valid contribution value in all stages, the offset field is written to the unattributed field table and does not participate in the generation of the responsibility offset result in this round.

[0077] The purpose of S2-5 is to perform a closure check on the propagation results and inversion results, resolve multi-stage attribution conflicts, and form a unique responsibility offset result. The inputs are the stage offset contribution set, the stage propagation offset set, and the convergence edge set. The processing actions are as follows: read the contribution direction of each offset field in the stage offset contribution set one by one, and perform a consistency check with the corresponding propagation direction in the stage propagation offset set. When the stage offset contribution direction is consistent with the stage propagation direction, the offset field is recorded as a consistent item; when the two are inconsistent, the offset field is recorded as a conflict item. Subsequently, conflict items are resolved according to the responsibility level correspondence and the stage attribution order. The responsibility level correspondence is used to determine the responsibility interval to which the offset field should be preferentially assigned, and the stage attribution order is used to determine the order of attribution within the same responsibility interval.

[0078] If a conflict item cannot be uniquely determined under both the responsibility level correspondence and the stage attribution order, then the node difference set is read back, and the stage with the larger node difference is taken as the stage to which the conflict item belongs; the consistent item is merged with the resolved conflict item to form the responsibility offset result, and the responsibility offset result is written to the S3 call area; the output is the responsibility offset result; the abnormal or missing handling is as follows: when a certain offset field lacks a propagation direction or contribution direction, it is directly marked as a field to be reviewed and removed from the responsibility offset result of this round;

[0079] Through the above steps, the current operating status can be transformed from the raw data layer into a responsibility offset result, clarifying how the offset propagates along the lifecycle stages and which stage dominates its formation, thus providing a unified input for the generation of subsequent candidate management paths. In practical applications: A five-axis CNC machine tool is currently undertaking the precision machining of thin-walled aerospace parts. The system first segments the spindle load, cutting force, and thermal deformation according to tool switching points and process switching points, extracts the state characteristics of each segment, and writes them into the current process type and responsibility level. Then, the state characteristics of each segment are mapped to the design and development stage, production and manufacturing stage, installation and commissioning stage, and operation and maintenance service stage nodes in the digital twin responsibility state diagram, and the target stage state nodes consistent with the current responsibility level are screened out. Then, the node difference is propagated along the stage inheritance edge set to obtain the stage propagation offset set corresponding to the thermal compensation correction offset and assembly deviation offset. Then, the contribution values ​​of each stage are traced back to determine that the current responsibility offset mainly comes from the compensation coverage of the installation and commissioning stage and the accumulation of assembly deviations in the production and manufacturing stage. Finally, the responsibility offset result is output after consistency verification, which can be directly called for the generation of subsequent continuing to undertake paths, maintenance-inherited paths, or task transfer paths.

[0080] S3. Based on the responsibility offset results, map the task responsibility constraint set to the digital twin responsibility state diagram, generate a continuation path for each stage state node, generate a downgraded acceptance path and a maintenance acceptance path for nodes with excessive responsibility offset, generate a task transfer path for nodes that do not meet delivery requirements, and output a candidate management path set.

[0081] In this embodiment, S3 is used to transform the responsibility offset result into an executable candidate management path set. Its processing logic is as follows: First, the task responsibility constraint set is written into the stage state nodes of the digital twin responsibility state diagram to form node-level constraint evaluation results; then, based on the node-level constraint evaluation results, continuation paths, downgraded paths, post-maintenance paths, and task transfer paths are generated; finally, closure verification and inversion screening are performed on each path to output a candidate management path set that can be used for subsequent adversarial simulation. This implementation process includes the following steps:

[0082] The purpose of S3-1 is to apply the responsibility offset results and the task responsibility constraint set to the status nodes of each stage, forming a node-level responsibility assignment evaluation result. The inputs are the responsibility offset results, the task responsibility constraint set, and the digital twin responsibility status diagram. The processing actions are as follows: First, read the accuracy requirements, process requirements, and delivery requirements from the task responsibility constraint set, and write the three types of requirements into the status nodes of each stage according to the task object. Specifically, the accuracy requirements are written into the capability boundary field of the stage status node, the process requirements are written into the process adaptation field of the stage status node, and the delivery requirements are written into the delivery closure field of the stage status node. Then, read the stage affiliation, offset direction, and offset amount corresponding to each offset field in the responsibility offset results, and write them into the offset occupancy field of the corresponding stage status node.

[0083] Subsequently, for each stage status node, the responsibility fulfillment value, delivery fulfillment value, and offset occupancy value are calculated. The responsibility fulfillment value is generated based on whether the capability boundary field of the current stage status node corresponds simultaneously with the accuracy requirements and process requirements; a correspondence indicates fulfillment, while a non-correspondence indicates non-fulfillment. The delivery fulfillment value is generated based on whether the delivery closure field of the current stage status node corresponds with the delivery requirements; a correspondence indicates fulfillment, while a non-correspondence indicates non-fulfillment. The offset occupancy value is generated as the ratio of the number of offset fields written to the stage status node in the responsibility offset result to the number of offset fields that the node can accept. The output quantity is the node... The constraint evaluation set requires each stage state node to have at least a corresponding responsibility fulfillment value, delivery fulfillment value, and offset occupancy value. The node constraint evaluation set is written to the path generation cache for S3-2 to read. Anomaly or missing field handling is as follows: if a stage state node is missing a capability boundary field, the responsibility fulfillment value for that node is not calculated and a missing field flag is added to the node; if a task object is missing delivery requirements, the task responsibility field source table is read back to complete the delivery requirements; if still missing, the task object is removed from this round of path generation; if a node does not have a responsibility offset result written, its offset occupancy value is recorded as zero and it is retained for subsequent path generation.

[0084] The purpose of S3-2 is to generate four types of candidate paths based on the node constraint evaluation set and form a path candidate set. The inputs are the node constraint evaluation set, the digital twin responsibility state graph, and the stage inheritance edge set. The processing actions are as follows: First, read the responsibility satisfaction value, delivery satisfaction value, and offset occupancy value of each stage state node one by one. For stage state nodes where both the responsibility satisfaction value and delivery satisfaction value are met, generate a continuing path. The generation method is: take the stage state node as the starting node, and connect the subsequent stage state nodes where the responsibility satisfaction value and delivery satisfaction value are met along the stage inheritance edge set until a node that does not meet the requirements or the stage end occurs. For stage state nodes where the responsibility satisfaction value does not meet the requirements but the delivery satisfaction value meets the requirements, generate a downgraded path and a maintained path respectively. The downgraded path is generated by: taking the current stage state node as the starting node, writing back the task responsibility constraint set corresponding to the lower responsibility level, and then connecting the stage state nodes where the responsibility satisfaction value is changed to meet the requirements along the stage inheritance edge set.

[0085] The maintenance-following path generation method is as follows: Starting with the current stage status node, a maintenance action node is inserted after it. The compensation correction field in the maintenance record is written to the subsequent stage status node. Then, the stage status nodes whose responsibility satisfaction value is satisfied are connected along the stage inheritance edge set. For stage status nodes whose delivery satisfaction value is not satisfied, a task transfer path is generated. The generation method is as follows: using the current stage status node as the breakpoint node, the connection is stopped at this node, and the current processing task is remapped to the corresponding stage status node of the same type of equipment object, forming a cross-equipment task transfer chain. Subsequently, the preceding and following stage status nodes in each path are connected according to the stage inheritance edge set to obtain a path candidate set. The output is the path candidate set, which is written to the path verification buffer for S3-3 to read. Anomaly or missing information handling is as follows: when a stage status node does not have a subsequent stage inheritance edge, the node is marked as a termination node and the path generation ends. When a equipment object does not have a similar stage status node that can receive task transfers, no task transfer path is generated and a transfer failure flag is written. The value of the downgraded responsibility level comes from the responsibility level correspondence table in the task responsibility constraint set. If there is no lower responsibility level, no downgraded successor path is generated.

[0086] The purpose of S3-3 is to perform path-level closure evaluation on the candidate path set, eliminate paths that cannot stably undertake tasks, and form a candidate management path set. The inputs are the candidate path set, responsibility offset results, and stage inheritance edge set. The processing actions are as follows: for each path in the candidate path set, calculate the path responsibility closure value, path delivery closure value, and path offset propagation value. The path responsibility closure value is generated by the ratio of the number of consecutive segments in the path where the responsibility of each stage state node satisfies the value to the total number of segments in the path. The path delivery closure value is generated by the ratio of the number of consecutive segments in the path where the delivery of each stage state node satisfies the value to the total number of segments in the path. The path offset propagation value is generated by the ratio of the number of edges in the path where the corresponding offset directions of adjacent stage inheritance edges are consistent to the total number of edges in the path.

[0087] Subsequently, each path is subject to elimination judgment: when there is no stage inheritance edge connecting the state nodes of the preceding and following stages in the path, it is marked as a stage break path and eliminated; when the responsibility satisfaction value of the state node of the preceding stage is met but the responsibility satisfaction value of the state node of the following stage is not met, and this change is not triggered by the downgraded or maintained successor path, it is marked as a responsibility reversal path and eliminated; for paths that are not eliminated, their path responsibility closure value, path delivery closure value, and path offset transfer value are retained and written into the candidate management path set; the output is the candidate management path set and written into the S4 call area; the exception or missing handling is as follows: when a path lacks a single segment evaluation value, the existing continuous segment evaluation values ​​are used to participate in the closure calculation and a segment missing mark is written; if all paths are eliminated, an empty candidate management path set is generated and a no valid path mark is written, so that the subsequent S4 can directly output the result of the unsolved target management path;

[0088] Through the above steps, the node-level responsibility evaluation results can be transformed into path-level management candidate results. This ensures that the four management strategies—continued acceptance, downgraded acceptance, post-maintenance acceptance, and task transfer—are all based on a unified responsibility caliber and a unified graph structure. Furthermore, before entering the adversarial simulation, path breaks and responsibility reversals are eliminated. In practical application: a five-axis CNC machine tool is currently undertaking a high-precision impeller machining task. The responsibility offset results show that the compensation offset during the installation and commissioning phase has approached the responsibility boundary, while the delivery closure field during the maintenance service phase still meets the current delivery requirements. This implementation method first determines the task's contour accuracy requirements, toolpath process requirements, and delivery time... The system requires writing status nodes at each stage to determine the responsibility fulfillment value, delivery fulfillment value, and offset occupancy value for the production and installation / commissioning stages. Then, for nodes where both responsibility and delivery fulfillment values ​​are met, a continuation path is generated. For nodes where responsibility fulfillment values ​​are not met but delivery fulfillment values ​​are met, a downgraded path and a post-maintenance path are generated, respectively. For nodes where delivery fulfillment values ​​are not met, a task transfer path is generated. Finally, the system calculates the path responsibility closure value, path delivery closure value, and path offset transfer value for each path, eliminating paths with stage breaks or responsibility reversals, forming a candidate management path set for direct use in subsequent target management path calculations.

[0089] S4. Perform constraint-driven adversarial simulation on each candidate management path in the candidate management path set, calculate the corresponding task completion results, accuracy mismatch results and resource consumption results, and perform multi-objective constraint screening and path advantage ranking to solve the candidate management path with the first-ranked result as the target management path.

[0090] In this embodiment, S4 is used to solve for the target management path that corresponds to the current processing task from the candidate management path set. Its processing logic is as follows: First, the current processing task and responsibility offset are written into the stage state nodes and stage inheritance edges of each candidate management path, forming a deducible adversarial deduction branch; then, path deduction results are generated from three dimensions: task completion, accuracy mismatch, and resource consumption; finally, constraint filtering and round-by-round sorting are performed on each candidate management path to output the target management path. This implementation process includes the following steps:

[0091] The purpose of S4-1 is to transform the candidate management path set into a comparable adversarial inference branch set, so that each path is simultaneously subjected to task acceptance verification and responsibility disturbance verification. The inputs are the candidate management path set, the digital twin responsibility state diagram, and the task responsibility constraint set. The processing actions are as follows: First, read the stage state node sequence and stage inheritance edge sequence in the candidate management path one by one, and write the accuracy requirements, process requirements, and delivery requirements in the current processing task into the corresponding stage state node one by one. Among them, the accuracy requirements are written into the capability boundary field, the process requirements are written into the process adaptation field, and the delivery requirements are written into the delivery closure field. Then, write the offset amount, offset direction, and stage affiliation corresponding to each offset field in the responsibility offset result into the corresponding stage inheritance edge one by one to form the responsibility disturbance writing result.

[0092] Subsequently, forward deduction is performed sequentially along the nodes of the candidate management path. During forward deduction, each node in the path is checked segment by segment to determine whether it still meets the responsibility assumption condition after writing the current processing task, and the deduction result of the previous node is passed to the next node, generating a forward deduction chain. Then, reverse perturbation is performed sequentially along the edges of the candidate management path. During reverse perturbation, the responsibility offset is written back edge by edge from the end stage of the path, and the reverse influence of the inherited edge of each stage on the state node of the previous stage after assuming the responsibility offset is examined, generating a reverse perturbation chain. The same path is then used to... The corresponding forward deduction chain and reverse perturbation chain are combined to form the adversarial deduction branch set of the path; the output is the adversarial deduction branch set corresponding to each candidate management path, and is written to the path deduction buffer for S4-2 to read; the abnormal or missing handling is as follows: when a candidate management path is missing a stage state node, the path deduction is terminated and a missing node mark is written; when a candidate management path is missing a stage inherited edge, the reverse perturbation is terminated and a missing edge mark is written; when any constraint field of the task responsibility constraint set is missing, the task responsibility field source table is read back and rewritten; if it is still missing, the corresponding candidate management path is removed.

[0093] The purpose of S4-2 is to transform the adversarial simulation branch set into a sortable path simulation result set. The inputs are the adversarial simulation branch set, the digital twin responsibility state diagram, and the responsibility offset result. The processing actions are as follows: For the forward simulation chain corresponding to each candidate management path, the number of process completion segments, the number of delivery closure segments, and the number of responsibility retention segments are counted respectively. Among them, the number of process completion segments is counted according to the number of consecutive node segments in the forward simulation chain where the process adaptation field is kept true, the number of delivery closure segments is counted according to the number of consecutive node segments where the delivery closure field is kept true, and the number of responsibility retention segments is counted according to the number of consecutive node segments where the responsibility satisfaction value is kept true. The three are combined to form the task completion result. For the reverse perturbation chain corresponding to each candidate management path, the precision offset increment and responsibility offset increment of the state nodes before and after the reverse perturbation are counted respectively. Among them, the precision offset increment is generated according to the difference between the precision boundary field of the node after the reverse perturbation and the precision boundary field of the node before the reverse perturbation, and the responsibility offset increment is generated according to the difference between the responsibility occupation field of the node after the reverse perturbation and the responsibility occupation field of the node before the reverse perturbation. The two are combined to form the precision mismatch result.

[0094] For each candidate management path, the number of maintenance insertions, task switching times, and stage duration are calculated. Maintenance insertions are counted based on the number of times maintenance action nodes appear in the path; task switching times are counted based on the number of times task objects change in the path; and stage duration is calculated by summing the time intervals corresponding to the state nodes of each stage in the path. These three are combined to form the resource usage result. The output is a path deduction result set. Each candidate management path corresponds to at least one task completion result, one precision mismatch result, and one resource usage result, which are written to the path filtering cache for S4-3 to read. Anomaly or missing information is handled as follows: if a path is interrupted in the forward deduction chain, only the number of consecutive segments established before the interruption is counted and written as a deduction interruption flag; if a path is missing a comparison field in the reverse perturbation chain, this field is not included in the incremental calculation for this round, and a field missing flag is written; if stage duration is missing, it is filled in using the median time of the adjacent stage states.

[0095] The purpose of S4-3 is to perform constraint filtering and round-by-round sorting on the path deduction result set to solve for the unique target management path. The inputs are the path deduction result set and the task responsibility constraint set. The processing actions are as follows: First, perform constraint filtering on each candidate management path one by one. If there is an interruption in the number of process completion segments and an interruption in the number of delivery closure segments in the task completion results, then the path is identified as a candidate management path with delivery breakage and is eliminated. If there is an increase in the precision offset increment and an increase in the responsibility offset increment in the precision mismatch results, and the direction of this change is opposite to the direction of the decrease in the original responsibility level of the path, then the path is identified as a candidate management path with responsibility reversal and is eliminated. If there are multiple stage status nodes occupying the same equipment object in the same time interval in the resource usage results, or maintenance action nodes and processing execution nodes overlap in the same time interval, then the path is identified as a candidate management path with stage conflict and is eliminated.

[0096] For candidate management paths that have not been eliminated, they are first sorted by task completion results, with the sorting rule being: priority given to the number of completed process segments, followed by the number of delivery closure segments, and then the number of responsibility maintenance segments. Next, they are sorted by precision mismatch results, with the sorting rule being: priority given to those with smaller precision offset increments, followed by those with smaller responsibility offset increments. Finally, they are sorted by resource usage results, with the sorting rule being: priority given to those with fewer maintenance insertions, followed by those with fewer task switching times, and then those with shorter stage durations. The above sorting uses a round-by-round sorting method, with the previous round's sorting result serving as the input for the next round. Sorting stops when the first and last two rounds' sorting positions are the same, and the candidate management path at the top of the sort is determined as the target management path. The output is the target management path, written to the S5 call area. Anomaly or missing path handling is as follows: if all candidate management paths are eliminated in the constraint screening, the "unsolved target management path" marker is output; if the top sorting positions are tied, the path with fewer task switching times in the path stage sequence is prioritized, and if still tied, the path with fewer maintenance insertion times is prioritized.

[0097] Through the above steps, a unified deduction caliber can be formed among candidate management paths such as continued acceptance, downgraded acceptance, post-maintenance acceptance, and task transfer, based on the same task responsibility constraints and the same responsibility offset conditions. Path selection is constrained by three types of results: task completion, accuracy mismatch, and resource usage, thus avoiding path selection based solely on a single indicator. In practical application: a five-axis CNC machine tool is currently undertaking impeller finishing tasks. The candidate management path set simultaneously includes continued acceptance, post-maintenance acceptance, and task transfer paths. This implementation first writes the task's accuracy requirements, process requirements, and delivery requirements into the stage status nodes corresponding to each path. Then, the previously obtained thermal compensation offset and assembly offset are written into the stage inheritance edges corresponding to each path. Subsequently, forward deduction and reverse perturbation are performed on each path. Statistical analysis shows that although the continued acceptance path has a complete number of completed process segments, the accuracy offset increment increases; although the post-maintenance acceptance path adds one maintenance insertion, the responsibility segment count remains complete; and although the task transfer path eliminates the current machine tool's responsibility offset, it generates two task switches. Finally, paths with stage conflicts are eliminated, and the post-maintenance acceptance path is determined as the target management path through round-by-round sorting.

[0098] S5. When the target management path is solved, the target five-axis CNC machine tool is identified as a machine tool that can accept responsibility and the target management action is output; when the target management path is not solved, the target five-axis CNC machine tool is identified as a machine tool that cannot accept responsibility and the maintenance intervention action or task transfer action is output, generating the responsibility acceptance result.

[0099] In this embodiment, S5 is used to determine whether the target five-axis CNC machine tool can undertake the current machining task based on whether the target management path forms a closed acceptance chain, and generates corresponding target management actions, maintenance intervention actions, or task transfer actions accordingly, ultimately forming a responsibility acceptance result. Its processing logic is as follows: First, perform path closure verification on the target management path to distinguish between candidate machine tools that can undertake the responsibility and candidate machine tools that cannot. Then, extract the path action nodes of the machine tools that can undertake the responsibility and the break positions and break types of the machine tools that cannot. Finally, write the action results and the machine tool object's responsibility assignment to form a responsibility acceptance result that can be directly used for subsequent feedback updates. This implementation process includes the following steps:

[0100] The purpose of S5-1 is to confirm whether the target management path constitutes a complete acceptance chain and to complete the acceptance grouping of machine tool objects accordingly. The input quantities are the target management path, the candidate management path set, and the digital twin responsibility state diagram. The processing actions are as follows: first, read the sequence of stage state nodes, the sequence of stage inheritance edges, and the sequence of path actions in the target management path, and perform a closure check on the three. The sequence of stage state nodes is used to check whether the preceding and following stages are arranged in a continuous manner according to the design and development stage, the production and manufacturing stage, the installation and commissioning stage, and the operation and maintenance service stage. The sequence of stage inheritance edges is used to check whether there are corresponding stage inheritance edges between adjacent stage state nodes. The sequence of path actions is used to check whether the processing execution action, the maintenance insertion action, and the path switching action correspond one-to-one with the corresponding stage state nodes and that no reversal has occurred.

[0101] When the nodes, edges, and actions are sequentially consecutive, the target management path is deemed to have passed verification, and the corresponding machine tool is marked as a candidate machine tool for responsibility acceptance. When the target management path is missing any stage state node, any necessary stage inherited edge, or any path action, the target management path is deemed to have failed verification, and the machine tool that has not formed the target management path is marked as a candidate machine tool for responsibility acceptance. The output is the intermediate result of the acceptance determination, which includes at least the machine tool identifier, acceptance mark, and path closure mark, and is written to the action generation buffer for S5-2 to read. The exception or missing handling is as follows: when the target management path is missing a stage state node, it is directly recorded as a node broken path; when the target management path is missing a stage inherited edge, it is directly recorded as an edge broken path; when the target management path is missing a path action, it is directly recorded as an action broken path. If the candidate management path set is empty, the corresponding machine tool is directly marked as a candidate machine tool for responsibility acceptance.

[0102] The purpose of S5-2 is to generate action results corresponding to the machine tool's acceptance status based on the intermediate results of the acceptance determination. The inputs are the intermediate results of the acceptance determination, the target management path, and the candidate management path set. The processing actions are as follows: First, for candidate machine tools that can accept responsibility, read the machining execution nodes, maintenance insertion nodes, and path switching nodes from the target management path, and generate target management actions according to the order in which the nodes appear in the path. Specifically, the machining execution node is used to generate the continue machining action, the maintenance insertion node is used to generate the planned maintenance action, and the path switching node is used to generate the stage switching action. Then, for candidate machine tools that cannot accept responsibility, read the actions following the last closed node in the candidate management path set. The location and type of the fracture are defined as follows: the last closed node is the last stage state node that maintains continuity in node order, edge order, and action order. The fracture location is determined by the first fracture node or the first fracture edge after the last closed node. The fracture type is determined by three rules: delivery fracture, responsibility reversal, and stage conflict. Specifically, if the number of delivery closure segments is interrupted in the task completion result, it is called a delivery fracture. If the accuracy offset increment and responsibility offset increment increase simultaneously in the accuracy mismatch result and are inconsistent with the direction of change of responsibility level, it is called a responsibility reversal. If there is overlapping occupation of equipment objects or overlapping maintenance actions and processing actions in the same time interval in the resource occupation result, it is called a stage conflict.

[0103] Subsequently, action results are generated according to the fracture type. Delivery fracture corresponds to the generation of task transfer action, responsibility reversal corresponds to the generation of maintenance intervention action, and stage conflict corresponds to the generation of a combination of maintenance intervention action and task transfer action. The output is an action result set, which includes at least machine tool identifier, action type, action start node, and action execution order, and is written to the responsibility attribution cache for S5-3 to read. The handling of anomalies or missing information is as follows: when the maintenance insertion node is missing for the candidate machine tool for which responsibility can be undertaken, only the continue processing action and stage switching action are generated; when the fracture type of the candidate machine tool for which responsibility cannot be undertaken cannot be determined, the path deduction result set is read back and re-determined. If it still cannot be determined, it is uniformly written as a task transfer action and marked as pending review.

[0104] The purpose of S5-3 is to bind the action results with the machine tool object's execution responsibility, forming a unique responsibility acceptance result. The input quantities are the action result set and the intermediate results of the acceptance judgment. The processing actions are as follows: read each action result in the action result set one by one, and for candidate machine tools whose responsibility can be accepted, write the target management action with the machine tool identifier, target management path identifier, and acceptance mark into the responsibility acceptable result; for candidate machine tools whose responsibility cannot be accepted, write the maintenance intervention action or task transfer action with the machine tool identifier, fracture location, fracture type, and unacceptable mark into the responsibility unacceptable result.

[0105] Subsequently, the results of acceptable and unacceptable responsibilities are merged according to the machine tool identifier to generate a responsibility acceptance result. The responsibility acceptance result includes at least the machine tool identifier, acceptance status, action type, action sequence, and responsibility attribution fields. The output is the responsibility acceptance result, which is written to the S6 call area for subsequent feedback and updates. The handling of exceptions or omissions is as follows: if the machine tool identifier is missing from the action result, the writing is not performed; if the action sequence is missing from the action result, the action sequence is filled in according to the node order in the target management path; if the acceptable and unacceptable responsibilities result fall on the same machine tool object at the same time, the determination result of whether the target management path is closed is used as the retention basis, and only the writing result consistent with the closure determination is retained.

[0106] Through the above steps, the derivation results of the target management path can be converted into clear responsibility acceptance status and execution actions. This ensures that whether a machine tool continues to accept the current processing task is no longer determined at the path level, but is further solidified into executable action results and rewritable responsibility acceptance results. In practical applications: After S4 calculation, the target management path corresponding to a certain five-axis CNC machine tool includes production and manufacturing stage nodes, installation and commissioning stage nodes, and operation and maintenance service stage nodes. After the installation and commissioning stage, a maintenance action node is inserted. In this implementation method, the continuity of the three types of sequences is first checked. If they are continuous, the machine tool is marked as a candidate machine tool that can accept responsibility, and target management actions are generated according to the processing execution node, maintenance insertion node, and stage switching node in the path. If another similar machine tool does not form a closed target management path, and the number of delivery closure segments is interrupted after the last closing node, the machine tool is marked as a candidate machine tool that cannot accept responsibility, and task transfer actions are generated according to the delivery break type. Finally, the two types of action results are written into the responsibility acceptable result and the responsibility unacceptable result, respectively, to form the responsibility acceptance result, which is used for subsequent updates of the digital twin responsibility status diagram.

[0107] S6. Read the processing feedback data, maintenance feedback data and task fulfillment data after the target management action is executed, calculate the result deviation of the corresponding stage state node, the transmission deviation of the corresponding stage inherited edge and the adaptation deviation of the task responsibility constraint set in the digital twin responsibility state diagram, perform recursive correction and continuous convergence update on the digital twin responsibility state diagram, and output the updated digital twin responsibility state diagram and digital twin management baseline.

[0108] In this implementation, S6 is used to write back the feedback results after the target management action is executed to the digital twin responsibility state graph, and to form a new digital twin management baseline when responsibility closure is achieved. The principle is as follows: First, processing feedback data, maintenance feedback data, and task fulfillment data are mapped to stage state nodes, stage inheritance edges, and task responsibility constraint sets, respectively, forming deviation association results at the node, edge, and constraint levels. Then, parameter write-back and boundary updates are performed according to the three types of deviations until the graph structure update results converge. Finally, responsibility closure verification is performed on the updated graph structure, and the node parameters, edge parameters, and constraint boundaries that pass the verification are written into the digital twin management baseline for subsequent responsibility offset solving and path generation. This implementation process includes the following steps:

[0109] The purpose of S6-1 is to reattach the actual feedback results after execution to the digital twin responsibility state diagram and solve the deviation association set required for subsequent write-back. The inputs are processing feedback data, maintenance feedback data, task fulfillment data, digital twin responsibility state diagram, and task responsibility constraint set. The processing actions are as follows: First, map the processing feedback data to the corresponding stage state nodes in the digital twin responsibility state diagram according to the equipment identifier, process identifier, and time identifier. The processing feedback data includes at least the processing result, actual accuracy result, actual process completion result, and actual occupied time. Then, map the maintenance feedback data to the corresponding stage inheritance edge according to the maintenance object identifier, maintenance stage, and maintenance time. The maintenance feedback data includes at least the maintenance action type, compensation correction amount, and post-maintenance state change. Finally, map the task fulfillment data to the task responsibility constraint set according to the task identifier. The task fulfillment data includes at least the task delivery result, delivery time, and process fulfillment result.

[0110] Subsequently, result deviation, transmission deviation, and adaptation deviation are calculated separately: result deviation is generated by comparing the target result field recorded in the stage status node with the actual result field in the processing feedback data item by item; transmission deviation is generated by comparing the offset transmission field recorded in the stage inherited edge with the post-maintenance status change field in the maintenance feedback data item by item; adaptation deviation is generated by comparing the accuracy requirements, process requirements, and delivery requirements recorded in the task responsibility constraint set with the actual fulfillment field in the task fulfillment data item by item; the output is a deviation association set, which includes at least the stage status node identifier, result deviation field group, stage inherited edge identifier, transmission deviation field group, task responsibility constraint identifier, and adaptation deviation field group, and is written to the write-back buffer for S6-2 to read; the abnormal or missing handling is as follows: when the processing feedback data is missing the actual accuracy result, the corresponding result deviation is not calculated and a missing field mark is written; when the maintenance feedback data cannot be mapped to the stage inherited edge, the adjacent stage inherited edge is read back according to the maintenance stage for secondary mapping, and if it still cannot be mapped, it does not participate in the current round of edge parameter write-back; when the task fulfillment data is missing the delivery time point, it is filled in according to the task completion record time and a fill-in mark is written;

[0111] The purpose of S6-2 is to perform write-back updates on node parameters, edge parameters, and constraint boundaries based on the deviation association set, and to form a converged digital twin responsibility state diagram. The inputs are the deviation association set and the digital twin responsibility state diagram. The processing actions are as follows: First, write back the node parameters of each stage state node according to the result deviation. The write-back objects include the capability boundary field, responsibility occupancy field, and result record field in the stage state node. The write-back method is to write the result deviation item by item into the corresponding field and replace the original field value with the written-back field value. Next, write back the edge parameters of each stage inherited edge according to the transmission deviation. The write-back objects include the offset field, direction field, and inheritance relationship field in the stage inherited edge. The write-back method is to write the transmission deviation item by item into the corresponding edge parameter. Finally, write back the constraint boundaries of the task responsibility constraint set according to the adaptation deviation. The write-back objects include the accuracy requirement boundary, process requirement boundary, and delivery requirement boundary. The write-back method is to write the adaptation deviation into each constraint boundary field.

[0112] After completing one round of write-back, all stage state nodes are sorted by result deviation, all stage inherited edges are sorted by propagation deviation, and all task responsibility constraints are sorted by adaptation deviation. The consistency of the node deviation sorting, edge deviation sorting, and constraint deviation sorting after the previous two rounds of write-back is used as a stopping condition. When the stopping condition is met, the updated digital twin responsibility state diagram is output. The output is the updated digital twin responsibility state diagram, written to the responsibility closure verification area for S6-3 to read. Anomaly or missing data handling is as follows: if a stage state node lacks a corresponding result deviation field, only the existing field is written back; if a stage inherited edge or a task responsibility constraint has no deviation data in this round, the parameters from the previous round remain unchanged; if the same object fails to write back in two consecutive rounds, the object is written to the object to be reviewed table and removed from the current round's sorting.

[0113] The purpose of S6-3 is to filter out node parameters, edge parameters, and constraint boundaries that meet the responsibility closure conditions from the updated digital twin responsibility state diagram, forming a reusable digital twin management baseline. The input is the updated digital twin responsibility state diagram. The processing actions are as follows: first, read the updated stage state nodes, stage inheritance edges, and task responsibility constraint sets, and perform responsibility closure verification on the three. Specifically, the stage state nodes are used to check whether the capability boundary field, responsibility occupancy field, and result record field are consistent. The stage inheritance edges are used to check whether the offset field, direction field, and inheritance relationship field maintain a continuous transmission relationship with the previous and subsequent stage state nodes. The task responsibility constraint sets are used to check whether the accuracy requirement boundary, process requirement boundary, and delivery requirement boundary are consistent with the corresponding stage state nodes and stage inheritance edges.

[0114] When there is a consistent correspondence between the stage state node, stage inheritance edge, and task responsibility constraint set, the responsibility closure is deemed valid. When any level has a broken field, mismatched boundaries, or inconsistent transmission directions, the responsibility closure is deemed invalid. For node parameters, edge parameters, and constraint boundaries of valid responsibility closure, they are written into the digital twin management baseline according to device identifier, stage identifier, and task identifier. For objects of invalid responsibility closure, they are retained in the updated digital twin responsibility state diagram but not written into the digital twin management baseline. The output is the digital twin management baseline and is written into the subsequent S2 and S3 call areas. The handling of anomalies or missing items is as follows: If a stage state node lacks an adjacent stage inheritance edge during responsibility closure verification, the node is recorded as a broken node and is not written into the digital twin management baseline. If the task responsibility constraint boundary is incomplete, the corresponding task identifier is written into the baseline supplement table and written again after the subsequent task fulfillment data is completed.

[0115] Through the above steps, the actual feedback results after processing execution can be continuously written back to the digital twin responsibility status diagram, so that the digital twin responsibility status diagram no longer remains at the initial modeling result, but forms a digital twin management baseline that can be recursively updated; subsequent responsibility offset results, candidate management path sets, and target management paths are all calculated based on this updated baseline, thereby ensuring that the full lifecycle management results are synchronously corrected with the actual execution process; in practical applications: after a five-axis CNC machine tool completes a high-precision impeller machining, the system maps the actual machining accuracy result and the actual process completion result to the installation and commissioning phase and the operation and maintenance service phase. The corresponding stage status nodes map the compensation correction amount after maintenance to the stage inheritance edge between the installation and commissioning stage and the operation and maintenance service stage, and map the task delivery result on schedule to the task responsibility constraint set corresponding to the current task; then solve the result deviation, transmission deviation and adaptation deviation, and write back the node parameters, edge parameters and constraint boundaries respectively; when the sorting results of the previous two rounds are consistent, output the updated digital twin responsibility status diagram; then write the node parameters, edge parameters and constraint boundaries of the responsibility closure into the digital twin management baseline, so that the machine tool can directly call the updated baseline when solving the responsibility offset of similar tasks in the future.

[0116] Furthermore, the present invention also includes a five-axis CNC machine tool full lifecycle management system, the system comprising a map construction module, a solution module, a path generation module, an optimization module, a judgment module, and a baseline update module:

[0117] The graph construction module is used to acquire the structural parameters, assembly deviations, compensation records, machining results, and maintenance records of the target five-axis CNC machine tool during the design and development, production and manufacturing, installation and commissioning, and operation and maintenance service stages, as well as the accuracy requirements, process requirements, and delivery requirements corresponding to the machining tasks to be performed. It performs time alignment, object alignment, and stage attribution processing on the data of each stage, constructs a digital twin responsibility state graph, and outputs the stage state node set, stage inheritance edge set, and task responsibility constraint set.

[0118] The solver module is used to read the current running data and the current processing context data, map the current running data to the corresponding stage state node in the digital twin responsibility state diagram, calculate the state difference between the current running data and each stage state node, and perform state propagation calculation and historical offset inversion calculation along the stage inheritance edge set, and output the responsibility offset result.

[0119] The path generation module is used to map the task responsibility constraint set to the digital twin responsibility status diagram based on the responsibility offset result, generate a continuation path for each stage status node, generate a downgraded succession path and a maintenance succession path for nodes with excessive responsibility offset, generate a task transfer path for nodes that do not meet the delivery requirements, and output a candidate management path set.

[0120] The optimization module is used to perform constraint-driven adversarial simulation on each candidate management path in the candidate management path set, calculate the corresponding task completion results, accuracy mismatch results and resource consumption results, and perform multi-objective constraint screening and path advantage ranking to solve the candidate management path with the first place in the ranking as the target management path.

[0121] The determination module is used to identify the target five-axis CNC machine tool as a machine tool that can assume responsibility and output target management actions when the target management path is solved; when the target management path is not solved, the target five-axis CNC machine tool is identified as a machine tool that cannot assume responsibility and output maintenance intervention actions or task transfer actions, generating responsibility assumption results.

[0122] The baseline update module is used to read the processing feedback data, maintenance feedback data and task fulfillment data after the execution of target management actions, calculate the result deviation of the corresponding stage state node, the transmission deviation of the corresponding stage inherited edge and the adaptation deviation of the task responsibility constraint set in the digital twin responsibility state diagram, perform recursive correction and continuous convergence update on the digital twin responsibility state diagram, and output the updated digital twin responsibility state diagram and digital twin management baseline.

[0123] Working principle: This solution first unifies the data generated during the design, manufacturing, debugging, and maintenance stages of a five-axis CNC machine tool, establishing a digital twin responsibility status diagram corresponding to this machine tool. Then, it incorporates the current task requirements and the machine tool's real-time status to determine which responsibility offsets the machine tool has made relative to the historical baseline. Next, based on the offsets, it generates multiple processing paths, such as continuing machining directly, performing maintenance before machining, reducing the responsibility level, or transferring the task to another machine tool. The system then analyzes each of these paths, selects the most suitable one, outputs the corresponding action, and writes the execution result back to the digital twin responsibility status diagram as the new baseline for the next judgment.

[0124] For example, a five-axis CNC machine tool is machining a high-precision impeller. The system first looks at the machine tool's previous assembly deviations, compensation records, maintenance records, and machining results. Then, it combines this with the current spindle load, thermal deformation, and cutting status to determine whether it can stably complete this task. If it can complete the task, it continues machining. If it can be restored after one maintenance, it performs maintenance first and then continues machining. If it is no longer suitable to continue undertaking this type of task, the task is switched to another machine tool. In this way, the system does not only look at whether the machine tool can move, but also judges whether it can complete the task according to the current requirements.

[0125] The above description is merely 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 managing the entire lifecycle of a five-axis CNC machine tool, characterized in that, include: S1. Obtain the structural parameters, assembly deviations, compensation records, machining results, and maintenance records of the target five-axis CNC machine tool during the design and development, production and manufacturing, installation and commissioning, and operation and maintenance service stages, as well as the accuracy requirements, process requirements, and delivery requirements corresponding to the machining tasks to be performed. Perform time alignment, object alignment, and stage attribution processing on the data of each stage, construct a digital twin responsibility state diagram, and output the stage state node set, stage inheritance edge set, and task responsibility constraint set. S2. Read the current running data and the current processing context data, map the current running data to the corresponding stage state node in the digital twin responsibility state diagram, calculate the state difference between the current running data and each stage state node, and perform state propagation calculation and historical offset inversion calculation along the stage inheritance edge set, and output the responsibility offset result. S3. Based on the responsibility offset results, map the task responsibility constraint set to the digital twin responsibility state diagram, generate a continuation path for each stage state node, generate a downgraded acceptance path and a maintenance acceptance path for nodes with excessive responsibility offset, generate a task transfer path for nodes that do not meet delivery requirements, and output a candidate management path set. S3 includes: S3-1. Based on the responsibility offset results, task responsibility constraint set and digital twin responsibility state diagram, write the accuracy requirements, process requirements and delivery requirements in the task responsibility constraint set into the state nodes of each stage respectively, calculate the responsibility satisfaction value, delivery satisfaction value and offset occupancy value of each stage state node for the current processing task, and output the node constraint evaluation set. S3-2. For the node constraint evaluation set, generate a continuing path for stage state nodes where both responsibility satisfaction value and delivery satisfaction value are met, generate a downgraded path and a maintenance path for stage state nodes where responsibility satisfaction value is not met but delivery satisfaction value is met, generate a task transfer path for stage state nodes where delivery satisfaction value is not met, and connect the corresponding stage state nodes according to the stage inheritance edge set, and output the path candidate set. S3-3. Calculate the path responsibility closure value, path delivery closure value, and path offset transfer value for each path in the path candidate set, eliminate paths with phase breaks or responsibility reversals, and write the remaining paths into the candidate management path set. S4. Perform constraint-driven adversarial simulation on each candidate management path in the candidate management path set, calculate the corresponding task completion results, accuracy mismatch results and resource consumption results, and perform multi-objective constraint screening and path advantage ranking to solve the candidate management path with the first-ranked result as the target management path. S4 includes: S4-1. Obtain the candidate management path set, digital twin responsibility state diagram and task responsibility constraint set. Write the current processing task into the stage state nodes in each candidate management path one by one, write the corresponding responsibility offset into each stage inheritance edge one by one, perform forward deduction along the nodes of each candidate management path in sequence, and perform reverse perturbation along the edges of each candidate management path in sequence to generate the adversarial deduction branch set corresponding to each candidate management path. S4-2. Calculate the number of process completion segments, delivery closure segments, and responsibility maintenance segments after forward deduction to form the task completion result. Calculate the precision offset increment and responsibility offset increment of each stage state node before and after reverse disturbance to form the precision mismatch result. Calculate the maintenance insertion number, task switching number, and stage occupation time corresponding to each candidate management path to form the resource occupation result. Output the path deduction result set. S4-3. Perform constraint screening on each candidate management path in the path deduction result set, and eliminate candidate management paths with delivery breaks in task completion results, candidate management paths with responsibility reversals in accuracy mismatch results, and candidate management paths with stage conflicts in resource usage results. Then, sort the remaining candidate management paths in the order of task completion results, accuracy mismatch results, and resource usage results in rounds. When the first position of the sorting is consistent in the previous and next rounds, the candidate management path with the first position is determined as the target management path. S5. When the target management path is solved, the target five-axis CNC machine tool is identified as a machine tool that can accept responsibility and the target management action is output; when the target management path is not solved, the target five-axis CNC machine tool is identified as a machine tool that cannot accept responsibility and the maintenance intervention action or task transfer action is output, generating the responsibility acceptance result. S6. Read the processing feedback data, maintenance feedback data and task fulfillment data after the execution of the target management action, calculate the result deviation of the corresponding stage state node, the transmission deviation of the corresponding stage inherited edge and the adaptation deviation of the task responsibility constraint set in the digital twin responsibility state diagram, perform recursive correction and continuous convergence update on the digital twin responsibility state diagram, and output the updated digital twin responsibility state diagram and digital twin management baseline.

2. The method for full lifecycle management of five-axis CNC machine tools according to claim 1, characterized in that: S1 includes: S1-1: Read the equipment identification field, component identification field, process identification field, and time identification field from the structural parameters, assembly deviations, compensation records, processing results, and maintenance records. Perform primary key merging on fields with the same name across stages and mapping merging on fields with different names across stages. Rearrange the data according to the order of occurrence of the design and development stage, production and manufacturing stage, installation and commissioning stage, and operation and maintenance service stage, and output a unified data unit set for each stage. S1-2. For the unified data unit set of the stage, extract the stable field representing the inherent capability, the offset field representing the stage change, and the constraint field representing the responsibility boundary. Generate stage state nodes from the stable field, generate stage inheritance edges from the offset fields of the same object in adjacent stages, and map the accuracy requirements, process requirements, and delivery requirements into task responsibility constraint edges, and output the initial structure of the digital twin responsibility state diagram. S1-3. Perform closure verification on the initial structure of the digital twin responsibility state graph, identify missing stage state nodes, broken stage inheritance edges, and unmapped task responsibility constraint edges, and write the verified stage state nodes, stage inheritance edges, and task responsibility constraint edges into the stage state node set, stage inheritance edge set, and task responsibility constraint set, respectively, and output the digital twin responsibility state graph.

3. The method for full lifecycle management of five-axis CNC machine tools according to claim 2, characterized in that: S2 includes: S2-1. Read the current running data and the current processing context data, perform context segmentation on the current running data, extract state features from each segment, and write the current processing context data into the corresponding state features to form the current state feature set. Compare the field differences before and after writing and remove mismatched fields that do not correspond to the responsibility level. S2-2. Map the current state feature set one by one to the stage state nodes in the digital twin responsibility state diagram, calculate the node difference between each stage state node and the current state feature set, perform node-by-node screening according to the node difference, process correspondence and responsibility level correspondence, retain the stage state nodes whose screening results have not changed, and generate the target stage state node and node difference set.

4. The method for full lifecycle management of five-axis CNC machine tools according to claim 3, characterized in that: S2 also includes: S2-3. Taking the target stage state node and node difference set as input, the node difference is passed along the stage inheritance edge set one by one. The change in difference between the two ends of each stage inheritance edge and the result of consistent direction are calculated. The edge passing result is updated according to the result of consistent change in difference and direction. Stop when the update results of the two rounds are consistent. Output the stage propagation offset set and convergence edge set. S2-4. Taking the stage propagation offset set and convergence edge set as input, trace back the offset source stage by stage along the reverse path of the stage inheritance edge set, calculate the contribution value of each stage to the current offset, perform round-by-round resolution according to the contribution value and stage attribution conflict relationship, stop when the stage attribution results of the previous and next rounds are consistent, and output the stage offset contribution set. S2-5. Perform consistency checks on the stage offset contribution set, stage propagation offset set, and convergence edge set. Identify conflict items where the stage offset contribution direction is inconsistent with the stage propagation direction. Perform conflict resolution according to the responsibility level correspondence and stage attribution order, and output the responsibility offset result.

5. The method for full lifecycle management of five-axis CNC machine tools according to claim 4, characterized in that: S5 includes: S5-1. Perform a closure check on the order of stage state nodes, the order of stage inheritance edges, and the order of path actions in the target management path. Mark the machine tool corresponding to the target management path that passes the check as a candidate machine tool that can take on responsibility, and mark the machine tool that has not formed a target management path as a candidate machine tool that cannot take on responsibility. Output the intermediate result of the acceptance judgment. S5-2. For the intermediate results of the acceptance judgment, extract the machining execution nodes, maintenance insertion nodes and path switching nodes corresponding to the target management path for candidate machine tools whose responsibilities can be accepted, and generate target management actions in the order of node appearance; for candidate machine tools whose responsibilities cannot be accepted, extract the fracture position and fracture type after the last closed node in the candidate management path set, generate maintenance intervention actions or task transfer actions according to the fracture type, and output the action result set. S5-3. Write the responsibility attribution for each action result in the action result set, write the target management action and the candidate machine tool for responsibility acceptance into the responsibility acceptance result, write the maintenance intervention action or task transfer action and the candidate machine tool for responsibility non-acceptance into the responsibility non-acceptance result, and generate the responsibility acceptance result.

6. The method for full lifecycle management of five-axis CNC machine tools according to claim 5, characterized in that: S6 includes: S6-1. Map the processing feedback data to the corresponding stage state node in the digital twin responsibility state diagram, map the maintenance feedback data to the corresponding stage inheritance edge, map the task fulfillment data to the task responsibility constraint set, calculate the result deviation of each stage state node, the transmission deviation of each stage inheritance edge and the adaptation deviation of the task responsibility constraint set, and output the deviation association set. S6-2. Write back the node parameters for each stage state node according to the result deviation, write back the edge parameters for each stage inherited edge according to the transmission deviation, and write back the constraint boundary for the task responsibility constraint set according to the adaptation deviation. The stop condition is that the node deviation sorting, edge deviation sorting and constraint deviation sorting after the two rounds of writing back are consistent. Output the updated digital twin responsibility state diagram. S6-3. Read the updated digital twin responsibility state diagram, perform responsibility closure verification on the updated state nodes of each stage, the inherited edges of each stage, and the task responsibility constraint set, write the node parameters, edge parameters, and constraint boundaries that pass the verification into the digital twin management baseline, and output the digital twin management baseline.

7. A five-axis CNC machine tool lifecycle management system, used to implement the five-axis CNC machine tool lifecycle management method according to any one of claims 1-6, the system comprising a map construction module, a solution module, a path generation module, an optimization module, a judgment module, and a baseline update module, characterized in that: The graph construction module is used to acquire the structural parameters, assembly deviations, compensation records, machining results, and maintenance records of the target five-axis CNC machine tool during the design and development, production and manufacturing, installation and commissioning, and operation and maintenance service stages, as well as the accuracy requirements, process requirements, and delivery requirements corresponding to the machining tasks to be performed. It performs time alignment, object alignment, and stage attribution processing on the data of each stage, constructs a digital twin responsibility state graph, and outputs the stage state node set, stage inheritance edge set, and task responsibility constraint set. The solver module is used to read the current running data and the current processing context data, map the current running data to the corresponding stage state node in the digital twin responsibility state diagram, calculate the state difference between the current running data and each stage state node, and perform state propagation calculation and historical offset inversion calculation along the stage inheritance edge set, and output the responsibility offset result. The path generation module is used to map the task responsibility constraint set to the digital twin responsibility status diagram based on the responsibility offset result, generate a continuation path for each stage status node, generate a downgraded succession path and a maintenance succession path for nodes with excessive responsibility offset, generate a task transfer path for nodes that do not meet the delivery requirements, and output a candidate management path set. The optimization module is used to perform constraint-driven adversarial simulation on each candidate management path in the candidate management path set, calculate the corresponding task completion results, accuracy mismatch results and resource consumption results, and perform multi-objective constraint screening and path advantage ranking to solve the candidate management path with the first place in the ranking as the target management path. The determination module is used to identify the target five-axis CNC machine tool as a machine tool that can assume responsibility and output target management actions when the target management path is solved; when the target management path is not solved, the target five-axis CNC machine tool is identified as a machine tool that cannot assume responsibility and output maintenance intervention actions or task transfer actions, generating responsibility assumption results. The baseline update module is used to read the processing feedback data, maintenance feedback data and task fulfillment data after the execution of target management actions, calculate the result deviation of the corresponding stage state node, the transmission deviation of the corresponding stage inherited edge and the adaptation deviation of the task responsibility constraint set in the digital twin responsibility state diagram, perform recursive correction and continuous convergence update on the digital twin responsibility state diagram, and output the updated digital twin responsibility state diagram and digital twin management baseline.

Images