An online calibration and closed-loop iteration method for ring die processing parameters

Abstract

The application relates to the technical field of numerical control machining process control, and discloses an online calibration and closed-loop iteration method for ring mold machining parameters. The method is used for solving the problems that the online closed-loop iteration of ring mold machining parameters in the traditional method depends on experience and lacks convergence criteria and step safety constraints. Firstly, the task is registered to lock the machining datum, measure calibration, partition table and boundary threshold caliber; in the machining, the load, vibration, cooling and tool withdrawal state are collected and are assigned with time identifiers and credibility, and the state section is cut based on the anchor point; only a single main variable is selected in the partition, and the variable is issued and read back according to the step interval and slope constraint and is confirmed; the window is determined according to the hard and soft boundary, and if protection or callback is triggered, the window is frozen and rechecked; after the convergence criterion is met, the version is solidified and the evidence index is archived, and the parameter traceable convergence is realized.

Description

Technical Field

[0001] This invention relates to the field of CNC machining process control technology, specifically to an online calibration and closed-loop iteration method for ring die machining parameters. Background Technology

[0002] The ring die is a key component of pelleting equipment. Its geometric quality, such as the die hole diameter, hole spacing, taper, and inlet morphology, directly affects pelleting efficiency and product consistency. Currently, ring die hole machining is mostly completed by CNC drilling or deep hole machining combined with processes such as reaming and boring. During the machining process, it is usually necessary to adjust machining parameters such as spindle speed, feed rate, and cooling flow rate according to material batch, tool wear, cooling and lubrication status, and changes in machine tool temperature to avoid problems such as overload tool breakage, hole wall burning, and hole diameter fluctuation.

[0003] To improve detection and quality control capabilities, existing technologies have developed detection schemes for ring die hole parameters. For example, the published invention patent application CN110006324B discloses a method and apparatus for detecting ring die hole parameters. By collecting information on the die hole array, it detects and statistically analyzes parameters such as hole spacing, row spacing, and number of holes, thereby improving the efficiency and consistency of die hole parameter acquisition. However, this type of scheme focuses on detecting and judging die hole parameters and is usually used as a means of quality inspection after processing or process sampling. It lacks a mechanism to establish a closed-loop relationship between the detection results and processing control variables. In particular, it lacks parameter correction logic, update step size constraints, and convergence criteria for the processing process, making it difficult to support online calibration and iterative optimization of processing parameters.

[0004] Meanwhile, to reduce the impact of errors during the machining process, some existing technologies have proposed machine tool error compensation and real-time modeling methods. For example, the published invention patent application CN102736558A discloses a CNC machine tool thermal error real-time compensation modeling method based on time series algorithm. By modeling and predicting thermal error data, compensation control can be achieved, which can reduce the impact of thermal drift on machining accuracy to a certain extent. However, this type of method is mainly aimed at compensation modeling for a single error source. It usually does not take into account multi-source disturbances such as tool wear, chip removal status, and cooling and lubrication changes in ring die deep hole machining. It also lacks a closed-loop control strategy that maps process monitoring signals with hole quality indicators for parameter iterative updates.

[0005] In actual ring die machining, online parameter adjustment still largely relies on experience-based trial and error. The adjustment basis often comes from the operator's subjective judgment of vibration, current, chip condition, or sampling results. However, due to the lack of verifiable convergence criteria and update step size constraints, parameter adjustment may experience repeated oscillations or overcorrection, making it difficult for closed-loop iteration to converge stably. This can lead to spindle overload, tool breakage, or hole quality fluctuations such as diameter taper, affecting batch consistency and production cycle time. Therefore, there is an urgent need for a technical solution that can establish a closed-loop iterative control mechanism for machining parameters based on online monitoring and measurement feedback, and set convergence criteria and safety boundary constraints to ensure that parameters such as speed, feed, and cooling are stably updated within a safe range and maintain consistent hole quality. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides an online calibration and closed-loop iteration method for ring die machining parameters, which solves the problems of traditional methods where online closed-loop iteration of ring die machining parameters relies on experience and lacks convergence criteria and step size safety constraints.

[0007] To achieve the goal mentioned in the background art of ensuring safe and stable convergence of ring die machining parameters and guaranteeing consistent die hole quality, the present invention provides the following technical solution:

[0008] An online calibration and closed-loop iteration method for ring die machining parameters includes:

[0009] S1: Establish a machining datum version, complete clamping datum alignment and measurement consistency calibration, generate task identifiers and partition tables, and load partition initial parameters and boundary rules;

[0010] S2: During machining execution, collect load, vibration, cooling supply and tool retraction status, write time stamps and reliability markers to the collected data, and generate status records;

[0011] S3: Enter the iterative control process by partition, determine the single main variable, generate parameter change instructions based on the upper limit of step size and the minimum adjustment interval, and issue them according to the slope limit and execute readback confirmation;

[0012] S4: Determine the status based on boundary rules within the confirmation window and the stable window. Freeze the upgrade and execute the review process when a downgrade callback or protection action is triggered.

[0013] S5: When the convergence criteria are met, solidify the partition convergence version and archive parameter change records, window summaries, and sampling records.

[0014] In a preferred embodiment, establishing a processing baseline version includes:

[0015] Before starting the machine, generate a task identifier and establish a task registration record, and register the processing baseline version;

[0016] Before clamping, check the cleanliness and contact status of the positioning surface. After alignment and locking, verify the posture stability and fix the machining coordinates for tool setting.

[0017] The machining reference version is rebuilt and the valid version is updated when the clamping is reset, the fixture is reinstalled, the tool is replaced, the equipment is restored from abnormality, or the task is interrupted and restarted.

[0018] In a preferred embodiment, the clamping datum alignment and measurement consistency calibration are completed, a task identifier and partition table are generated, and initial partition parameters and boundary rules are loaded, including:

[0019] After locking the reference object set, the measuring device zero point is reset and the range is confirmed, multi-directional repeated sampling is performed. The consistency is judged according to the calibration rules of the threshold parameter set. If it fails, the fixed point is reset and the installation and communication are checked and retested. A measurement calibration version is generated and bound to the processing reference version.

[0020] Create a partition table to record the set of sampling locations and the number of partition switching interlock entries, load the initial version of the partition parameter group and the version of the boundary rule, and write them into the reference locking rule.

[0021] In a preferred embodiment, load, vibration, cooling supply, and tool retraction status are collected during machining execution. The collected data is then time-stamped and confidence-tagged, and a status record is generated, including:

[0022] When processing starts, write the time identifier and partition start identifier, and verify the consistency of the processing reference version, measurement calibration version, boundary rule version and threshold parameter set version;

[0023] Collect load, vibration, cooling supply and tool retraction status and write them into time stamps and confidence markers, and divide the status into segments according to event anchor points;

[0024] Anomalies are generated by confidence gating and repetition gating. If the partition interlock is not satisfied, the collected segment is marked as a verification segment and the upward adjustment entry is closed.

[0025] In a preferred embodiment, the iterative control process is initiated by partition, and a single dominant variable is determined, including:

[0026] When entering a partition, write the partition start event anchor point, lock the referenced version and generate an iteration number based on the task registration record, verify the entry condition according to the interlocked entries, and if it fails, set the iteration status to pause and close the up-entry entry.

[0027] After passing, a single dominant variable is determined based on the priority entry, and a locking flag is applied to the remaining variables. When a protection entry is triggered, downscaling actions are allowed on non-dominant variables, and the trigger entry number is recorded.

[0028] In a preferred embodiment, a parameter change instruction is generated based on the upper limit of the step size and the minimum adjustment interval, and then issued according to the slope limit and a readback confirmation is performed, including:

[0029] Based on the unlocking conditions of the upper limit of step size and the minimum adjustment interval, a parameter change instruction is generated and the referenced entry number is recorded. After a preliminary check, the instruction is issued according to the slope transition and written to the transition start and end anchor points and the parameter effective anchor points.

[0030] If the double read confirmation fails, write the freeze reason identifier, and verify the interface status, machine tool lock status and parameter interlock conditions according to the review items, and perform re-issuance, conservative maintenance or version rollback.

[0031] In a preferred embodiment, state determination is performed based on boundary rules within the confirmation window and the stable window, including:

[0032] After the parameter readback is confirmed, the confirmation window and the stabilization window are started with the parameter effective event anchor point;

[0033] Based on the version of the boundary rules and the version of the threshold parameter set associated with them, the state segments that meet the credibility requirements are selected from the state log, and the hard boundary entries and soft boundary entries are matched and the trigger entry number is recorded.

[0034] Once the window is confirmed to be passed and not frozen, perform the sampling inspection according to the sampling interlock item, and generate a window summary to write to the window log.

[0035] In a preferred embodiment, upon triggering a downgrade callback or protection action, the upgrade process is frozen and a review process is executed, including:

[0036] When a hard boundary entry is triggered, the protection action sequence bound to the entry is frozen and processed, and written to the protection event log.

[0037] When a soft boundary entry reaches the restriction level, it is frozen and the main variable is reversed. After slope transition and readback confirmation, the validity, credibility, transition end and communication readback status of the calibration are verified according to the review entry, and the freeze is lifted or maintained according to the verification conclusion.

[0038] In a preferred embodiment, the partition convergence version is solidified when the convergence criterion is met, including:

[0039] When the convergence criterion is met, version consistency verification is performed based on the window summary, sampling record and parameter change record of the continuous and stable window chain, and gating is performed according to the confidence level.

[0040] When both the sampling consistency entry and the parameter static stability entry are satisfied, the convergence anchor point is written, a version freeze lock is applied, and the current parameter version is solidified as the partition convergence version.

[0041] In a preferred embodiment, the archived parameter change record, window summary, and sampling inspection record include:

[0042] After the solidified partition convergence version is completed, the parameter change records, window summaries and sampling inspection records are archived in chronological order to generate the evidence index version. The process package is then sealed by establishing a reference relationship with the status log, window log and sampling inspection log. Only appending is allowed to the sealed content.

[0043] When reusing across tasks, the converged version is loaded after the reuse conditions are locked according to the task registration and the verification is passed again.

[0044] Compared with the prior art, the present invention provides an online calibration and closed-loop iteration method for ring die machining parameters, which has the following beneficial effects:

[0045] 1. This invention, by uniformly locking the machining datum version, measurement calibration version, partition table version, and boundary rules and threshold calibers through task registration records during the machine-on stage, avoids the difficulty in tracing coordinate and measurement caliber drifts caused by clamping reset, tool replacement, or equipment recovery. During machining, time stamps and reliability markers are written for load, vibration, cooling supply, and tool retraction status, and status segments are divided by event anchor points and abnormal segment numbers are marked, enabling multi-source disturbances and data loss to be identified and eliminated according to rules. In partition-level iteration, only a single dominant variable is selected, subject to step size upper limit and minimum adjustment interval. The method incorporates slope transition constraints and suppresses parameter trial-tuning oscillations through a double-readback confirmation and verification process. Within the confirmation window and stable window, it makes judgments based on hard and soft boundary entries. When protection or downgrade callbacks are triggered, the upward adjustment is frozen and the sampling verification is tightened. After meeting the convergence criteria, the convergence version is solidified and the evidence index is archived to support cross-task reuse and re-verification. This transforms the ring die hole machining parameters from experience-based trial-tuning to traceable, controllable, and convergent online closed-loop iterative control, solving the problems of traditional methods where online closed-loop iteration of ring die machining parameters relies on experience and lacks convergence criteria and step size safety constraints.

[0046] 2. This invention integrates parameter updates, window determination, spot check verification, and convergence solidification within a single closed-loop chain. It uses entry numbers and evidence indexes to record the reasons for parameter changes, trigger boundaries, readback confirmations, and spot check results, ensuring that each parameter adjustment corresponds to a specific state segment and rule definition. This facilitates the tracing of anomaly sources and the reproduction of the judgment process. After convergence, a version freeze lock is set, and a reuse and re-verification mechanism is introduced to isolate and manage short-term correction versions from the convergence baseline, preventing on-site handling from overriding the baseline process. Furthermore, during cross-task calls, the validity verification of benchmarks and calibrations, credibility gating, and window verification are completed. This improves the reusability and audit consistency of process knowledge accumulation, reduces the risk of process drift and batch fluctuations, and forms a reusable, verifiable, and auditable process package management method. Attached Figure Description

[0047] Figure 1 This is a schematic diagram of the online calibration and closed-loop iteration method for ring die processing parameters according to the present invention. Detailed Implementation

[0048] 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.

[0049] Example 1: Figure 1 An online calibration and closed-loop iteration method for ring die machining parameters is presented, including:

[0050] S1: Establish a machining datum version, complete clamping datum alignment and measurement consistency calibration, generate task identifiers and partition tables, and load initial partition parameters and boundary rules. The specific implementation is as follows:

[0051] Before machining, a task registration record is established and a task identifier is generated. The task registration record uniformly records the machining reference version, measurement calibration version, partition table version, parameter group initial version, boundary rule version, and threshold parameter set version, and is used to limit the version reference relationship of subsequent data and actions. The task identifier is generated with a traceable code, associated with the ring mold batch, tool batch, machine tool number, and process type, and serves as a unified index key for status logs, parameter change logs, window logs, and sampling logs, so as to trace back to the corresponding task and version set.

[0052] During the preparation stage, the ring die is clamped onto a special fixture and locked in the clamping position. The fixture positioning surfaces are checked sequentially, as are the alignment and locking of the ring die reference surfaces. Before clamping, the contact and cleanliness of the positioning surfaces are verified. If contamination or burrs are found, they are cleaned and rechecked before proceeding with the alignment and locking operations. After locking, the stability of the position is verified, including whether the ring die's position swings freely, whether the locked position rebounds, and whether the contact relationship with the reference surfaces is stable. If the verification fails, the locking is released, the clamping reference is realigned, and the locking is repeated, and the verification is repeated. Upon successful verification, the machine tool tool setting and coordinate establishment process begins. The machine tool probes or sets the tool on the reference surface according to the preset tool setting path, determining the machining coordinate origin, reference direction, and attitude reference, and then solidifies these as the machining reference version. The machining reference version is registered and a reconstruction trigger event is recorded. Reconstruction trigger events include clamping reset, fixture reloading, tool replacement, equipment abnormality recovery, and task interruption restart. When a reconstruction event is triggered, the machining reference version is re-established, and the new valid version is written into the task registration record for subsequent steps to reference the then-valid machining coordinate caliber.

[0053] After the machining datum version is established, measurement consistency calibration is performed to ensure that the measurement coordinates and machining coordinates maintain the same caliber and are aligned within the same task. Measurement consistency calibration first determines the set of datum reference objects and locks the reference object types. The datum reference objects are selected from one or more combinations of fixture datum blocks, ring die datum holes, and ring die datum step surfaces to form the reference object set. Once the reference object set is locked, it cannot be changed arbitrarily; any changes trigger recalibration. Subsequently, the zero-point reset and range confirmation of the measuring device are performed in a preset sequence. Zero-point reset restores the internal reference state of the measuring device to its initial state, and range confirmation verifies that the output response of the measuring device is within the normal operating range. After completing zero-point reset and range confirmation, attitude consistency verification is carried out. Attitude consistency verification uses the same reference object as the datum, and... Repeated sampling is performed under the same direction and different path conditions to verify the correspondence between the measurement coordinate direction and the machining coordinate direction, as well as the axial consistency. The direction identifier, path identifier, and time identifier of each sampling are written into the calibration log. The calibration log is used to support the subsequent judgment of the usability of the measurement caliber. At least the sampling sequence and sampling conditions are recorded to facilitate anomaly tracing and location. After repeated sampling, consistency is judged according to the calibration rules. The calibration rules are defined in the threshold parameter set and are fixed in a versioned manner. The calibration rules include the comparison caliber of repeated sampling differences, the judgment caliber of directional consistency, and the judgment caliber of bias stability, and the judgment boundary is limited by the consistency threshold. When there is reverse direction, axial inconsistency, bias drift, or repeated differences exceeding the consistency threshold, the measurement consistency calibration is judged to be unsuccessful.

[0054] When the measurement consistency judgment fails, the calibration correction process is initiated and a correction record is generated. The calibration correction process sequentially executes the measurement device fixed point reset, installation status verification, and interface status verification. The installation status verification is used to confirm the fastening status, installation posture, and reference surface contact relationship. The interface status verification is used to confirm communication continuity and check for abnormal readback delay and data jump abnormality prompts. During the calibration correction process, the reset action, verification conclusion, and abnormal prompts are written to the correction log, and the correction log is associated with the calibration log to form a traceable correction link. After the correction is completed, the measurement consistency verification and consistency judgment are re-executed until the consistency judgment passes. After the consistency judgment passes, a measurement calibration version is generated, and a binding relationship is established between the measurement calibration version and the currently valid machining reference version. The binding relationship is registered in the task registration record so that subsequent steps can synchronously trace the corresponding machining reference version when referencing the measurement calibration version. The measurement calibration version registers reconstruction trigger events, including probe replacement, measurement device displacement, clamping reset, reference object set change, and abnormal disturbance within the window. After a reconstruction event is triggered, the calibration process is re-executed and a new measurement calibration version is generated to maintain consistency between the measurement caliber and the machining caliber.

[0055] After completing the benchmark and calibration registration, a partition table is established so that subsequent closed-loop iterations are organized and executed according to the spatial differences of the ring die hole array. The partition table uses a combination of circumferential sectors and radial rings to divide the partitions, and records the partition range, processing sequence, set of sampling locations, and partition switching interlock conditions for each partition. The set of sampling locations is kept fixed within the task and is used for consistency verification of sampling locations in the subsequent stable window. When the set of sampling locations needs to be adjusted, it is updated according to the rule change process and enters the review process. It can only take effect after passing the review. The partition switching interlock conditions are derived from the interlock entries in the boundary rules. The interlock entries are written into the partition table with the entry number and serve as the basis for the check before switching. Before the partition switch, the pre-state is verified item by item according to the interlock entries. The pre-state includes at least the parameters being in a stable state, the measurement calibration version being in a valid state, the tool retraction action being completed and the machine tool being in a safe stopping posture, and the communication and readback status meeting the reference requirements. After the verification is passed, the next partition is entered, thereby avoiding the overlap of partition switching and parameter switching, which would lead to the confusion of state caliber.

[0056] After the partition table is established, the initial parameters, boundary rules, and threshold parameter sets for the partition are loaded. The initial parameters are read from the process library in the form of initial versions of parameter groups. The process library entries establish a correspondence based on material type, aperture range, pore density level, and process type, and output the initial constraint configurations for rotation speed, feed, cooling supply, and retraction rhythm. The initial constraint configurations are characterized by range, gear, and allowed switching conditions, and serve as the version reference starting point for subsequent parameter change instructions. Boundary rules are recorded as boundary rule versions and bound to the partition. Boundary rules are divided into hard boundary entries, soft boundary entries, interlock entries, and review entry entries, and each entry is assigned an entry number. Hard boundary entries limit the insurmountable safety restrictions and specify the action constraints after triggering. Soft boundary entries specify the warning and degradation trigger conditions. Interlock entries specify the pre-state for change issuance, spot check execution, partition switching, and review initiation. Review entry entries specify the freezing trigger conditions. The threshold parameter set is recorded with different versions and referenced with the boundary rule versions. It is used to unify the threshold criteria for calibration consistency judgment, communication and readback judgment, security boundary judgment, interlock verification, and review entry triggering. The threshold parameter set records the indicator name, applicable scope identifier, comparison criteria, trigger type, continuous conditions, recovery conditions, permission level, and source identifier in structured fields. The source identifier indicates the basis for threshold formation, including equipment or tool safety specifications, process verification records, historical process package stability window evidence, and on-site calibration records. Threshold changes enter the review entry through change registration. After review approval, a new threshold parameter set version is generated and the reference in the task registration record is updated, ensuring that threshold adjustments are auditable and traceable, while also being constrained by permission levels and security boundaries to prevent threshold settings from entering risky areas.

[0057] The locking rules are referenced and written to the task registration record; subsequent log records, parameter changes, and window judgments are only allowed to reference the processing benchmark version and measurement calibration version that are currently bound and in a valid state in the task registration record, and the boundary rule version and threshold parameter set version bound to the partition are required to be consistent with them; the sampling record is only allowed to enter the window judgment when the measurement calibration version is in a valid state; when version inconsistency, version failure, or missing binding relationship occurs, the upward entry is closed and the process is transferred to the review entry; after the review is passed and the new version binding registration is completed, the process execution is resumed.

[0058] S2: During machining execution, load, vibration, cooling supply, and tool retraction status are collected. The collected data is then written with a time stamp and a reliability marker, and a status record is generated. The specific implementation is as follows:

[0059] After the task registration record takes effect, the status acquisition process begins. When processing starts, the processing start time identifier and the current partition start identifier are written, and the currently bound and valid processing benchmark version and measurement calibration version are read from the task registration record. At the same time, the boundary rule version and threshold parameter set version bound to the partition are also read. Then, version consistency verification is performed. If the verification fails, the current acquisition segment is marked as a review segment and the up-adjustment entry is closed. The status log and interlock check record are still continuously written to maintain the chain of evidence. After the new version binding registration is completed and the verification is passed again, the subsequent status segments are marked as candidate available segments.

[0060] During the zonal machining process, load status, vibration status, cooling supply status, and tool retraction status are continuously collected according to task-level acquisition rules. Load status is formed by the spindle load, feed axis load, or equivalent load index read back from the machine tool, and written into the status record according to a unified field caliber to support data alignment from different equipment sources. Vibration status is formed by machine tool monitoring or external sensors, and written into the status record according to a unified sampling caliber, which can be time-aligned with load status. Cooling supply status records the medium opening and closing status, supply level, and abnormal markers. Abnormal markers cover situations such as supply interruption, supply fluctuation, and supply switching transition, and are used to identify stable supply ranges and supply disturbance ranges. Tool retraction status records the start and end of the action, tool retraction segment identifier, and tool retraction reason identifier. The tool retraction reason identifier is mapped to the boundary rule entry number so that the tool retraction trigger basis and corresponding handling path can be traced according to the entry number later.

[0061] Each status record is written with a unified time identifier, which is generated by the machine tool control clock or a unified time source. This time identifier is used to align the sequential relationship of multiple source statuses and determine the start and end intervals of each status segment. When a critical control event occurs, an event anchor point is written. The event anchor point records the event type, event number, and corresponding time identifier, covering events such as partition switching, parameter change taking effect, tool retraction start and end, and verification entry start and end. Status records are associated with adjacent event anchor points in chronological order. The event anchor points are used as boundaries to divide status segments and establish a correspondence between status segments and control events.

[0062] To enable the status records to be used for subsequent iterative control and judgment, a credibility tag is written to each piece of collected data. The credibility tag includes a credibility level and a credibility reason code. The credibility level is divided into a credible state, a restricted state, and an unavailable state. The credibility reason code is used to identify the reason for degradation, which includes insufficient data collection integrity, insufficient communication continuity, handover transition period, missing segments, jump segments, and device start-up / shutdown transition. Data collection integrity is used to determine whether key status fields are complete and meet the continuity requirements within the same time period. If a continuous status segment cannot be formed, the credibility is marked as unavailable. Communication continuity is used to determine whether there is a disconnection, delay accumulation, or readback in the backlink. When a readback delay anomaly occurs and the actual occurrence time cannot be determined, the credibility is marked as limited. The switching transition period is used to identify unstable intervals such as equipment start-up and shutdown, parameter switching ramp segments, and tool retraction return segments. Within these intervals, the credibility is marked as limited and the corresponding event anchor point is bound. Missing segments and jump segments are used to identify sudden changes caused by sensor or interface anomalies. When a sudden change overlaps with a transition event anchor point, the credibility is marked as limited and a review prompt mark is written. When a sudden change occurs repeatedly in a non-transition interval, the credibility is marked as limited and an anomaly segment number is generated. Data corresponding to limited and unavailable states are not included in the upward adjustment basis set, and related records are used for window judgment and review traceability.

[0063] All anomaly detection, credibility degradation detection, and interlock check detection refer to the corresponding threshold entries in the threshold parameter set version locked in the task registration record, and the threshold entry number and source identifier are synchronously written in the status log, anomaly segment number record, and interlock check record. The source identifier is used to indicate the category to which the threshold is based, so that the detection process can be reproduced under the same standard and the version change trajectory of the threshold can be traced.

[0064] After the status record is formed, it is written to the status log according to the partition identifier and time sequence, and a version reference relationship is established with the task registration record. The status log is collected by task identifier and located by partition identifier and time identifier. At the same time, it records the reference information of the processing benchmark version, measurement calibration version, boundary rule version and threshold parameter set version, so that each status segment can be traced to the corresponding caliber and rule basis. In addition to saving instantaneous status points, the status log is also divided according to event anchor points to form status segments, and records the start and end range, duration interval and confidence summary of the status segment. At the same time, a segment type label is written to the status segment. The segment type label includes stable segment, transition segment, abnormal segment and verification segment. The stable segment is used for continuity verification of window determination, the transition segment is used to isolate short-term fluctuations caused by parameter changes and tool retraction transition, the abnormal segment is used to register continuous abnormal characteristics and carry the abnormal segment number, and the verification segment is used to mark the status data generated during the verification entry period.

[0065] Anomaly segments are formed using an itemized gating mechanism. Anomaly spike judgment and repetitive gating judgment refer to spike trigger entries and repetitive trigger entries in the threshold parameter set, respectively. The spike trigger entries define the triggering and recovery conditions for a single mutation, while the repetitive trigger entries define the repetitive triggering criteria and summary conditions within the non-transition interval. Anomaly identification first undergoes credibility gating. When anomaly spikes occur in the load or vibration and the credibility is limited or unavailable, only the anomaly is recorded and a review prompt mark is written, without generating an anomaly segment number. After passing the credibility gating, repetitive gating is then performed. When anomaly spikes meet the repetitive triggering conditions in the non-transition interval and span multiple state segments, the relevant segments are summarized into a continuous anomaly segment and assigned an anomaly segment number. At the same time, the trigger type identifier and its positional relationship relative to the event anchor point are written into the anomaly segment number record, so that subsequent window judgments can continuously verify the anomaly segment according to the same item caliber and keep anomaly identification and action execution independent of each other.

[0066] When switching partitions, an event anchor point is written to the partition switching event, and a switching interlock check is performed. The interlock check is based on the interlock entry number and threshold parameter set caliber bound to the partition, and the check conclusion and corresponding entry number are written to the interlock check record. The interlock check covers conditions such as tool retraction completion status, parameter stability status, confidence level meeting the iteration entry conditions, and measurement calibration version being in a valid state. When the interlock check passes, the first segment of the next partition is marked as a candidate usable segment. When the interlock check fails, the subsequent acquisition segment is marked as a verification segment and handled according to the verification segment handling rules. The candidate usable segment marking is restored after the interlock entry is satisfied. The status log, abnormal segment number record, and interlock check record serve as the input basis for subsequent partition iteration control and window determination.

[0067] S3: Enter the iterative control process by partition, determine the single dominant variable, generate parameter change instructions based on the upper limit of the step size and the minimum adjustment interval, issue them according to the slope limit and perform readback confirmation. The specific implementation is as follows:

[0068] Once processing enters a certain partition, the iterative control process is initiated according to the partition, and the parameter adjustment is organized using the partition as the smallest unit for closed-loop updates. The iterative control references the currently locked and valid processing benchmark version, measurement calibration version, parameter group version, boundary rule version, and threshold parameter set version in the task registration record to generate a partition iterative record and write the partition iterative number. The partition iterative number is used to associate records of entry condition verification, instruction generation, transition issuance, readback confirmation, and review and processing.

[0069] At the start of a partition iteration, the starting event anchor point of that partition is located in the status log. The processing order and sampling location set in the partition table are read, and the set of adjustable variables limited by interlock entries in the boundary rules is also read. Subsequently, entry condition verification is performed. The entry condition verification is conducted item by item in an interlocked manner, and an entry condition verification record is generated. The entry condition verification covers the partition iteration permission status, status log credibility level, verification entry end status, tool retraction completion status, parameter switching non-transitional state status, and measurement calibration version validity status. Each verification item references interlock entries, credibility entries, and transitional state entries, and is specified in the entry condition. The item verification record includes an item number and a judgment conclusion field. The judgment conclusion field includes a trigger type identifier, a continuous condition satisfaction flag, and a recovery condition satisfaction flag to support subsequent review entries in reproducing the verification process using the same criteria. If any verification is not satisfied, the partition iteration control state is set to pause and a pause reason identifier is written. The pause reason identifier is bound to the event anchor point. In the paused state, the upward adjustment entry is closed, and only conservative maintenance instructions or action instructions corresponding to the protection item permitted by the boundary rules are allowed to be executed. After the event anchor point is written at the end of the review, the entry condition re-examination is triggered. After the re-examination is passed, the parameter update stage begins.

[0070] After entering the parameter update phase, a single dominant variable is first determined and variable locking is performed. The dominant variable is determined based on the priority entries in the boundary rule version. The priority entries are used to map the risk type and its state trend to the candidate variables and give the priority order of the candidate variables. The risk type is determined based on the state segment label, abnormal segment number, and retraction reason identifier. After the dominant variable is determined, it is written into the current iteration record, and a variable locking list is generated and written with a locking mark. The locking mark restricts the generation of upward adjustment or modification instructions for other variables besides the dominant variable during the current iteration to avoid unclear causal relationships caused by the superposition of multiple variables. When a protection entry is triggered, downward adjustment or retraction actions are allowed for non-dominant variables according to the protection entry, and an exception reason identifier and trigger entry number are written for each exception action to distinguish the exception action from the normal iteration record.

[0071] After the primary variable is locked, a parameter change instruction is generated. Before generating the instruction, the upper limit of the step size and the minimum adjustment interval are verified. The upper limit of the step size and the minimum adjustment interval are limited by the partition rule entries. The upper limit of the step size is expressed using gear level differences and allowed cross-level rules to limit the range of gears that a single change can cross. The minimum adjustment interval is expressed using unlocking conditions to limit the interval constraints that must be met before adjusting upwards and to verify the completion of verification. The unlocking conditions reference window interlock entries and use the window number as the unlocking credential to write to the parameter change log, so as to form a constraint relationship of alternating progress of change and verification. After the verification is passed, a parameter change instruction is generated and written to the instruction metadata. The instruction metadata includes instruction sequence number, task identifier, partition identifier, iteration number, primary variable type, target gear identifier, change reason identifier, set of referenced entry number, planned effective time point identifier, and associated status segment number or abnormal segment number. The change reason identifier is expressed using rule-based enumeration and is bound to the referenced priority entry number to ensure that the change trigger semantics are reproducible.

[0072] The parameter change command is executed according to the issued control flow, which includes pre-check and slope transition. The pre-check is completed based on the interlock entries and a pre-check record is formed. The pre-check is used to confirm that the equipment is in a switchable state. The switchable state must at least meet the following conditions: not in the retraction execution segment, not in the partition switching transition segment, not in the verification segment, not in the emergency stop recovery transition segment, and the communication continuity meets the threshold entry requirements. The judgments related to communication continuity, readback delay, and readback consistency are written into the pre-check record with the corresponding threshold entry number and source identifier. The source identifier is used to indicate the category to which the threshold is based. After the pre-check is passed, the slope transition begins. The slope transition is executed according to the switching rhythm defined by the boundary rule entries, and the transition process is described by stage division and pre-conditions for switchable state, so that the parameters smoothly transition from the current level to the target level. The slope transition writes the transition start event anchor point and the transition end event anchor point, and marks the state segment of this interval as the transition segment. The state record in the transition segment is used for traceability and verification, and is not included in the upward adjustment basis set, so that the transition fluctuation does not participate in the upward adjustment judgment caliber.

[0073] After the parameter transition is completed, a readback confirmation is performed and the parameter effective anchor point is solidified. The readback confirmation adopts a dual confirmation mechanism. The first confirmation is used to verify that the machine tool control terminal has been successfully written and the readback is consistent. The second confirmation is used to verify that the parameter has not been overwritten or rolled back, and during the confirmation, the continuity and reliability of communication are verified to meet the requirements of the entry. The second confirmation uses the non-transitional state segment in the status log as the confirmation interval to avoid the transition segment affecting the confirmation caliber. The readback confirmation completes the judgment and leaves a trace based on the threshold parameter set entries. After both confirmations are passed, the parameter effective event anchor point is written, and this change is solidified as a new parameter version. The parameter version identifier is generated by the iteration number and the instruction sequence number. The parameter change log writes the version identifier before the change, the version identifier after the change, the change reason identifier, the set of referenced entry numbers, and the effective anchor point, and establishes an association with the status log according to the partition identifier and time anchor point, so that subsequent window judgments can distinguish the state segment before the change, the transition segment, and the candidate state segment after the change according to the anchor point boundary.

[0074] If the readback confirmation fails, the current iteration is terminated and the system enters the review entry point, writing a freeze reason identifier. The review entry point sequentially performs interface status verification, machine tool lock status verification, and parameter interlock condition verification according to the review entry entry items, and writes the verification conclusions and the referenced entry numbers to the review log. Interface status verification includes communication link status, readback delay status, and readback consistency status. Machine tool lock status verification includes parameter write protection status and safety interlock restriction status. Parameter interlock condition verification includes the constraint status of the main variable being restricted by partition interlock or boundary rules. Corrective actions are performed based on the verification conclusions, including reissuing instructions, switching to conservative hold instructions, or reverting to the previous stable version, and the corrective actions are written to the review log and parameter change log. After the review entry point ends and the entry conditions are met, the iteration number is regenerated and the iteration entry point is restored.

[0075] S4: Within the confirmation window and the stable window, the state is determined according to the boundary rules. When a downgrade callback or protection action is triggered, the upgrade is frozen and a review process is executed. The specific implementation is as follows:

[0076] After the parameter change takes effect and the readback confirmation is completed, the partition window process is started from the parameter effect event anchor point. The window process only uses the boundary rule version bound to the partition and its associated threshold parameter set version as the comparison caliber, and filters stable evidence segments with credibility at the allowable level from the status log according to the gating rules. After filtering the transition segment and the review segment, it enters the confirmation window. When the confirmation window passes and is not in a frozen state, it moves to the stable window.

[0077] The confirmation window uses the parameter effective event anchor point as the start reference point. After startup, window metadata is written and the confirmation window is marked as active. The window metadata includes window number, window type identifier, partition identifier, parameter version identifier, boundary rule version identifier, threshold parameter set version identifier, and window start time identifier. Within the confirmation window, the status is judged item by item according to the boundary rule entries. The entry types include hard boundary entries and soft boundary entries. The judgment objects cover load status, vibration status, cooling supply status, and tool retraction status. During the judgment process, the trigger entry number, trigger type identifier, associated status segment number or abnormal segment number, and trigger time anchor point are written into the window judgment record to establish the correspondence between trigger entries and evidence segments and support subsequent traceability.

[0078] When a hard boundary entry is matched, the process enters a protection procedure and terminates the confirmation window. The protection procedure enters a frozen state and writes a freeze reason identifier, which is bound to the trigger entry number for subsequent tracing of the freeze source. In the frozen state, the upward adjustment entry is closed and the iteration entry is paused, allowing only protection permission actions to be executed. Protection actions are executed according to the fixed action sequence bound to the trigger entry. The action sequence includes limiting feed, maintaining the cooling supply at a safe allowable level, performing tool retraction, and entering a safe dwell position. During the execution of protection actions, action legality verification and action result confirmation are performed. Action legality verification is performed according to the verification sequence given by the interlock entry and written to the interlock check record. The verification sequence covers the start conditions, execution conditions, and completion confirmation conditions of the action; the action result confirmation is performed according to the confirmation sequence and consistency criteria given by the readback confirmation item, and written into the readback confirmation record, so that the action execution process and completion status have verifiable and reproducible record evidence; after the protection handling is completed, it is written into the protection event record, which is associated with the trigger item number and trigger time identifier, and records the parameter version identifier, action sequence identifier, action start and end event anchor point, interlock check conclusion, readback confirmation conclusion and freeze reason identifier at the time of triggering, and establishes a reference relationship with the parameter change log, so that the protection trigger can correspond to the specific parameter version and specific trigger item;

[0079] When a soft boundary entry is matched, it is handled according to the entry's level rules. If the soft boundary trigger does not meet the restriction level conditions, the parameters remain unchanged and observation continues. At the same time, the trigger statistics are written into the window judgment record and included in the window summary data source. The trigger statistics are registered according to the counting criteria defined in the entry fields, including at least one of the trigger count, duration segment number, and repeated trigger count. When the soft boundary trigger meets the restriction level conditions, it enters the downgrade callback handling and writes the freeze reason identifier. Then, it performs a reverse small-step callback on the main variable. The callback amplitude is constrained by the step size upper limit entry, and the callback rhythm is executed according to the controlled switching criteria. The controlled switching completes the preparatory check and slope transition in sequence. After the preparatory check passes the interlock entry verification, it enters the slope transition. The slope transition interval is marked as the transition segment and written into the transition start and end event anchor points. After the callback is completed, a readback confirmation is performed and written into the callback effective event anchor point. After the readback confirmation is passed, a callback version identifier is generated, and a reference relationship is established between the callback version identifier and the original parameter version identifier to indicate that the callback belongs to the security correction link within the same partition.

[0080] After the downgrade callback is completed, the review entry is initiated, and the sampling strategy is switched to the tightened sampling level identifier preset by the partition rules. Tightened sampling takes effect by updating the reference relationship of the sampling location set and the sampling trigger interlock item number, so that the sampling covers the key sampling location set and simultaneously tightens the pre-sampling status requirements. The review entry organizes the execution process according to the review entry items and writes them to the review log. The review verification items include at least the measurement calibration version being in a valid state, the status log credibility being at an allowable level, the transition phase being completed, the clamping status having no loosening prompts, the tool status having no replacement prompts, and the communication and readback status meeting the threshold item caliber. The review log writes the verification conclusion and the referenced item number set to ensure that the review caliber is consistent with the partition binding relationship and is traceable. When the review passes, the frozen state is unfrozen and the confirmation window entry is reopened for re-judgment of the callback version in the current partition. When the review fails, the frozen state is maintained and manual processing is initiated. The stable window is not entered until the manual processing is completed and the review entry item requirements are met, in order to avoid consistency judgment under unreliable baseline or status conditions.

[0081] After the confirmation window closes and no hard boundary entries are triggered, and the frozen state is lifted, the stable window is started. The stable window completes registration based on window metadata and enters the active state. At the same time, it calls the sampling location set from the partition table to organize the sampling action. Before the sampling action starts, the sampling interlock verification is performed. The interlock verification is carried out according to the verification sequence defined by the interlock entry, and the verification conclusion is written to the sampling interlock inspection record. The interlock verification requires at least that the machine tool is in a safe stopping posture, the spindle status meets the sampling conditions, the cooling supply is in the sampling allowable state, and it is not currently in the parameter switching transition section or the tool retraction return section. The start and end of the sampling are written to the event anchor point, and the sampling results are written to the sampling log. The sampling log is associated with the partition identifier, sampling hole position identifier, sampling time identifier, measurement calibration version identifier, and parameter version identifier to establish a traceable association between the sampling results and the corresponding status segment and parameter segment, and to serve as the input basis for consistency verification.

[0082] Within the stable window, continuous state determination is performed, and joint verification of continuous stability and sampling consistency is conducted. State determination follows the boundary rule entries bound to the partition and their associated threshold parameter set entries. Continuous stability verification targets the state segments covered by the stable window, requiring that no hard boundary entries are triggered during the window period, and that soft boundary entries do not meet the restriction-level triggering conditions, while maintaining the allowable level of credibility. Sampling consistency verification is performed according to the sampling consistency criterion determined by the partition rules. The sampling consistency criterion references the threshold parameter set entries consistent with the boundary rule version as the comparison criterion, and uses the set of sampling record numbers within the stable window as the judgment range. At the same time, consistency recovery judgment is performed according to the recovery criterion defined by the threshold entry field, so that the sampling consistency verification has a fixed input range and a fixed comparison criterion. Once the entry conditions for protection handling or downgrade callback handling are met within the stable window, the stable window is immediately terminated and the window termination reason is written. The window termination reason is associated with the trigger entry number, and the window enters a frozen state and starts the review entry, which can complete downgrade and blocking upgrade when abnormal disturbances occur.

[0083] After a window ends, a window summary is generated and written to the window log. The window summary records the window number, start and end time identifier, partition identifier, parameter version identifier, boundary rule version identifier, and threshold parameter set version identifier. It also summarizes the set of trigger entry numbers, trigger type statistics, soft boundary trigger statistics, abnormal segment number references, sampling record number set references, and freeze status changes and freeze reason identifiers within the window. The window summary establishes reference relationships with the status log, parameter change log, protection event record, sampling log, interlock check record, and review log under the task identifier to support the traceability and review of the window determination process.

[0084] S5: When the convergence criteria are met, solidify the partition convergence version and archive parameter change records, window summaries, and sampling records. The specific implementation is as follows:

[0085] After the stable window ends and the window summary registration is completed, the partition convergence judgment entry is initiated. The convergence judgment uses the task identifier, partition identifier, and current parameter version identifier as indexes to read the window summary, sampling record, and parameter change behavior record corresponding to the continuous stable window chain, and uses the task registration record as the caliber to verify the consistency of the evidence citation relationship. The continuous stable window chain uses the stable window summary as the scope boundary, only including records within the coverage area of ​​the stable evidence segment, and excluding records within the coverage area of ​​the transitional event anchor point and the review segment. When the window summary has a termination reason, or the boundary rule version and threshold parameter set version referenced by the window summary do not meet the partition binding relationship, the current window chain stops continuing and restarts. The consistency verification requires that the processing benchmark version, measurement calibration version, boundary rule version, and threshold parameter set version referenced by the window summary, sampling record, and parameter change behavior record are in a valid state and the binding relationship is complete. When there is version failure, version inconsistency, or missing binding relationship, the review entry is entered and the solidification process is prohibited.

[0086] The convergence criteria are given by the boundary rule version bound to the partition. The boundary rule version adopts an itemized structure, which gives the entry conditions for convergence determination, the scope of the evidence chain, and the reference relationship between the determination items. The convergence determination adopts a parallel approach of quality sampling conditions and parameter change behavior conditions. Only when both types of conditions are met simultaneously within the same set of continuous stable window chains is the convergence determination allowed. The determination only references stable evidence segments with credibility at the allowable level. Restricted evidence segments are not included in the determination criteria.

[0087] The quality sampling inspection criteria use the consistency of sampling records as the core of verification. Sampling records reference a fixed set of sampling locations in the partition table, and this set maintains a stable standard during task execution. When a review entry is triggered, causing a change in the set of sampling locations, the current window chain is discontinued, and a new window chain is established based on the new set of sampling locations, participating in subsequent judgments again. The quality sampling inspection criteria require that all sampling records within the window chain fall within the target tolerance band, and that the differences between sampling records do not exceed the upper limit of fluctuation. The target tolerance band and the upper limit of fluctuation are given by the sampling consistency entries in the threshold parameter set and are specified in the window summary. The corresponding entry number is registered in the system; the threshold parameter set entry records the indicator name, applicable scope identifier, comparison caliber, trigger type, continuous condition and recovery condition in a structured field, and records the source identifier to indicate the basis for threshold formation; the basis for threshold formation corresponds to at least one of the following: equipment or tool safety specifications, process verification records, historical process package stability window evidence and on-site calibration records; threshold adjustment enters the review entry through change registration, and after the review is approved, a new threshold parameter set version is generated and the reference pointer of the task registration record is updated, so that the threshold caliber has auditable and traceable conditions and is subject to the constraints of permission level and security boundary;

[0088] The quality sampling inspection conditions also require that the measurement calibration version referenced in the sampling inspection record is in a valid state and that the binding relationship is consistent with that in the task registration record; when the measurement calibration version is invalid, the measurement calibration version is replaced but the binding registration is not completed, or the version referenced in the sampling inspection record is inconsistent with the binding relationship, the quality sampling inspection conditions are determined to be unmet and the process is transferred to the review entry point.

[0089] The core verification of parameter change behavior is whether the closed-loop iteration enters a controllable statically stable state. During verification, the change direction identifier, change type identifier, and freeze identifier in the parameter change log are read and compared with the soft boundary trigger statistics in the window summary. The parameter change behavior condition requires that there are no upward-adjusting parameter change records within the same window chain. When an upward-adjusting record exists within the window chain, the upward adjustment magnitude must not exceed the minimum adjustment unit, and no new limit-level soft boundary triggers must be added within the window chain. The minimum adjustment unit is defined by the statically stable parameter entries in the threshold parameter set according to the category of the primary variable and bound to the partition, serving as a unified standard for judging whether an upward adjustment has entered the statically stable interval. When there is an inconsistency between the upper limit of the step size and the minimum adjustment unit, a more conservative constraint is used as the judgment standard, and the referenced entry number is registered in the window summary. The soft boundary trigger statistics are defined by the soft boundary statistics entries in the threshold parameter set, and the statistics entries are clearly defined. The trigger type, counting caliber, and persistence caliber must be clearly defined. The statistical scope only includes stable evidence segments referenced in the window summary, excluding transitional and review coverage areas to maintain consistency between the statistical caliber and the window chain caliber. Soft boundary trigger statistics are based on window summary registration, and the evidence segments included in the statistics must meet the acceptable credibility level to avoid noise triggering introduced by restricted evidence segments. Protection direction actions and downgrade callback actions are not included in the adjustment judgment, but the corresponding trigger entry number and freeze reason identifier must be traceable in the window summary and parameter change log. When both the quality sampling condition and the parameter change behavior condition are met, the partition status is marked as entering the convergence state and written to the convergence judgment event anchor point. The convergence judgment event anchor point records the window number set, sampling record number set, current parameter version identifier, boundary rule version identifier, threshold parameter set version identifier, and convergence criterion entry number set, which are used as input credentials for subsequent solidification actions.

[0090] Upon entering the convergence state, the partition convergence version is solidified. During solidification, a version freeze lock is applied to the partition. The version freeze lock is used to close the upward entry and lock the reference to the current parameter group version. Modification of this version by overwriting is not allowed during task execution. In the freeze state, only protection actions triggered by boundary rules or review and correction processes are allowed to generate callback or rollback versions. Each correction must reference the trigger entry number and be written to the log. The version freeze lock is written to the task control log and bound to the partition identifier, while the freeze reason identifier is registered as convergence solidification. After the freeze lock takes effect, the current parameter group version is marked as the convergence version and a convergence version number is generated. The convergence version number is bound to the task identifier and partition identifier, and the parent version identifier and convergence judgment event anchor point reference are recorded, enabling the convergence version to locate the corresponding window chain evidence and sampling chain evidence. Simultaneously, during the solidification process, the version freeze lock is written to the task control log and bound to the partition identifier. The reference locking clause defines the legal usage conditions of the converged version. These conditions include consistency with the baseline version family, consistency with the measurement and calibration version family, consistency between the boundary rule version and the threshold parameter set version and the partition binding relationship, a state credibility level that meets the entry judgment requirements, and the requirement to complete window re-verification when reused across tasks. The reference locking clause is written into the converged version metadata as a set of entry numbers, ensuring that the converged version is continuously constrained by security boundaries and threshold standards when invoked as a controlled baseline. After the converged version is solidified, the partition enters a hold-up state. During the hold-up state, only the current parameters are maintained or corrections are performed according to the protection action and review / correction process. Temporary versions generated during the hold-up state are registered with the converged version as the parent version and written with the trigger entry number, but do not replace the converged version itself, to avoid short-term corrections changing the converged baseline and affecting reuse and auditing.

[0091] Once the converged version is finalized, it enters the archiving process. The archiving process organizes the parameter change records, window summaries, and sampling records for that partition in chronological order from the start of iteration control to the convergence determination, and generates an evidence index version. The archived content of the parameter change records includes the change cause entry number, change direction identifier, controlled switching gear identifier, readback confirmation conclusion, effective event anchor point, and the state segment number and abnormal segment number associated with the change. The archived content of the window summaries includes the confirmation window and stable window numbers, start and end time identifiers, trigger entry number set, soft boundary trigger statistics and corresponding entry numbers, freeze status changes, and freeze reason identifiers. The termination reason identifier is set, and the parameter version identifier, boundary rule version identifier, and threshold parameter set version identifier referenced in the window summary are registered. The archived content of the sampling record includes the sampling well set version reference, the sampling interlock inspection conclusion, the measurement calibration version identifier referenced during the sampling, and the parameter version identifier corresponding to the sampling. During the archiving process, the original log is kept untouched and modified, and only the archived index and summary are generated. The evidence index version establishes a reference relationship with the parameter change log, status log, window log, sampling log, interlock inspection record, and review log under the task identifier, which is used to locate the window chain, sampling chain, and change chain in the process of convergence version formation.

[0092] After archiving is completed, a process package is formed and archived. The process package is organized in a versioned structure, recording the processing baseline version reference, measurement calibration version reference, partition table version, boundary rule version number, threshold parameter set version number, convergence version number of each partition, and evidence index version reference. The archived event anchor is written to solidify the archived scope and the archived version set. After archiving, only new process package versions are allowed to be generated in an append-only manner. Overwriting or modifying the archived content is not allowed, so that historical process packages can be retained as a reference baseline for long-term retention and used for audit backtracking.

[0093] When subsequent tasks require the reuse of the converged version, the reuse entry point is locked based on the task registration record. Only when the reuse conditions are met and the referenced version caliber is consistent, the corresponding partition converged version is allowed to be loaded as the initial parameter version. After loading, the re-verification process is entered. The re-verification process is based on the binding caliber of boundary rules and threshold parameter sets. The validity verification of the processing benchmark version and measurement calibration version is completed according to the preconditions in the task registration record, and the state credibility is gated. Then, the judgment process of confirmation window and stability window is used to verify the applicability of the converged version in the current task. If the verification is successful, the entry point for keeping running or iteration is allowed. If the verification fails, the use of the converged version is prohibited and the process is rolled back to the basic parameter version or conservative parameter version.

[0094] The technical solution of this embodiment addresses the online calibration and closed-loop iterative control of ring die hole machining. Before machining, a task identifier is generated and a task registration record is established. The machining reference version and the measurement calibration version are bound together. A partition table is established, and the initial parameter sets, boundary rules, and threshold parameter sets for each partition are loaded, ensuring consistent standards and traceability for subsequent judgments and actions. After machining begins, load, vibration, cooling supply, and tool retraction status are continuously collected. Time identifiers are written to the collected data, and credibility markers are generated. Simultaneously, event anchors are set, linking partition switching, parameter activation, tool retraction start / end, and verification start / end to the same time series, forming status segment and abnormal segment numbers that can be used for judgment. When entering a partition, iteration is initiated only when credibility is allowed and interlocking conditions are met. A single dominant variable is determined in speed, feed, cooling supply, and tool retraction rhythm according to boundary rule priority, while other variables are locked. Subsequently, the step size is adjusted at the upper limit and lower limit. Parameter change instructions are generated under small adjustment interval constraints, issued according to slope limits, and confirmed twice. After confirmation, the effective anchor point is fixed and written to the parameter change log. After the parameters take effect, they enter the confirmation window and the stabilization window in sequence. The confirmation window gates hard and soft boundaries. When a hard boundary is triggered, a protection action is executed and the adjustment is frozen. When the soft boundary reaches the limit level, a downgrade callback is executed and the process enters the review. When the confirmation is successful and the parameter is not frozen, the stabilization window completes the sampling inspection and continuous stability verification and generates a window summary under the condition that the sampling inspection interlock is satisfied. When the continuous window meets the convergence criteria, the current partition parameter version is fixed as the converged version and the adjustment is frozen. The archived parameter changes, window summary, and sampling inspection records form a process package and are sealed. When subsequent tasks need to reuse the converged version, the reuse condition matching and benchmark caliber and credibility gating are completed first. Then, the window is re-verified and the process enters the "keep running" or "continue iteration" entry point to prevent the direct reuse of versions that do not meet the conditions.

[0095] It should be noted that this invention can be deployed on the device itself to realize embedded applications, or it can run on a PC or other terminal with a user interface, thereby meeting various hardware environments and usage requirements.

[0096] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wireless or wired transmission; wired transmission methods include optical fiber, twisted pair, coaxial cable, etc.; wireless transmission includes infrared, microwave, etc. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center containing one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0097] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0098] In the embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0099] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0100] In addition, the functional modules in the embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0101] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0102] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0103] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for online calibration and closed-loop iteration of ring die machining parameters, characterized in that, include: S1: Establish a machining datum version, complete clamping datum alignment and measurement consistency calibration, generate task identifiers and partition tables, load initial partition parameters and boundary rules, lock the reference object set, perform multi-directional repeated sampling after zero-point reset and range confirmation of the measuring device, determine consistency according to the calibration rules of the threshold parameter set, and if it fails, perform fixed-point reset and retest after installation and communication verification, generate a measurement calibration version and bind it to the machining datum version, establish a partition table, record the sampling location set and partition switching interlock entry number, the partition switching interlock condition comes from the interlock entry in the boundary rules, and the interlock entry is written into the partition table with the entry number and used as the switching... The pre-check basis; before switching partitions, verify the pre-state item by item according to the interlock items. The pre-state includes at least the parameters being in a stable state, the measurement calibration version being in a valid state, the tool retraction action being completed and the machine tool being in a safe stopping posture, and the communication and readback status meeting the reference requirements. Load the initial version of the partition parameter group and the boundary rule version and write the reference locking rule definition to define the legal use conditions of the converged version. The legal use conditions include the consistency of the baseline version family, the consistency of the measurement calibration version family, the consistency of the boundary rule version and the threshold parameter set version with the partition binding relationship, the state credibility meeting the level requirements of the entry judgment, and the window re-verification must be completed when reusing across tasks. S2: During machining execution, load, vibration, cooling supply, and tool retraction status are collected. The collected data is written with time stamps and reliability markers. Anomalies are generated into anomaly segment numbers through reliability gating and repeatability gating. Anomaly segments are formed using an itemized gating mechanism. Anomaly spike judgment and repeatability gating judgment respectively refer to spike trigger entries and repeatability trigger entries in the threshold parameter set. The spike trigger entries define the triggering and recovery conditions for a single mutation, and the repeatability trigger entries define the repeatability trigger criteria and summary conditions within the non-transition interval. Anomaly identification first undergoes credibility gating. When an abnormal spike occurs in the load or vibration and the credibility is limited or unavailable, only the anomaly is recorded and a review prompt mark is written, without generating an anomaly segment number. After passing the credibility gating, repeatability gating is then performed. When the abnormal spike meets the repeat triggering condition in the non-transition interval and spans multiple state segments, the relevant segments are summarized into a continuous anomaly segment and assigned an anomaly segment number. At the same time, the trigger type identifier and its positional relationship relative to the event anchor point are written into the anomaly segment number record, so that subsequent window judgments can continuously verify the anomaly segment according to the same entry caliber, and keep the anomaly identification and action execution independent of each other. When the partition interlock is not satisfied, the acquisition segment is marked as a review segment and the upward adjustment entry is closed. When the partition is switched, the partition switching event anchor point is written, and the switching interlock check is performed. The interlock check is performed according to the interlock entry number and threshold parameter set caliber bound to the partition. The check conclusion and the corresponding entry number are written into the interlock check record, and a status record is generated. S3: Enter the iterative control process by partition, determine the single main variable, write the partition start event anchor point when entering the partition, the event anchor point records the event type, event number and corresponding time identifier, covering partition switching, parameter change effect, tool retraction action start and end and review entry start and end, lock the reference version based on the task registration record and generate the iteration number, verify the entry condition according to the interlock item, the interlock item specifies the pre-state of change issuance, spot check execution, partition switching and review start, the review entry item specifies the freeze trigger condition, verification items and recovery path, if it fails, set the iteration state to pause and close the upward adjustment entry, generate parameter change instructions according to the step size upper limit and minimum adjustment interval, issue according to the slope limit and execute readback confirmation, generate parameter change instructions according to the step size upper limit and minimum adjustment interval unlock condition and record the reference item number, the unlock condition reference window interlock item, and write the window number as the unlock credential to the parameter change log to form a constraint relationship of alternating progress of change and verification, after pre-check, issue according to the slope limit and write to the transition start and end anchor point and parameter effect anchor point; S4: Within the confirmation window and the stable window, the state is determined according to the boundary rules. The confirmation window uses the parameter effective event anchor point as the start reference point. After starting, the window metadata is written and the confirmation window is marked as active. The window metadata includes the window number, window type identifier, partition identifier, parameter version identifier, boundary rule version identifier, threshold parameter set version identifier, and window start time identifier. After the confirmation window ends and no hard boundary entry is triggered, and the frozen state is lifted, the stable window is started. When the soft boundary trigger reaches the limit level condition, the downgrade callback is processed and the freeze reason identifier is written. Then, the reverse small step callback is executed on the main variable. In addition to saving the instantaneous state point, the state log is also divided according to the event anchor point to form state segments. According to the boundary rule version and its associated threshold parameter set version, the state segments with the required credibility are selected from the state log, and the hard boundary entry and soft boundary entry are matched and the trigger entry number is recorded. When the downgrade callback or protection action is triggered, the upgrade is frozen and the review process is executed. S5: When the convergence criterion is met, the partition convergence version is solidified. When both the consistency entry and the static stability entry of the parameters are met, the convergence anchor point is written, a version freeze lock is applied, and the current parameter version is solidified as the partition convergence version. The quality sampling condition takes the consistency of the sampling record as the core of verification. The sampling record references the fixed sampling position set version in the partition table. The sampling position set version maintains a stable caliber during task execution. The parameter change behavior condition takes whether the closed-loop iteration enters a controllable static stability state as the core of verification. During verification, the change direction identifier, change type identifier, and freeze identifier in the parameter change log are read and compared with the soft boundary trigger statistics in the window summary for consistency. The parameter change record, window summary, and sampling record are archived.

2. The online calibration and closed-loop iteration method for ring die processing parameters according to claim 1, characterized in that, Establish a processing baseline version, including: Before starting the machine, generate a task identifier and establish a task registration record, and register the processing baseline version; Before clamping, check the cleanliness and contact status of the positioning surface. After alignment and locking, verify the posture stability and fix the machining coordinates for tool setting. The machining reference version is rebuilt and the valid version is updated when the clamping is reset, the fixture is reinstalled, the tool is replaced, the equipment is restored from abnormality, or the task is interrupted and restarted.

3. The online calibration and closed-loop iteration method for ring die processing parameters according to claim 1, characterized in that, During machining, load, vibration, cooling supply, and tool retraction status are collected. The collected data is then time-stamped and reliability-marked, and a status record is generated, including: When processing starts, write the time identifier and partition start identifier, and verify the consistency of the processing reference version, measurement calibration version, boundary rule version and threshold parameter set version; Collect load, vibration, cooling supply and tool retraction status and write them into time stamps and reliability markers, and divide the status into segments according to event anchor points.

4. The online calibration and closed-loop iteration method for ring die processing parameters according to claim 1, characterized in that, The iterative control process is initiated by partition, and a single dominant variable is determined, including: After passing, a single dominant variable is determined based on the priority entry, and a locking flag is applied to the remaining variables. When a protection entry is triggered, downscaling actions are allowed on non-dominant variables, and the trigger entry number is recorded.

5. The online calibration and closed-loop iteration method for ring die processing parameters according to claim 1, characterized in that, Parameter change instructions are generated based on the upper limit of the step size and the minimum adjustment interval, and issued according to the slope limit with readback confirmation, including: If the double read confirmation fails, write the freeze reason identifier, and verify the interface status, machine tool lock status and parameter interlock conditions according to the review items, and perform re-issuance, conservative maintenance or version rollback.

6. The online calibration and closed-loop iteration method for ring die processing parameters according to claim 1, characterized in that, Within the confirmation window and the stable window, state determination is performed according to boundary rules, including: After the parameter readback is confirmed, the confirmation window and the stabilization window are started with the parameter effective event anchor point; Once the window is confirmed to be passed and not frozen, perform the sampling inspection according to the sampling interlock item, and generate a window summary to write to the window log.

7. The online calibration and closed-loop iteration method for ring die processing parameters according to claim 1, characterized in that, When a downgrade callback or protection action is triggered, the upgrade process is frozen and a review process is executed, including: When a hard boundary entry is triggered, the protection action sequence bound to the entry is frozen and processed, and written to the protection event log. When a soft boundary entry reaches the restriction level, it is frozen and the main variable is reversed. After slope transition and readback confirmation, the validity, credibility, transition end and communication readback status of the calibration are verified according to the review entry, and the freeze is lifted or maintained according to the verification conclusion.

8. The online calibration and closed-loop iteration method for ring die processing parameters according to claim 1, characterized in that, When the convergence criteria are met, the converged version of the partition is solidified, including: When the convergence criterion is met, version consistency verification is performed based on the window summary, sampling record and parameter change record of the continuous stable window chain, and gating is performed according to the credibility.

9. The online calibration and closed-loop iteration method for ring die processing parameters according to claim 6, characterized in that, Archived parameter change records, window summaries, and sampling inspection records include: After the solidified partition convergence version is completed, the parameter change records, window summaries and sampling inspection records are archived in chronological order to generate the evidence index version. The process package is then sealed by establishing a reference relationship with the status log, window log and sampling inspection log. Only appending is allowed to the sealed content. When reusing across tasks, the converged version is loaded after the reuse conditions are locked according to the task registration and the verification is passed again.

Images