Gear machining correlation correction and numerical control program closed-loop regeneration method and system based on measured residual error of gear grinding travel state prediction

Abstract

The application discloses a kind of based on the gear machining associated correction and numerical control program closed loop regeneration method and system of residual error of gear grinding stroke state prediction and measurement.This application method realizes the same semantic comparison of measured data and predicted data by gauge semantic mapping, generates multiple-gauge prediction vector according to the cumulative feed driven virtual workpiece update, and constructs multiple-gauge residual by pairing the measured value and the predicted value under the same stroke;Under the multiple constraints of single correction limit, cumulative feed continuity, worm available area, etc., the correction amount is inversely solved, the associated correction strategy is executed according to the parent entity and semantics of the parameter to be corrected, and the dependent parameters are updated synchronously;After closed loop verification, the numerical control program is regenerated, and when it fails, it is rolled back to the last valid version.The application optimizes precision correction and tool resource management cooperatively, ensures process parameter consistency, improves the stability of closed loop correction and the reliability of numerical control program, and is suitable for batch precision machining of worm grinding machine.

Description

Technical Field

[0001] This invention belongs to the field of gear precision machining and CNC machining technology, specifically relating to a gear machining correlation correction and CNC program closed-loop regeneration method and system based on gear grinding stroke state prediction and measured residual. It is applicable to gear grinding machining of CNC worm gear grinding machines, and is especially suitable for batch gear production with high requirements for precision and consistency. Background Technology

[0002] As a core component of mechanical transmission systems, gears' machining accuracy directly determines the system's load-bearing capacity, operational smoothness, and service life. Among various gear finishing processes, continuous generating grinding (worm wheel grinding) has become the mainstream finishing method for small-module hardened gears in automotive transmissions, new energy vehicle reducers, and aerospace transmissions due to its high processing efficiency and stable accuracy. With the continuous improvement of NVH performance and load-bearing capacity requirements for gears in applications such as new energy vehicles and high-speed transmissions, gear machining accuracy levels are becoming increasingly stringent. In continuous generating grinding, the final gear accuracy depends not only on the machine tool's initial accuracy and the initial settings of machining parameters, but also on the significant combined effects of multiple factors during grinding, including wheel wear, thermal deformation, cutting force fluctuations, and uneven material hardness. Therefore, how to perceive the gear's accuracy status in real time during machining and dynamically adjust subsequent machining strategies based on actual measurement results has become a key technical issue in the field of gear precision manufacturing.

[0003] Existing technologies have been extensively studied in terms of quality control and parameter correction in gear grinding.

[0004] (1) Tooth surface measurement feedback and grinding parameter correction. Gear tooth surface error data is obtained through in-machine or external measurement, and grinding parameters are corrected accordingly. For example, one solution for the forming grinding of topologically modified helical gears obtains the tooth surface error through in-machine measurement, and calculates the correction amount in an optimized manner based on the sensitivity relationship between the axis motion polynomial coefficients and the tooth surface error, forming a closed loop of "in-machine measurement - sensitivity analysis - parameter correction - re-grinding". Another solution for cycloidal gear grinding establishes a motion and tooth profile model based on the continuous closed tooth profile of the cycloidal gear, and calculates the quantitative correction value of the machine tool grinding setting through numerical optimization. These solutions achieve accuracy correction based on measurement feedback to a certain extent, but their limitations are: they are mainly for forming grinding or specific tooth profiles, and do not involve operations such as constructing virtual workpieces by cumulative feed according to stroke in continuous generating grinding, predicting multiple gauge dimensions at the same time, maintaining the remaining worm gear tool movement area, etc., nor do they involve the propagation of workpiece, process and tool related parameters in CAM and CNC program version management.

[0005] (2) Evaluation of continuous generating grinding process and adjustment of feed / tool ​​movement. This focuses on evaluating the grinding wheel condition or workpiece quality through process signals and adjusting grinding parameters accordingly. For example, some schemes use near-process acoustic emission sensors and grinding wheel spindle current to evaluate the grinding wheel condition and workpiece quality, and identify vibration frequency bands related to tooth surface ripples through simulation process models, thereby adapting the tool movement and feed motion of continuous generating grinding. Other schemes study the wear evolution of worm grinding wheels by cumulative material removal volume, change grinding speed, normal allowance, feed speed and tool movement amount through orthogonal experiments, quantify the worm wear state using fractal dimension, and analyze the relationship between wear state and tooth surface roughness. These schemes are suitable for process monitoring and offline parameter recommendation, but their core feedback is process sensor signals or wear characterization, and they do not use the residual between gauge prediction value and measured value to drive the next stroke correction, nor do they involve online recalculation of the remaining worm area, CAM associated parameter propagation and NC generation verification.

[0006] (3) Improvement of gear measuring device and equipment structure. In the field of gear measurement, a large number of patent documents have involved structural improvements, pose alignment and calibration methods of gear measuring systems. For example, some solutions unify the measurement and machining reference of large gears through auxiliary centering structures and square ruler touch measurement; some solutions map displacement parameters and mechanical parameters under different inclination angles to a coordinate system consistent with the gear installation reference to describe the slope angle distribution; and some solutions improve the efficiency of three-dimensional gear detection through dual loading areas and laser contour detection mechanisms. Some other patent documents involve improvements to the dressing device, shaft system structure or line contact machining mechanism of generating gear grinding machines. The above solutions focus on the calibration accuracy, angle detection, detection efficiency or machine tool structure improvement of the measurement system, but none of them involve the prediction of stroke-level multi-gauges, the correlation correction of measured residual drive and the constraint of the remaining worm region.

[0007] In summary, while existing technologies have made progress in areas such as tooth surface measurement feedback, process evaluation, tool shifting management, and measuring devices, a unified and coordinated systematic solution has yet to be formed that integrates "stroke-level accuracy prediction, multi-gauge measurement feedback, CAM-related parameter propagation, worm gear resource management, and CNC program closed-loop regeneration." Specifically, the following are the main drawbacks.

[0008] (1) Measurement data and process prediction data lack unified semantics. Existing systems may have the ability to import gear measurement results and predict gear grinding processes, but the configuration of the measurement item name, unit, measurement radius, number of teeth spanned and ball bar diameter may not correspond one-to-one with the predicted quantity. Directly assigning values ​​by name can easily mix data under different measurement conditions, leading to errors in correction decisions.

[0009] (2) A single error quantity is difficult to reliably reflect the actual machining deviation. Different gauges respond differently to tooth thickness, measurement radius and gear geometry. Single gauge correction is easily affected by measurement abnormalities or local model errors. The lack of multi-gauge consistency verification will lead to misjudgment or overcorrection.

[0010] (3) Precision correction and worm gear resource management are separated. Existing feed or tooth thickness correction usually only focuses on the workpiece size, while tool shifting management only focuses on the available area of ​​the grinding wheel. Correction actions may cause future feed plans to be discontinuous, or cause new tool shifting positions to enter the used area, the safety margin area, or exceed the allowable range, resulting in waste of tool resources or interruption of processing.

[0011] (4) Parameter modification lacks cross-entity correlation propagation. Machining modifications may involve workpieces, features, processes, machining types and shared tools. If only the current interface parameters are modified without updating the relevant process and dependent parameters, it will cause inconsistencies in parameters under the same tool or machining type, leading to process conflicts.

[0012] (5) The NC program regeneration lacks closed-loop verification and traceability mechanisms. If the NC program is generated directly after the parameters are written without verifying whether the residual converges, whether the cumulative feed is monotonic, whether the remaining worm gear area is valid, and whether the new parameters meet the physical boundaries, it may generate an unexecutable program or a program that is inconsistent with the actual grinding state; it is also difficult to restore the previous valid version when the generation fails, which affects the reliability of production.

[0013] To address the aforementioned technical issues, there is an urgent need for a systematic solution that can organically integrate gear grinding stroke state prediction, multi-gauge measured residual feedback, worm gear resource management, CAM associated parameter propagation, and CNC program closed-loop regeneration. Summary of the Invention

[0014] In view of this, the purpose of this invention is to provide a method and system for gear machining correlation correction and CNC program closed-loop regeneration based on gear grinding stroke state prediction and measured residual. By establishing a closed-loop feedback mechanism of stroke-level multi-gauge prediction and semantically equivalent measured residual, and integrating worm gear resource management and CAM correlation parameter propagation, stable correction of machining accuracy, collaborative optimization of tool resources, and closed-loop reliability of CNC program are achieved, thereby improving the accuracy consistency and process stability of gear grinding.

[0015] To achieve the above objectives, the present invention provides the following technical solution: This invention first proposes a method for gear machining correlation correction and CNC program closed-loop regeneration based on gear grinding stroke state prediction and measured residual, including: S1, Acquire and parse gear measurement data: Monitor or read gear measurement files, parse the evaluation items output by the measuring equipment into records containing measurement keys, values, units, measurement configurations, time and source files, and mark the validity of the data and the consistency of the configuration; S2, Perform gauge semantic mapping: Maintain a semantic dictionary between the measuring device identifier and the internal gauge identifier of the CAM, and map the measurement data to a unified gauge semantic space based on the semantic dictionary to generate a set of actual measurements; the mapping includes comparison and unification of gauge type, unit, measurement direction, measurement radius or tooth tip height, number of teeth spanned, ball or roller diameter, applicable tooth profile and allowable tolerance; S3, Maintain process plan and cumulative feed status: Save incremental feed, cumulative feed, machining stage, allowable allowance and current CNC version according to grinding stroke; lock executed strokes to prevent changes, and create or update the initial process plan based on total feed depth and number of strokes; S4, Update the virtual workpiece: Perform equivalent material removal on the initial gear geometry based on the current cumulative feed to obtain a virtual workpiece corresponding to the current stroke state; the equivalent material removal is achieved by converting the radial cumulative feed into the normal tooth thickness change, and then mapping it into the displacement coefficient change; S5, Perform multi-gauge prediction: On the same virtual workpiece, based on a unified geometric datum and gauge configuration, calculate the predicted values ​​of multiple gauges to form a multi-gauge prediction vector; the gauges include at least two of the following: cross tooth thickness, normal chord tooth thickness, normal tooth thickness, transverse tooth thickness, single ball, double ball, single roller, and double roller measurement dimensions. S6, Maintain worm gear single-tool status and remaining available area: Based on the worm gear's allowable tool movement range, starting position, step distance, number of machined workpieces, used area, dressing record, and safety margin, maintain the current tool movement position and remaining available area or set of areas; S7, Construct multi-gauge residuals: Pair the actual measurement set under the same travel and the same gauge semantics with the multi-gauge prediction vector, calculate the physical residual and tolerance normalized residual, and generate an effective gauge set; if the number of effective gauges is insufficient or there is a configuration conflict, output an automatic reverse calculation prohibition status. S8, Constraint Correction Inverse Calculation: Establish the response relationship of the gauge residual to the parameter to be corrected, and solve the correction amount of the process parameter under the constraints of single correction limit, parameter physical boundary, cumulative feed continuity, remaining machining allowance and worm gear usable area; if there is no feasible solution, perform downgrade processing. S9, Execute the correlation correction strategy: Based on the parent entity and semantic selection of the parameter to be corrected, determine the propagation range of the correction effect, write the target parameter and update the dependent parameters synchronously; the propagation range includes the current workpiece, the current feature, the current process, related processes under the same processing type, or related processes using the same tool; S10, Update Future Plans and Closed-Loop Verification: Allocate the obtained feed correction to strokes that have not yet been executed, and recalculate the future cumulative feed; recalculate the remaining worm gear area based on the new tool shifting position, step distance, and used area; at least verify the parameter boundaries, cumulative feed monotonicity, remaining margin, tool shifting position feasibility, consistency of associated parameters, and convergence of predicted residuals; only proceed to the CNC regeneration step after verification is passed. S11, Perform CNC regeneration and version management: Call the CAM CNC generation mechanism to generate a new CNC program corresponding to the corrected process state, and record the old version, new version, correction trigger reason, predicted and measured vectors, residuals, parameter changes, future feed plan and remaining worm gear area; if CNC generation, syntax check or process verification fails, restore the previous valid parameter snapshot and CNC version.

[0016] Furthermore, in step S2, the gauge semantic mapping also includes: when the measurement units are inconsistent but the conversion relationship can be determined, the units are unified first and then compared; when the number of teeth spanned, the measurement radius, the diameter of the ball bar, or the measurement direction are inconsistent, the measurement item does not participate in the residual construction and is marked as a configuration conflict.

[0017] Furthermore, in step S4, based on the current cumulative radial feed... Equivalent material removal is performed on the initial gear geometry to obtain the displacement coefficient of the virtual workpiece. :

[0018] in: These are the initial displacement coefficients; This is the conversion factor for radial feed to normal tooth thickness reduction; Normal modulus; It is the normal pressure angle; The cumulative radial feed The result is obtained by accumulating the radial feed increments of each stroke:

[0019] in, For the first The stroke increments radial feed, and satisfy... ; The cumulative radial feed at the end of stroke 0 is defined as 0. The maximum allowed cumulative radial feed; This is the index for the current itinerary.

[0020] Furthermore, in step S7, the physical residual and tolerance normalized residuals They are respectively:

[0021] in: For the first Itinerary No. Predicted values ​​for individual gauges; These are the corresponding measured values; For the first Normalized scale or tolerance of a gauge; The set of valid gauges It consists of gauges with consistent configuration, valid values, and capable of participating in identification; The automatic correction is based on the maximum absolute normalized residual:

[0022] when Keep the current parameters; when When entering the constraint correction inverse calculation, where This is the preset dead zone threshold.

[0023] Furthermore, in step S8, the constraint correction inverse calculation obtains the correction vector by solving a weighted optimization problem with regularization terms. :

[0024] in: For the first The credibility weight of each metric; For the first Itinerary No. The parameter for the first Sensitivity matrix of each normalized predictor; For the first The current correction amount for each process parameter; For the first Regularization scale for each parameter; The regularization coefficient is used. Tolerance normalization residuals; For a set of valid gauges; This represents the total number of process parameters to be corrected. And each of the corrected parameters satisfies:

[0025] in: and The first The lower and upper limits allowed for each parameter; For the first The maximum single correction magnitude for each parameter.

[0026] Furthermore, the sensitivity matrix Obtained through any of the following methods: Analytical differentiation is performed based on gear geometry and the grinding motion model; Apply a small perturbation to the parameter to be corrected on the virtual workpiece, calculate the change of each gauge prediction value with finite difference, and divide by the corresponding normalized scale; Offline calibration is performed using standard parts, historical workpieces, or trial grinding data. Updated iteratively based on the measured changes before and after each revision; When the number of effective gauges is less than the number of parameters to be corrected, or when the sensitivity matrix is ​​ill-conditioned, reduce the number of parameters to be corrected, add regularization terms, correct only the highest priority parameters, or request additional testing.

[0027] Furthermore, in step S8, the cumulative feed continuity constraint is:

[0028] in, For the revised future Cumulative radial feed; To maintain the maximum permissible cumulative radial feed, the cumulative feed of the executed stroke remains unchanged; To be allocated to the future The cumulative feed adjustment amount of the stroke; For the first Cumulative radial feed of the stroke; This represents the total number of grinding strokes. The usable area constraint of the worm gear is:

[0029] in: and These are the current cutter position and the planned next cutter position, respectively. For the next step of the cutting tool movement; The remaining available area before selecting the next cutter position; and These are the lower and upper bounds of the remaining available area, respectively. For safety margins, for reverse or segmented tool shifting, direction transformation or interval set calculation is used.

[0030] Furthermore, in step S10, the closed-loop verification also includes predicting improvement conditions:

[0031] in: This is the weighted sum of squared residuals before correction; This is the weighted sum of squared residuals after the candidate corrections and re-prediction; To predict the improvement acceptance coefficient, and ; For the first The credibility weight of each metric; Tolerance normalization residuals; For the first Itinerary No. The tolerance normalized residuals obtained from the re-prediction of each gauge; For a set of valid gauges; If the preset maximum number of retries is exceeded If all constraints cannot be met simultaneously, automatic correction will terminate, and the system will either maintain or roll back to the previous valid state.

[0032] This invention also proposes a gear machining correlation correction and CNC program closed-loop regeneration system based on grinding stroke state prediction and measured residual, used to implement the gear machining correlation correction and CNC program closed-loop regeneration method based on grinding stroke state prediction and measured residual as described above, including: The measurement result acquisition module is used to monitor or read gear measurement files and parse the evaluation items output by the measuring equipment into records containing measurement keys, values, units, measurement configurations, time, and source files. The gauge semantic mapping module is used to maintain a semantic dictionary between the measurement equipment identifier and the gauge identifier inside the CAM, map the measurement data to a unified gauge semantic space, and generate a set of actual measurements. The process planning and cumulative feed module is used to save incremental feed, cumulative feed, machining stage, allowable allowance and current CNC version according to the grinding stroke, and to lock the executed stroke; The virtual workpiece update module is used to perform equivalent material removal on the initial gear geometry based on the current cumulative feed, so as to obtain a virtual workpiece corresponding to the current stroke state. The multi-gauge prediction module is used to calculate the predicted values ​​of multiple gauges on the same virtual workpiece, forming a multi-gauge prediction vector. The Single Tool Status and Remaining Available Area module is used to maintain the current tool position and the remaining available range or set of ranges for the worm gear; The residual construction module is used to pair measured values ​​and predicted values ​​under the same travel and the same gauge semantics, calculate physical residuals and tolerance normalized residuals, and generate a set of valid gauges; The constraint correction inverse calculation module is used to establish the response relationship of the gauge residual to the parameter to be corrected, and to solve the correction amount of the process parameter under constraint conditions. The associated correction strategy module is used to set the strategy based on the parent entity and semantic selection of the parameter to be corrected, determine the propagation scope, write the target parameter and synchronously update the dependent parameters. Future plans include updating and implementing a closed-loop verification module to allocate feed corrections to strokes that have not yet been executed, recalculate future cumulative feeds and remaining worm gear area, and perform multiple constraint verifications. The NC regeneration and versioning module is used to call CAM's NC generation mechanism to generate new NC programs and perform version management, logging, and rollback operations.

[0033] Furthermore, the mapping performed by the gauge semantic mapping module includes comparison and unification of gauge type, unit, measurement direction, measurement radius or tooth tip height, number of teeth spanned, ball or roller diameter, applicable tooth profile and allowable tolerance; when measurement configurations are not equivalent, the merging and use of corresponding measurement data are prevented; The constraints executed by the constraint correction and reverse calculation module include: single correction limit, parameter physical boundary, cumulative feed monotonicity, remaining machining allowance, and worm gear available area constraint; the worm gear available area constraint supports unidirectional continuous tool movement, reverse tool movement, segmented tool movement, and topological diagonal tool movement.

[0034] The beneficial effects of this invention are as follows: This invention relates to a gear machining correlation correction and NC program closed-loop regeneration method based on gear grinding stroke state prediction and measured residual. By integrating gear grinding stroke state prediction, multi-gauge residual feedback, worm gear resource constraints, CAM correlation propagation and NC closed-loop regeneration into a unified control link, it brings the following systematic technical effects.

[0035] (1) Improved reliability of correction through consistency judgment of multiple gauges. This invention generates multiple gauge prediction values ​​such as cross tooth thickness, chord tooth thickness, normal tooth thickness and ball bar size under the same virtual workpiece and unified gauge configuration, and performs semantic matching with the measured values. Compared with the correction methods of single gauge or inconsistent gauge configuration in the prior art, the multi-gauge method can effectively identify single gauge measurement anomalies or local model errors, reduce the probability of incorrect correction and overcorrection, and make the correction decision more robust.

[0036] (2) Stroke state-driven precision closed loop. This invention uses the cumulative feed as the driving variable for updating the virtual workpiece, so that each stroke has a clear geometric reference and a corresponding multi-gauge prediction value. Compared with the existing technology where the cumulative feed is only used as a process record, this scheme realizes the quantitative correspondence between each measurement result and a specific stroke state, providing a traceable stroke-level reference for subsequent residual calculation and correction back-calculation, and improving the precision of closed-loop control.

[0037] (3) Collaborative optimization of accuracy correction and resource management. This invention considers single correction limit, parameter boundary, cumulative feed monotonicity, and remaining usable area of ​​the worm gear simultaneously in the constraint correction inverse calculation. The correction action not only makes the tooth thickness approach the target, but also ensures that the tool position always falls within the usable range and does not destroy the used area and safety margin constraints, avoiding problems such as "the size is qualified but the tool goes out of bounds" or "resource constraints destroy the continuity of processing", and realizing the joint optimization of processing accuracy and grinding wheel life.

[0038] (4) CAM-related propagation ensures process consistency. This invention selects a propagation strategy based on the parent entity (workpiece, feature, process, machining type, tool) of the parameter to be corrected, and synchronously updates dependent parameters when writing the target parameter. Compared with the defects in the prior art where local value changes may cause parameter conflicts under the same tool or machining type, this solution ensures the consistency of process data through the related correction strategy and the maintenance of correction source records.

[0039] (5) Closed-loop verification and version management improve system reliability. This invention allows NC to be regenerated only after verification is passed, and can roll back to the previous valid version in case of failure. This mechanism effectively avoids the problem of generating unexecutable programs due to residual non-convergence, discontinuous feed plan or worm gear failure, and at the same time makes each NC version traceable to the corresponding prediction, actual measurement, residual and parameter snapshot, improving the traceability of the production process and the fault recovery capability. Attached Figure Description

[0040] To make the objectives, technical solutions, and beneficial effects of this invention clearer, the following figures are provided for illustration: Figure 1 This is a flowchart of the method for gear machining correlation correction and CNC program closed-loop regeneration based on gear grinding stroke state prediction and measured residual of the present invention. Detailed Implementation

[0041] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention. It should be understood that the specific embodiments described herein are only for explaining the present invention and are not intended to limit the present invention.

[0042] This embodiment provides a method and system for gear machining correlation correction and CNC program closed-loop regeneration based on grinding stroke state prediction and measured residuals. The method and system of this embodiment are applicable to CNC worm gear grinding machines. Specifically, in this embodiment, the method for gear machining correlation correction and CNC program closed-loop regeneration based on grinding stroke state prediction and measured residuals is executed by a system for gear machining correlation correction and CNC program closed-loop regeneration based on grinding stroke state prediction and measured residuals. This system consists of a measurement and semantic layer, a stroke state prediction layer, a worm gear resource layer, a closed-loop decision layer, and a CAM execution layer. The measurement and semantic layer transforms external measured data into unified gauge semantics; the stroke state prediction layer generates multi-gauge references corresponding to the current stroke; the worm gear resource layer maintains the available tool shifting area; the closed-loop decision layer forms residuals and inversely calculates the action; the CAM execution layer propagates parameters, updates future plans, verifies, and generates a new CNC. The process plan corresponding to the new NC re-enters the next stroke state calculation, forming a closed loop.

[0043] Specifically, this embodiment is a gear machining correlation correction and CNC program closed-loop regeneration system based on gear grinding stroke state prediction and measured residuals. It includes a measurement result acquisition module, a gauge semantic mapping module, a process planning and cumulative feed module, a virtual workpiece update module, a multi-gauge prediction module, a single tool state and remaining available area module, a residual construction module, a constraint correction and reverse calculation module, a correlation correction strategy module, a future plan update and closed-loop verification module, and a CNC regeneration and version module.

[0044] The measurement result acquisition module monitors or reads gear measurement files, parsing the evaluation items output by the measuring equipment into records containing measurement keys, values, units, measurement configurations, time, and source files. The module retains the original measurement files or summary identifiers to facilitate traceability between subsequent corrections and NC versions. For data that is not numerical, not finite, has duplicate keys, or lacks key configurations, this module marks its validity rather than using it directly for parameter correction.

[0045] The gauge semantic mapping module maintains a semantic dictionary between the measuring equipment identifier and the internal gauge identifier of the CAM, mapping measurement data to a unified gauge semantic space and generating a set of actual measurements. Specifically, each dictionary entry must contain at least the gauge type, unit, measurement direction, measurement radius or tooth tip height, number of teeth spanned, ball or roller diameter, applicable tooth profile, and allowable tolerance. Mapping compares not only names but also configurations and units; data with convertible units but inequivalent measurement configurations must not be merged. This module outputs a set of actual measurements within the same semantic space to the residual construction module.

[0046] The process planning and cumulative feed module is used to save incremental feed, cumulative feed, machining stage, allowable allowance, and current NC version by grinding stroke, and to lock executed strokes. The initial plan can come from CAM process settings or be obtained from the total feed depth and stroke allocation. Each revision only allows changes to the plan after the current stroke; executed strokes remain immutable; the updated cumulative feed should monotonically remain constant and be constrained by the maximum total feed and remaining machining allowance.

[0047] The virtual workpiece update module performs equivalent material removal on the initial gear geometry based on the current cumulative feed, resulting in a virtual workpiece corresponding to the current travel state. A preferred approach is to convert the radial cumulative feed into a normal tooth thickness variation, and then map it to a displacement coefficient variation; alternatively, the tooth thickness, tooth space width, or other geometric parameters that can drive gauge calculations can be directly updated. The virtual workpiece serves as a unified geometric reference for the prediction module, preventing inconsistent material removal assumptions from being applied to different gauges.

[0048] The multi-gauge prediction module is used to calculate predicted quantities of multiple gauges on the same virtual workpiece, forming a multi-gauge prediction vector. Specifically, the multi-gauge prediction module calculates the measured dimensions of cross tooth thickness, normal chord tooth thickness, normal tooth thickness, transverse tooth thickness, single ball, double ball, single roller, and double roller on the same virtual workpiece. Each predicted quantity is output along with its measurement configuration. The prediction set can be a subset selected based on tooth profile, gauge, and on-site measurement capabilities, but the number and independence of effective measurements used for reverse engineering should be sufficient to identify the parameters to be corrected.

[0049] The Single Tool Status and Remaining Available Area module is used to maintain the current tool movement position and the remaining available range or set of ranges for the worm gear. For forward tool movement, the available area can be represented as the range from the unused starting point to the allowed ending point; for reverse or segmented tool movement, multiple ranges with different directions can be maintained. Any new tool movement must fall within the available area and avoid overlapping with the used area and the reserved safety zone.

[0050] The residual construction module pairs measured and predicted values ​​under the same travel distance and gauge semantics, calculates physical residuals and tolerance-normalized residuals, and generates a set of valid gauges. The residual construction module can set weights based on measurement repeatability, gauge sensitivity, and historical reliability. If the number of valid gauges is insufficient, there are configuration conflicts, or the residuals contradict each other, the module outputs a "Disable Automatic Inverse Calculation" status and switches to manual verification, supplementary measurement, or only recording mode.

[0051] The constraint correction inverse calculation module is used to establish the response relationship between the gauge residual and the parameter to be corrected, and to solve for the correction amount of the process parameter under constraints. Specifically, the constraint correction inverse calculation module uses analytical sensitivity, finite difference sensitivity, calibration matrix or history identification model to establish the response relationship between the gauge residual and the parameter to be corrected, and calculates the correction amount through weighted least squares, optimization with regularization term or equivalent constraint solution method. The module simultaneously executes dead zone, amplitude limit, physical boundary, cumulative feed continuity, remaining margin and worm gear usable area constraints; if no feasible solution exists, the correction amount is not forcibly generated, but instead it is downgraded to adjusting measurement weights, narrowing the correction range, reallocating the tool shifting area, triggering dressing / tool ​​change or stopping machining.

[0052] The associated correction strategy module is used to select and set strategies based on the parent entity and semantics of the parameter to be corrected, determine the propagation scope, write the target parameter, and synchronously update dependent parameters. The propagation scope can be the current workpiece, the current feature, the current process, related processes under the same machining type, or related processes using the same tool. After the parameter is written, dependent parameters and visibility are updated synchronously, and a record of "correction source - target object - propagation scope" is retained to avoid updating only local interface values.

[0053] The future update and closed-loop verification module is used to allocate feed corrections to strokes that have not yet been executed, recalculate future cumulative feed and remaining worm gear region, and perform multiple constraint checks. Specifically, the future update and closed-loop verification module allocates the obtained feed corrections to subsequent strokes, recalculates the cumulative feed, and recalculates the remaining worm gear region based on the new tool shifting start point, step distance, and used area. The closed-loop verification checks at least: parameter boundaries, cumulative feed monotonicity, remaining margin, feasibility of tool shifting position, consistency of associated parameters, whether the predicted residual decreases as expected, and whether the data required for NC generation is complete. Only after the verification passes can NC regeneration be allowed.

[0054] The NC regeneration and versioning module is used to invoke the CAM's NC generation mechanism to generate new NC programs and perform version management, recording, and rollback operations. Specifically, the NC regeneration and versioning module calls the CAM's NC generation mechanism to generate and correct new programs corresponding to the revised process state, and records the old version, new version, correction trigger reason, predicted and measured vectors, residuals, parameter changes, future feed plans, and remaining worm gear area. If NC generation, syntax checking, or process verification fails, the previous valid parameter snapshot and NC version are restored. For operating modes requiring manual approval, the module first generates a pending approval plan, and replaces the effective version after approval.

[0055] like Figure 1 As shown, this embodiment describes a gear machining correlation correction and CNC program closed-loop regeneration method based on gear grinding stroke state prediction and measured residual, which includes the following steps.

[0056] S1. Acquire and parse gear measurement data. Monitor or read gear measurement files, parse the evaluation items output by the measuring equipment into records containing measurement keys, values, units, measurement configurations, time, and source files, and mark the validity and configuration consistency of the data. For measurement data that is not numerical, not finite, contains duplicate keys, or lacks key configurations, mark its validity and prevent it from being directly used for parameter correction; the measurement file retains the original file or summary identifier to establish a traceability relationship between measurement data, correction actions, and NC versions.

[0057] In this step, the gear grinding closed-loop control system monitors or reads the gear measurement file. The measuring device outputs the gear measurement file during its stroke or after workpiece machining. This file contains the measuring device's evaluation items for each gear under test. The control system parses the evaluation items output by the measuring device into data records containing "measurement key - value - unit - measurement configuration - time - source file". The measurement configuration includes at least information such as measurement radius or tooth tip height, number of teeth spanned, ball or roller diameter, and measurement direction.

[0058] For the parsed data records, the control system marks their validity. Specifically: when the data is a non-numerical type, the numerical value is infinite (such as infinity or non-numerical), the same measurement key appears repeatedly with inconsistent values, or key configuration items are missing (such as the number of teeth spanned, measurement radius), the control system marks the data as invalid, preventing it from being directly used for parameter correction. At the same time, the control system retains the original measurement file or generates a summary identifier to facilitate traceability between subsequent corrections and NC versions.

[0059] Preferably, the measurement file is a gear measurement data exchange format conforming to the ISO 18653 standard, and the control system parses it according to this format specification.

[0060] S2, Perform gauge semantic mapping. Maintain a semantic dictionary between the measuring device identifier and the internal gauge identifier of the CAM. Based on the semantic dictionary, map the measurement data to a unified gauge semantic space to generate a set of actual measurements. The mapping includes comparison and unification of gauge type, unit, measurement direction, measurement radius or tooth tip height, number of teeth spanned, ball or roller diameter, applicable tooth profile, and allowable tolerance. Gauge semantic mapping also includes: when the measurement units are inconsistent but the conversion relationship can be determined, unify the units before comparison; when the number of teeth spanned, measurement radius, ball diameter, or measurement direction is inconsistent, the measurement item does not participate in the residual construction and is marked as a configuration conflict.

[0061] In this step, the control system maintains a semantic dictionary between the measuring equipment identifier and the internal gauge identifier of computer-aided manufacturing. Each dictionary entry contains at least the gauge type, unit, measurement direction, measurement radius or tooth tip height, number of teeth spanned, ball or roller diameter, applicable tooth profile, and permissible tolerance.

[0062] During mapping, the control system compares not only names but also configurations and units. When units are inconsistent but the conversion relationship can be determined (e.g., between millimeters and micrometers), the control system unifies the units before comparison. When the number of teeth spanned, measurement radius, ball diameter, or measurement direction is inconsistent, the measurement item is considered a configuration conflict and does not participate in subsequent residual construction. Through gauge semantic mapping, the control system maps all valid measurement data to a unified gauge semantic space, generating a set of actual measurements.

[0063] The core function of this step is to avoid data misuse caused by different naming conventions or configurations of the same gauge between the measuring equipment and the CAM. For example, when the "tooth thickness" in the measurement file corresponds to the tooth thickness across a specific number of teeth, while the "tooth thickness" in the CAM corresponds to the normal chord tooth thickness, the control system maps the two to their respective correct gauge identifiers through a semantic dictionary, without performing a direct comparison across gauge types.

[0064] In addition, the control system can automatically calculate the conversion value under the equivalent measurement configuration according to the gauge type and measurement configuration, but it is necessary to confirm that the conversion relationship is traceable and the error is acceptable before conversion.

[0065] S3 maintains the process plan and cumulative feed status. Saves incremental feed, cumulative feed, machining stage, allowable allowance, and current CNC version according to the grinding stroke; locks executed strokes to prevent changes, and creates or updates the initial process plan based on the total feed depth and number of strokes.

[0066] In this step, the control system saves the incremental feed, cumulative feed, machining stage, allowable margin, and current NC version according to the grinding stroke. The initial process plan can be derived from CAM process settings or allocated from the total depth of feed and number of strokes. The control system locks the executed strokes into an unchangeable state, meaning that the incremental feed and cumulative feed of the executed strokes remain unchanged.

[0067] The cumulative radial feed is obtained by summing the radial feed increments of each stroke:

[0068] in: For the first Incremental radial feed of the stroke, satisfying ; For the first Cumulative radial feed at the end of the stroke; The cumulative radial feed at the end of stroke 0 is defined as 0. The maximum allowed cumulative radial feed; Index for the current itinerary; This represents the total number of grinding strokes.

[0069] When the control system determines that the incremental feed of the current stroke has caused the cumulative feed to exceed... If this occurs, the system will reject the incremental feed and indicate a process plan error. Each correction only allows changes to the plan after the current stroke; already executed strokes remain unchanged.

[0070] S4, Update the virtual workpiece. Based on the current cumulative feed, perform equivalent material removal on the initial gear geometry to obtain a virtual workpiece corresponding to the current stroke state; the equivalent material removal is achieved by converting the radial cumulative feed into normal tooth thickness variation, and then mapping it into displacement coefficient variation.

[0071] In this step, the control system bases its decisions on the current cumulative radial feed. Equivalent material removal is performed on the initial gear geometry to obtain a virtual workpiece corresponding to the current travel state. The specific method of equivalent material removal is as follows: the radial cumulative feed is converted into the normal tooth thickness variation, and then mapped into the displacement coefficient variation.

[0072] Preferably, when the gear being ground is an involute cylindrical gear, the displacement coefficient of the virtual workpiece is calculated as follows:

[0073] in: This is the initial displacement coefficient (i.e., the displacement coefficient of the gear in its raw or heat-treated state). The conversion factor from radial feed to normal tooth thickness reduction is determined by calibration or process mechanism; Normal modulus; This is the normal pressure angle.

[0074] The virtual workpiece serves as the unified geometric reference for the multi-gauge prediction module. By mapping the stroke state to the geometric state of the virtual workpiece, the control system avoids the use of inconsistent material removal assumptions for each gauge. The virtual workpiece includes at least basic geometric parameters such as gear module, number of teeth, pressure angle, displacement coefficient, and tooth width.

[0075] S5, Perform multi-gauge prediction. On the same virtual workpiece, based on a unified geometric datum and gauge configuration, calculate the predicted values ​​of multiple gauges to form a multi-gauge prediction vector; the gauges include at least two of the following: cross tooth thickness, normal chord tooth thickness, normal tooth thickness, transverse tooth thickness, single ball, double ball, single roller, and double roller measurement dimensions.

[0076] In this step, the control system calculates the predicted values ​​of multiple gauges on the same virtual workpiece, based on a unified geometric datum and gauge configuration, forming a multi-gauge prediction vector. The gauges include at least two of the following: span tooth thickness, normal chord tooth thickness, normal tooth thickness, transverse tooth thickness, single-ball, double-ball, single-roller, and double-roller measurement dimensions.

[0077] Gauge prediction is based on geometric prediction functions Sure: .in: This is the initial set of gear geometry parameters (including module, number of teeth, pressure angle, etc.). The virtual workpiece displacement coefficient obtained in step S4; For the first Measurement configuration of each gauge (including measurement radius, number of teeth spanned, or diameter of the ball bar, etc.).

[0078] Each prediction is output along with its measurement configuration. The prediction set can be a subset selected based on tooth profile, gauge, and field measurement capabilities, but the number and independence of valid measurements used for reverse engineering should be sufficient to identify the parameter to be corrected. In one example of this implementation, when the module of the gear being ground... Number of teeth Pressure angle At that time, the multi-gauge prediction module calculates the tooth thickness across the virtual workpiece. Normal chord tooth thickness Normal tooth thickness As a prediction vector.

[0079] S6, Maintain worm gear single-tool status and remaining available area: Based on the worm gear's allowable tool movement range, starting position, step distance, number of machined workpieces, used area, dressing records, and safety margin, maintain the current tool movement position and remaining available area or set of areas. In this step, the control system maintains the current tool movement position and the remaining available range or set of ranges based on the worm gear's allowable tool movement range, starting position, step distance, number of processed workpieces, used area, dressing records, and safety margin.

[0080] For a forward unidirectional continuous tool movement, the remaining usable area is represented as follows: ,in and These are the lower and upper boundaries of the remaining available area, respectively. The new cutter position must be selected within the available area and must avoid overlapping with the used area and the reserved safety belt.

[0081] For reverse tool tracing, the control system calculates the coordinates after transformation. For segmented tool tracing, the control system represents the remaining available area as multiple non-overlapping intervals and selects from the intervals that satisfy the machining stage, roughing and finishing zones, and safety margins. For topological diagonal tool tracing, the control system discretizes the two-dimensional or multi-segment trajectory into a set of indexed available positions.

[0082] In this step, the control system also records the grinding wheel dressing history, including the number of dressings, the amount removed during dressing, and the outer diameter of the grinding wheel after dressing, as a reference for judging the remaining usable area.

[0083] S7. Construct multi-gauge residuals. Pair the actual measurement set under the same travel and the same gauge semantics with the multi-gauge prediction vector, calculate the physical residual and the tolerance normalized residual, and generate a valid gauge set; if the number of valid gauges is insufficient or there is a configuration conflict, output a state that prohibits automatic reverse calculation.

[0084] In this step, the control system pairs the measured values ​​with the predicted values ​​under the same stroke and the same gauge semantics, and calculates the physical residual and the tolerance normalized residual.

[0085] Physical residuals and tolerance normalized residuals They are respectively:

[0086] in: For the first Itinerary No. Predicted values ​​for individual gauges; These are the corresponding measured values; For the first Normalized scale or tolerance of a gauge .

[0087] The control system also generates a set of valid gauges. It consists of gauges with consistent configuration, valid values, and capable of participating in identification. The control system sets confidence weights based on measurement repeatability, gauge sensitivity, and historical reliability. , Preferably, the normalized weights of each gauge satisfy the following conditions: .

[0088] The automatic correction criterion is the maximum absolute normalized residual. :

[0089] when While maintaining the current parameters, only update the status record; when Then, the constraint correction and inverse calculation are performed. The preset dead zone threshold can be set according to measurement repeatability and process tolerance; a typical value is [value to be filled in]. .

[0090] If the number of valid gauges is insufficient, there are configuration conflicts, or the residuals are contradictory, the system will output "Automatic reverse calculation prohibited" status and switch to manual confirmation, supplementary measurement, or recording mode.

[0091] Preferably, when the normalized residual directions of multiple gauges are in obvious conflict, the system checks the gauge configuration, measurement file version, and virtual workpiece status, and requests a retest if necessary.

[0092] S8, Constraint Correction Inverse Solution. Establish the response relationship between the gauge residual and the parameter to be corrected. Under the constraints of single correction limit, parameter physical boundary, cumulative feed continuity, remaining machining allowance, and worm gear usable area, solve for the correction amount of the process parameter; if there is no feasible solution, perform downgrade processing.

[0093] In this step, the control system establishes the response relationship between the gauge residual and the parameter to be corrected, and solves for the correction amount of the process parameter under multiple constraints.

[0094] The control system establishes a sensitivity matrix using analytical sensitivity, finite difference sensitivity, calibration matrix, or historical identification model. Sensitivity matrix Indicates the first Itinerary No. The parameter for the first The rate of change of the normalized predictor. Preferably, when the number of effective gauges is less than the number of parameters to be corrected, or when the sensitivity matrix is ​​ill-conditioned, the control system reduces the number of parameters to be corrected, adds regularization terms, corrects only the highest priority parameters, or requests additional measurements, without forcing a solution for full parameter correction.

[0095] The constraint correction inverse problem obtains the correction vector by solving a weighted least squares optimization problem with regularization terms. :

[0096] in: For the first The credibility weight of each metric; For the first Itinerary No. The parameter for the first Sensitivity matrix of each normalized predictor; For the first The current correction amount for each process parameter; For the first Regularization scale for each parameter ; The regularization coefficient is . ; Tolerance normalization residuals; For a set of valid gauges; The total number of process parameters to be corrected, which is a positive integer.

[0097] Each corrected parameter must simultaneously satisfy both the physical boundary and the single correction limit:

[0098] in: and The first The lower and upper limits allowed for each parameter; For the first The maximum single correction magnitude of each parameter. .

[0099] The control system simultaneously enforces dead zone, amplitude limit, physical boundary, cumulative feed continuity, remaining margin, and worm gear availability zone constraints. The cumulative feed continuity constraint is as follows:

[0100] in, For the revised future Cumulative radial feed; To maintain the maximum permissible cumulative radial feed, the cumulative feed of the executed stroke remains unchanged; To be allocated to the future The cumulative feed adjustment amount of the stroke; For the first Cumulative radial feed of the stroke; This represents the total number of grinding strokes.

[0101] That is, the revised future cumulative feed should monotonically remain unchanged and not exceed the maximum permissible cumulative radial feed. The cumulative feed of the executed stroke remains unchanged.

[0102] The usable area constraint of the worm gear is:

[0103] in: and These are the current cutter position and the planned next cutter position, respectively. For the next step of the cutting tool movement; The remaining available area before selecting the next cutter position; and These are the lower and upper bounds of the remaining available area, respectively. To allow for a safety margin, for reverse or segmented tool movement, direction transformation or interval set calculations are used. The proposed next tool movement position. Must belong to the remaining availability zone before the update The updated remaining availability zone is from start, For safety margin, .

[0104] If no feasible solution exists, the control system performs a degradation process: adjusting measurement weights, narrowing the correction range, reallocating the tool shifting zone, triggering trimming / tool ​​change, or stopping machining, without forcibly generating a correction amount. In one example of this implementation, Current cumulative feed Corrected cumulative feed , All do not exceed The constraints are satisfied.

[0105] S9, Execute the correlation correction strategy. Based on the parent entity and semantic selection settings of the parameter to be corrected, determine the propagation range of the correction effect, write the target parameter and synchronously update the dependent parameters; the propagation range includes the current workpiece, the current feature, the current process, related processes under the same processing type, or related processes using the same tool.

[0106] In this step, the control system determines the propagation range of the correction effect based on the parent entity and semantic selection setting strategy of the parameter to be corrected, writes the target parameter, and synchronously updates the dependent parameters.

[0107] The scope of application can be: the current workpiece, the current feature, the current process, related processes under the same machining type, or related processes using the same tool. The control system automatically determines the propagation scope based on the CAM data structure hierarchy to which the parameter belongs. For example, when the correction parameter belongs to the cutting parameter at the tool level, the control system propagates the correction to all related processes using the same tool; when the correction parameter only belongs to the feed rate of the current process, the control system only modifies the parameters of the current process.

[0108] After the parameters are written, the control system synchronously updates the dependent parameters and visible status, and retains a record of "correction source - target - propagation range" to avoid inconsistencies caused by modifying only local interface values ​​without updating related processes and dependent parameters. In one example of this implementation, the control system writes the feed correction to the current gear grinding process and updates the corresponding dependent parameters based on the same machining type and shared tool relationships.

[0109] S10, Update Future Plans and Closed-Loop Verification. Allocate the obtained feed correction to strokes that have not yet been executed, and recalculate the future cumulative feed; recalculate the remaining worm gear area based on the new tool shifting position, step distance, and used area; at least verify the parameter boundaries, cumulative feed monotonicity, remaining margin, tool shifting position feasibility, consistency of associated parameters, and convergence of predicted residuals; only proceed to the CNC regeneration step after verification is passed.

[0110] In this step, the control system allocates the obtained feed correction to the strokes that have not yet been executed and recalculates the future cumulative feed; it also recalculates the remaining worm gear area based on the new tool position, step distance, and used area.

[0111] Closed-loop verification should at least check: parameter boundaries (whether all corrected parameters meet the upper and lower limit constraints) and cumulative feed monotonicity (whether the future cumulative feed will remain monotonically constant and not exceed the limit). ), remaining margin (whether it meets the minimum margin requirement), feasibility of tool shifting position (whether the new tool shifting position belongs to the usable area before the update), consistency of associated parameters (whether the parameters under the same tool or machining type are consistent), and whether the predicted residual decreases as expected.

[0112] The predicted improvement conditions are:

[0113] in: This is the weighted sum of squared residuals before correction; This is the weighted sum of squared residuals after the candidate corrections and re-prediction; To predict the acceptance coefficient for improvement, Typical value ; For the first The credibility weight of each metric; Tolerance normalization residuals; After applying candidate modifications, the first Itinerary No. The tolerance normalized residuals obtained from the re-prediction of each gauge; This is the set of valid gauges.

[0114] Only after all checks pass can the NC regeneration step be initiated. (If the maximum number of retries is preset...) If all constraints cannot be met simultaneously, automatic correction will terminate, and the system will either maintain or roll back to the previous valid state. Preferably, when the verification fails, the control system will execute the following in order of priority: reduce the correction amount, reallocate the tool shifting position, trigger trimming / tool ​​change, or proceed to manual confirmation.

[0115] S11, perform CNC regeneration and version management. Call the CAM CNC generation mechanism to generate a new CNC program corresponding to the corrected process state, and record the old version, new version, correction trigger reason, predicted and measured vectors, residuals, parameter changes, future feed plan and remaining worm gear area; if CNC generation, syntax check or process verification fails, restore the previous valid parameter snapshot and CNC version.

[0116] In this step, the control system calls the NC generation mechanism of CAM to generate a new NC program corresponding to the corrected process state, and records the old version, new version, correction trigger reason, predicted and measured vectors, residuals, parameter changes, future feed plan and remaining worm gear area.

[0117] If NC generation, syntax checking, or process verification fails, the control system restores the previous valid parameter snapshot and NC version. For operating modes requiring manual approval, the control system first generates a plan to be approved, and then replaces the effective version after approval.

[0118] Each closed-loop calculation establishes a unique record, which includes at least the workpiece identifier, process identifier, tool identifier, stroke index, measurement file summary, gauge configuration, predicted value, measured value, residual, sensitivity model version, parameters before and after correction, future feed plan, remaining worm gear area, verification result, and NC file summary. The new NC only replaces the effective version after generation and verification are completed; in case of failure, the snapshot before correction is restored.

[0119] In one example of this implementation, when the weighted residual objective satisfies When the parameter boundaries, feed monotonicity, and worm gear usable area all pass the verification, the control system generates a new NC version. This record can be used for manual review, subsequent sensitivity calibration, and quality traceability.

[0120] In summary, this implementation method, through the sequential execution of steps S1 to S11, constructs the cumulative feed state, updates the virtual workpiece, predicts multiple gauges, maps semantic measurements, reverses residuals, propagates CAM associations, updates the future feed and worm gear area synchronously, and verifies and regenerates NC in sequence, forming a complete closed-loop control system for gear grinding.

[0121] In a specific application scenario of this embodiment: Suppose that an involute cylindrical gear is produced by continuous generating grinding, and the total number of strokes is... Normal module Normal pressure angle Initial displacement coefficients Conversion factor for radial feed to normal tooth thickness reduction The initial incremental feed is: , , , The cumulative feed is , , , After the second stroke, the virtual workpiece displacement coefficient is approximately... The multi-gauge prediction module obtains three gauge prediction values ​​on the virtual workpiece: tooth thickness. Normal chord tooth thickness Normal tooth thickness The semantic mapping module confirms that the measured data uses the same number of teeth spans, measurement radius, and unit to obtain the measured values. , , If the normalized scales of the three gauges are respectively , , The physical residual is , , The normalized residual is approximately , , Set a dead zone threshold. ,but Then, by performing constraint correction and inverse calculation, the increase in the subsequent total radial feed is obtained. The corrected cumulative feed is , If the remaining available area before the worm gear update is Current position of the cutting tool The next step is... Safety margin Then determine the position of the next slash. Remaining available areas after the update If all checks pass, a new NC version will be generated.

[0122] Specifically, regarding the acquisition of the sensitivity matrix, this embodiment is not limited to a single acquisition method. Those skilled in the art can choose according to actual conditions: analytically differentiate based on gear geometry and grinding motion model; apply small perturbations to the parameters to be corrected on a virtual workpiece, calculate the changes in the predicted values ​​of each gauge using finite difference and divide by the corresponding normalized scale; perform offline calibration using standard parts, historical workpieces, or trial grinding data; or, when stability and data quality requirements are met, recursively update based on the measured changes before and after each correction.

[0123] Regarding data anomaly handling, when measurement units are inconsistent but the conversion relationship can be determined, the system will first unify the units before comparison; when the number of teeth spanned, measurement radius, ball diameter, or measurement direction is inconsistent, the measurement item will not be included in the valid gauge set. When multiple gauge residual directions conflict significantly, the system checks the gauge configuration, measurement file version, and virtual workpiece status, and requests a remeasurement if necessary. When the remaining worm gear zone is insufficient, the system prioritizes reallocating the tool shifting position, and then triggers trimming or tool change, without bypassing resource constraints by expanding the allowable zone.

[0124] Regarding NC version traceability, a unique record is created for each closed-loop calculation, including at least the workpiece identifier, process identifier, tool identifier, stroke index, measurement file summary, gauge configuration, predicted value, measured value, residual, sensitivity model version, parameters before and after correction, future feed plan, remaining worm gear area, verification result, and NC file summary. A new NC only replaces the effective version after generation and verification; in case of failure, the snapshot before correction is restored. This record can be used for manual review, subsequent sensitivity calibration, and quality traceability.

[0125] This embodiment of the gear machining correlation correction and CNC program closed-loop regeneration method and system based on gear grinding stroke state prediction and measured residual has the following advantages compared with the prior art.

[0126] (1) Achieve consistency judgment of multiple gauges. By comparing the cross tooth thickness, chord tooth thickness, normal tooth thickness and ball bar measurement under the same virtual workpiece and the same measurement configuration, it is possible to identify single gauge abnormalities or configuration inconsistencies and reduce the probability of error correction.

[0127] (2) Transform stroke prediction into a machining state that can be used in a closed loop. Cumulative feed is no longer just a process record, but a state variable that drives virtual workpiece and gauge prediction, so that each measurement can correspond to a clear stroke reference.

[0128] (3) Coordinate precision correction with worm gear resource management. The correction back-calculation takes into account the remaining margin, future feed continuity and worm gear availability area to avoid tool overflow, overlapping usage areas or waste of tool resources caused by machining dimension improvement schemes.

[0129] (4) Ensure consistency of CAM associated parameters. Select the propagation range based on the parent entity, process, machining type and tool relationship, and update the dependent parameters to reduce process conflicts caused by local value changes.

[0130] (5) Improve the stability of automatic correction. Suppress frequent back corrections or divergence caused by measurement noise by using dead zone, weight, regularization term, amplitude limit, prediction improvement conditions and maximum number of retries.

[0131] (6) Enhance the reliability and traceability of NC. A new NC is generated only after the geometry, process, resources and associated constraints are passed; each version can be traced back to the corresponding prediction, measurement, residual and parameter snapshot, and can be rolled back in case of failure.

[0132] (7) It has good scalability. The gauge set, inverse calculation algorithm, tool shifting method and parameter propagation strategy can all be replaced. As long as the unified semantics, constraint inverse calculation and closed-loop verification relationship are maintained, it can be applied to different tooth profiles, machine tools and measuring equipment.

[0133] The above-described embodiments are merely preferred embodiments provided to fully illustrate the present invention, and the scope of protection of the present invention is not limited thereto. Equivalent substitutions or modifications made by those skilled in the art based on the present invention are all within the scope of protection of the present invention. The scope of protection of the present invention is defined by the claims.

Claims

1. A method for gear machining correlation correction and CNC program closed-loop regeneration based on gear grinding stroke state prediction and measured residual, characterized in that: include: S1, Acquire and parse gear measurement data: Monitor or read gear measurement files, parse the evaluation items output by the measuring equipment into records containing measurement keys, values, units, measurement configurations, time and source files, and mark the validity of the data and the consistency of the configuration; S2, Perform gauge semantic mapping: Maintain a semantic dictionary between the measuring device identifier and the internal gauge identifier of the CAM, and map the measurement data to a unified gauge semantic space based on the semantic dictionary to generate a set of actual measurements; the mapping includes comparison and unification of gauge type, unit, measurement direction, measurement radius or tooth tip height, number of teeth spanned, ball or roller diameter, applicable tooth profile and allowable tolerance; S3, Maintain process plan and cumulative feed status: Save incremental feed, cumulative feed, machining stage, allowable margin and current CNC version according to grinding stroke; Lock the executed strokes so they cannot be changed, and create or update the initial process plan based on the total feed depth and number of strokes; S4, Update the virtual workpiece: Perform equivalent material removal on the initial gear geometry based on the current cumulative feed to obtain a virtual workpiece corresponding to the current stroke state; the equivalent material removal is achieved by converting the radial cumulative feed into the normal tooth thickness change, and then mapping it into the displacement coefficient change; S5, Perform multi-gauge prediction: On the same virtual workpiece, based on a unified geometric datum and gauge configuration, calculate the predicted values ​​of multiple gauges to form a multi-gauge prediction vector; the gauges include at least two of the following: cross tooth thickness, normal chord tooth thickness, normal tooth thickness, transverse tooth thickness, single ball, double ball, single roller, and double roller measurement dimensions. S6, Maintain worm gear single-tool status and remaining available area: Based on the worm gear's allowable tool movement range, starting position, step distance, number of machined workpieces, used area, dressing record, and safety margin, maintain the current tool movement position and remaining available area or set of areas; S7, Constructing multi-gauge residuals: Pair the actual measurement set under the same travel and the same gauge semantics with the multi-gauge prediction vector, calculate the physical residual and the tolerance normalized residual, and generate an effective gauge set; If the number of valid gauges is insufficient or there is a configuration conflict, the output will show an "automatic reverse calculation prohibited" status. S8, Constraint Correction Inverse Calculation: Establish the response relationship of the gauge residual to the parameter to be corrected, and solve the correction amount of the process parameter under the constraints of single correction limit, parameter physical boundary, cumulative feed continuity, remaining machining allowance and worm gear usable area; If no feasible solution is found, then a downgrade process is performed; S9, Execute the correlation correction strategy: Based on the parent entity and semantic selection of the parameter to be corrected, determine the propagation range of the correction effect, write the target parameter and update the dependent parameters synchronously; the propagation range includes the current workpiece, the current feature, the current process, related processes under the same processing type, or related processes using the same tool; S10, Update Future Plans and Closed-Loop Verification: Allocate the obtained feed correction to strokes that have not yet been executed, and recalculate the future cumulative feed; recalculate the remaining worm gear area based on the new tool shifting position, step distance, and used area; at least verify the parameter boundaries, cumulative feed monotonicity, remaining margin, tool shifting position feasibility, consistency of associated parameters, and convergence of predicted residuals; only proceed to the CNC regeneration step after verification is passed. S11, Perform CNC regeneration and version management: Call the CAM CNC generation mechanism to generate a new CNC program corresponding to the corrected process state, and record the old version, new version, correction trigger reason, predicted and measured vectors, residuals, parameter changes, future feed plan and remaining worm gear area; if CNC generation, syntax check or process verification fails, restore the previous valid parameter snapshot and CNC version.

2. The method for gear machining correlation correction and CNC program closed-loop regeneration based on gear grinding stroke state prediction and measured residual as described in claim 1, characterized in that: In step S2, the gauge semantic mapping further includes: when the measurement units are inconsistent but the conversion relationship can be determined, the units are unified before comparison; when the number of teeth spanned, the measurement radius, the diameter of the ball bar, or the measurement direction are inconsistent, the measurement item does not participate in the residual construction and is marked as a configuration conflict.

3. The method for gear machining correlation correction and CNC program closed-loop regeneration based on gear grinding stroke state prediction and measured residual as described in claim 1, characterized in that: In step S4, based on the current cumulative radial feed... Equivalent material removal is performed on the initial gear geometry to obtain the displacement coefficient of the virtual workpiece. : in: These are the initial displacement coefficients; This is the conversion factor for radial feed to normal tooth thickness reduction; Normal modulus; It is the normal pressure angle; The cumulative radial feed The result is obtained by accumulating the radial feed increments of each stroke: in, For the first The stroke increments radial feed, and satisfy... ; The cumulative radial feed at the end of stroke 0 is defined as 0. The maximum allowed cumulative radial feed; This is the index for the current itinerary.

4. The method for gear machining correlation correction and CNC program closed-loop regeneration based on gear grinding stroke state prediction and measured residual as described in claim 1, characterized in that: In step S7, the physical residual and tolerance normalized residuals They are respectively: in: For the first Itinerary No. Predicted values ​​for individual gauges; These are the corresponding measured values; For the first Normalized scale or tolerance of a gauge; The set of valid gauges It consists of gauges with consistent configuration, valid values, and capable of participating in identification; The automatic correction is based on the maximum absolute normalized residual: when Keep the current parameters; when When entering the constraint correction inverse calculation, where This is the preset dead zone threshold.

5. The method for gear machining correlation correction and CNC program closed-loop regeneration based on grinding stroke state prediction and measured residual as described in claim 1, characterized in that: In step S8, the constraint correction inverse problem is solved by solving a weighted optimization problem with regularization terms to obtain the correction vector. : in: For the first The credibility weight of each metric; For the first Itinerary No. The parameter for the first Sensitivity matrix of each normalized predictor; For the first The current correction amount for each process parameter; For the first Regularization scale for each parameter; The regularization coefficient is used. Tolerance normalization residuals; For a set of valid gauges; This represents the total number of process parameters to be corrected. And each of the corrected parameters satisfies: in: and The first The lower and upper limits allowed for each parameter; For the first The maximum single correction magnitude for each parameter.

6. The method for gear machining correlation correction and CNC program closed-loop regeneration based on gear grinding stroke state prediction and measured residual as described in claim 5, characterized in that: The sensitivity matrix Obtained through any of the following methods: Analytical differentiation is performed based on gear geometry and the grinding motion model; Apply a small perturbation to the parameter to be corrected on the virtual workpiece, calculate the change of each gauge prediction value with finite difference, and divide by the corresponding normalized scale; Offline calibration is performed using standard parts, historical workpieces, or trial grinding data. Updated iteratively based on the measured changes before and after each revision; When the number of effective gauges is less than the number of parameters to be corrected, or when the sensitivity matrix is ​​ill-conditioned, reduce the number of parameters to be corrected, add regularization terms, correct only the highest priority parameters, or request additional testing.

7. The method for gear machining correlation correction and CNC program closed-loop regeneration based on grinding stroke state prediction and measured residual as described in claim 1, characterized in that: In step S8, the cumulative feed continuity constraint is: in, For the revised future Cumulative radial feed; To maintain the maximum permissible cumulative radial feed, the cumulative feed of the executed stroke remains unchanged; To be allocated to the future The cumulative feed adjustment amount of the stroke; For the first Cumulative radial feed of the stroke; This represents the total number of grinding strokes. The usable area constraint of the worm gear is: in: and These are the current cutter position and the planned next cutter position, respectively. For the next step of the cutting tool movement; The remaining available area before selecting the next cutter position; and These are the lower and upper bounds of the remaining available area before selecting the next cutter position, respectively. For safety margins, for reverse or segmented tool shifting, direction transformation or interval set calculation is used.

8. The method for gear machining correlation correction and CNC program closed-loop regeneration based on grinding stroke state prediction and measured residual as described in claim 1, characterized in that: In step S10, the closed-loop verification further includes predicting improvement conditions: in: This is the weighted sum of squared residuals before correction; This is the weighted sum of squared residuals after the candidate corrections and re-prediction; To predict the improvement acceptance coefficient, and ; For the first The credibility weight of each metric; Tolerance normalization residuals; For the first Itinerary No. The tolerance normalized residuals obtained from the re-prediction of each gauge; For a set of valid gauges; If the preset maximum number of retries is exceeded If all constraints cannot be met simultaneously, automatic correction will terminate, and the system will either maintain or roll back to the previous valid state.

9. A gear machining correlation correction and CNC program closed-loop regeneration system based on gear grinding stroke state prediction and measured residual, used to implement the gear machining correlation correction and CNC program closed-loop regeneration method based on gear grinding stroke state prediction and measured residual as described in any one of claims 1-8, characterized in that: include: The measurement result acquisition module is used to monitor or read gear measurement files and parse the evaluation items output by the measuring equipment into records containing measurement keys, values, units, measurement configurations, time, and source files. The gauge semantic mapping module is used to maintain a semantic dictionary between the measurement equipment identifier and the gauge identifier inside the CAM, map the measurement data to a unified gauge semantic space, and generate a set of actual measurements. The process planning and cumulative feed module is used to save incremental feed, cumulative feed, machining stage, allowable allowance and current CNC version according to the grinding stroke, and to lock the executed stroke; The virtual workpiece update module is used to perform equivalent material removal on the initial gear geometry based on the current cumulative feed, so as to obtain a virtual workpiece corresponding to the current stroke state. The multi-gauge prediction module is used to calculate the predicted values ​​of multiple gauges on the same virtual workpiece, forming a multi-gauge prediction vector. The Single Tool Status and Remaining Available Area module is used to maintain the current tool position and the remaining available range or set of ranges for the worm gear; The residual construction module is used to pair measured values ​​and predicted values ​​under the same travel and the same gauge semantics, calculate physical residuals and tolerance normalized residuals, and generate a set of valid gauges. The constraint correction inverse calculation module is used to establish the response relationship of the gauge residual to the parameter to be corrected, and to solve the correction amount of the process parameter under constraint conditions. The associated correction strategy module is used to set the strategy based on the parent entity and semantic selection of the parameter to be corrected, determine the propagation scope, write the target parameter and synchronously update the dependent parameters. Future plans include updating and implementing a closed-loop verification module to allocate feed corrections to strokes that have not yet been executed, recalculate future cumulative feeds and remaining worm gear area, and perform multiple constraint verifications. The NC regeneration and versioning module is used to call CAM's NC generation mechanism to generate new NC programs and perform version management, recording, and rollback operations.

10. The gear machining correlation correction and CNC program closed-loop regeneration system based on grinding stroke state prediction and measured residual as described in claim 9, characterized in that: The mapping performed by the gauge semantic mapping module includes the comparison and unification of gauge type, unit, measurement direction, measurement radius or tooth tip height, number of teeth spanned, ball or roller diameter, applicable tooth profile and allowable tolerance; When measurement configurations are not equivalent, prevent the merging and use of corresponding measurement data; The constraints executed by the constraint correction and reverse calculation module include: single correction limit, parameter physical boundary, cumulative feed monotonicity, remaining machining allowance, and worm gear available area constraint; the worm gear available area constraint supports unidirectional continuous tool movement, reverse tool movement, segmented tool movement, and topological diagonal tool movement.

Images