An adaptive scheduling and collaborative control system for a multi-process gear forging and finishing production line

Through an adaptive scheduling and collaborative control system, physical signals of forging billets are collected piece by piece, deformation is calculated and predicted, and process adjustment instructions are generated. This enables personalized scheduling and collaborative control of gear forging and finishing production lines, solving the problems of quality variation and high scrap rate, and improving production efficiency and finished product consistency.

CN123113767APending Publication Date: 2026-11-17LUAN HELI MASCH MFG CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611451033.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-18
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing technologies cannot effectively address the quality variations caused by differences in materials, temperature, and deformation history among different forgings within the same batch in the scheduling and control of gear forging and finishing production lines. Furthermore, they lack a complete control loop from online sensing to real-time decision-making, resulting in high finishing scrap rates and low production line efficiency.

Method used

An adaptive scheduling and collaborative control system is adopted. The data acquisition module acquires the measured physical signals of the forging billet, the deformation prediction module calculates the predicted deformation, and the collaborative decision-making module generates process adjustment instructions, including active buffer control and quality gate mechanism, to realize personalized processing schemes for each piece and closed-loop linkage across processes.

Benefits of technology

It significantly reduced the scrap rate in finishing, improved production line efficiency, ensured product consistency and system adaptability, and overcame the model degradation problem in traditional solutions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123113767A_ABST
    Figure CN123113767A_ABST
Patent Text Reader

Abstract

This invention relates to the field of intelligent gear manufacturing technology, specifically to an adaptive scheduling and collaborative control system for a multi-process forging and finishing production line for gears. The system includes a data acquisition module that collects the geometric dimensions of the forging billet, cooling temperature gradient, internal defects, and final forging process parameters to generate a structured state feature dataset; a deformation prediction module that uses a prediction model to calculate the predicted deformation amount during subsequent finishing; a collaborative decision-making module that compares the predicted deformation amount with dual thresholds to generate a process adjustment instruction package containing forging allowance, holding time, and finishing compensation parameters; a process execution module that executes forging-end adjustments and / or finishing-end compensations and determines the buffer position; and a model iteration module that iteratively updates the prediction model using actual processing deviations as feedback samples. This invention achieves predictive closed-loop collaborative control throughout the forging-finishing process, improving gear machining accuracy and production line stability.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of intelligent gear manufacturing technology, and more specifically, to an adaptive scheduling and collaborative control system for a multi-process gear forging and finishing production line. Background Technology

[0002] As a core component of mechanical transmission systems, the manufacturing quality of gears directly affects the performance, lifespan, and reliability of the entire machine. The typical manufacturing process of gears involves multiple steps, including blanking, heating, multi-station forging, heat treatment, turning, gear hobbing, gear grinding, and inspection. Among these, the residual stress introduced during the forging stage is gradually released during subsequent finishing processes and is one of the main causes of tooth profile deviation, tooth direction deviation, and out-of-tolerance precision of the final product.

[0003] Currently, the scheduling and control of gear forging and finishing production lines generally adopt the following technical approach:

[0004] Firstly, there are general production scheduling solutions based on Manufacturing Execution Systems (MES) or Advanced Planning and Scheduling Systems (APS). These solutions primarily use work order delivery dates, equipment load, and material availability as constraints, generating scheduling plans through rule tables or optimization algorithms. However, existing ERP and MES systems in the forging industry suffer from problems such as large deviations in plan execution and a lack of coordination between forging and machining workshops. Their scheduling decisions are mainly based on the "time" dimension, failing to take into account the physical quality state of the forgings.

[0005] Secondly, there are quality control solutions for gear forging processes. These solutions reduce residual stress and its resulting deformation by optimizing forging die design, adjusting forging process parameters (such as final forging temperature and holding time), or improving heat treatment regimes. However, these solutions fall under the "process solidification" approach—ensuring batch quality consistency by setting a uniform process window, which cannot address the quality variations caused by differences in materials, temperatures, and deformation histories among different forgings within the same batch.

[0006] Third, there are thermal processing control solutions based on digital twins or big data. These solutions construct a full-process dataset and digital twin model through multi-sensor fusion, enabling the simulation and optimization of process parameters. However, their focus is on offline optimization in the "process parameter design" stage, and they have not yet established a complete control loop from "online sensing → piece-by-piece prediction → real-time decision-making → closed-loop iteration".

[0007] Therefore, an adaptive scheduling and collaborative control system for a multi-process gear forging and finishing production line is designed. Summary of the Invention

[0008] The purpose of this invention is to provide an adaptive scheduling and collaborative control system for a multi-process gear forging and finishing production line, so as to solve the problems mentioned in the background art.

[0009] To achieve the above objectives, the present invention aims to provide an adaptive scheduling and collaborative control system for a gear multi-process forging and finishing production line, comprising: a data acquisition module, set at the post-forging inspection station, for acquiring the measured physical signals of the current forging billet and performing local signal processing to generate and output a structured forging billet state feature dataset.

[0010] The measured physical signals include at least the forging billet geometric dimension signal, the temperature gradient distribution signal during the post-forging cooling stage, the internal defect echo signal, and the final forging process parameter signal.

[0011] The deformation prediction module, located in the edge industrial server, is connected to the data acquisition module. It is used to receive the forging billet state feature dataset, the batch information of the current workpiece, and the finished product accuracy level required by the order. It calls the deformation prediction model to calculate the predicted deformation of the current forging billet due to the release of residual stress during the subsequent finishing process.

[0012] The predicted deformation includes at least the estimated value of the tooth profile deviation and the estimated value of the tooth direction deviation after finishing. The input feature vector of the deformation prediction model includes at least the measured values ​​of the outer diameter of the forging billet, the measured values ​​of the inner hole of the forging billet, the end face runout value of the reference surface of the tooth billet, the characteristic value of the ultrasonic echo shrinkage and porosity probability, the characteristic value of the infrared temperature gradient variance of the post-forging cooling, the characteristic value of the temperature time-domain attenuation rate, the peak value of the final forging press pressure, the final forging holding time, and the historical statistical average value of the residual stress of the current batch, as well as the finished tooth profile accuracy grade and the finished tooth direction accuracy grade required by the order. The model iteration module uses the deviation value between the measured value of the finished tooth profile accuracy and the measured value of the tooth direction accuracy of the current workpiece after finishing and the predicted deformation as feedback samples to iteratively update the internal weights of the deformation prediction model.

[0013] The input feature vector of the deformation prediction model includes the time-domain decay feature value of the temperature gradient distribution signal.

[0014] The collaborative decision-making module, located in the edge industrial server, communicates with the deformation prediction module and is used to compare the predicted deformation amount with preset first threshold A and second threshold B, and generate corresponding process adjustment instruction packages based on the comparison results.

[0015] The active buffer control module is communicatively connected to the collaborative decision-making module and the execution mechanism of the finishing buffer area. When the predicted deformation exceeds the preset buffer intervention threshold, it sends an environmental control command to the execution mechanism of the buffer area to adjust the cooling rate of the workpiece in the buffer area.

[0016] The actuator of the finishing buffer area includes a heat insulation cover drive motor, a variable frequency cooling fan, and a zoned stacking and positioning device; the environmental control command includes the target value of the opening and closing angle of the heat insulation cover, the target value of the speed of the variable frequency cooling fan, and the zoned stacking density level. The actuator adjusts the temperature field distribution and cooling rate in the buffer area in a closed loop according to the environmental control command.

[0017] The process execution module is communicatively connected to the collaborative decision-making module and the finishing buffer area, and is set in the CNC of the forging press and / or the CNC of the finishing machine tool. It is used to receive and parse the process adjustment instruction package, execute the forging end adjustment action and / or finishing end compensation action in the process adjustment instruction package, and adjudicate and allocate buffer space conflict requests of the finishing machine tool according to the preset arbitration rules.

[0018] The quality access control module is communicatively connected to the collaborative decision-making module and the process execution module. It is used to intercept the current workpiece from flowing downstream and generate a return command when the predicted deformation exceeds the preset rework judgment threshold, so as to return the current workpiece to the re-inspection station. It is also used to route the current workpiece to the roughing process and reallocate the subsequent cutting allowance when the predicted deformation exceeds the preset downgrade judgment threshold but the interception is not triggered.

[0019] The model iteration module, located in the edge industrial server, is communicatively connected to the process execution module. It is used to receive the execution report containing actual processing parameters returned by the process execution module, and to use the deviation between the measured data of the finished product accuracy after finishing and the predicted deformation amount as feedback samples to iteratively update the internal weights of the deformation prediction model.

[0020] As a further improvement to this technical solution, the forging billet state feature dataset includes at least the following fields: workpiece unique identifier, measured value of forging billet outer diameter, measured value of forging billet inner hole, runout value of tooth billet reference surface end face, characteristic value of ultrasonic echo shrinkage and porosity probability, characteristic value of infrared temperature gradient variance after forging cooling, characteristic value of temperature time-domain attenuation rate, peak value of final forging press pressure, final forging holding time, and historical statistical average value of residual stress in the current batch.

[0021] As a further improvement to this technical solution, the first threshold A and the second threshold B are dynamic thresholds, which are generated in real time by superimposing a preset safety offset with the statistical average of the residual stress-deformation of the historical forging billets in the current processing batch as the benchmark value; the A and B threshold values ​​of the current batch are automatically refreshed once after processing N workpieces, where N is an integer greater than or equal to 1.

[0022] For example: , ,in This represents the statistical average of residual stress-deformation for the current batch. The standard deviation of residual stress-deformation for this batch is given. , Preset safety factor (preferred) =0.8, =1.5).

[0023] As a further improvement to this technical solution, the specific logic of the collaborative decision-making module in generating the corresponding process adjustment instruction package based on the comparison results is as follows: when the predicted deformation amount is ≤ A, an instruction to maintain the standard forging allowance parameter is generated, the current workpiece is marked as a normal part and transferred to the finishing buffer area.

[0024] When A < the predicted deformation amount ≤ B, an instruction package containing the adjusted target forging allowance setting value and / or forging holding time correction value is generated and sent to the forging press CNC to control the final forging pressure stop position of the subsequent workpieces to be forged in this batch to increase the single-sided forging allowance, and / or adjust the holding time of the forging press; the current workpiece is transferred to the finishing buffer area as a normal part.

[0025] When the predicted deformation is greater than B, the quality gate module is triggered to perform an interception judgment: if the interception is successful, a reflow command is generated to return the current workpiece to the re-inspection station; if the interception is unsuccessful, a high-risk label is marked for the current workpiece, and the instruction package includes recommended parameters for finishing feed compensation, grinding wheel linear speed correction value and coolant flow rate correction value, reducing the cutting feed rate of the current workpiece in finishing, lowering the grinding wheel linear speed and increasing the coolant flow rate.

[0026] As a further improvement to this technical solution, the process adjustment instruction package includes the target forging allowance setting value, the forging holding time correction value, the risk marking label, the recommended parameters for finishing feed compensation, the grinding wheel linear speed correction value, and the coolant flow rate correction value.

[0027] As a further improvement to this technical solution, the process execution module includes: a margin adjustment execution unit, which is set in the CNC of the forging press and is communicatively connected to the collaborative decision module. It is used to receive and parse the target forging margin setting value and the forging holding time correction value in the process adjustment instruction package, control the pressing stop position of the main oil cylinder according to the target forging margin setting value to adjust the single-sided forging margin of the subsequent workpieces to be forged in this batch, and / or control the holding time of the forging press according to the forging holding time correction value, and generate a first execution report containing the actual final forging pressing amount and the actual holding time after forging is completed, and send it back to the model iteration module.

[0028] The compensation execution unit, located in the CNC of the finishing machine tool, is communicatively connected to the collaborative decision-making module. It receives and parses the recommended finishing feed compensation parameters, grinding wheel linear speed correction value, and coolant flow correction value in the process adjustment instruction package. Based on the recommended finishing feed compensation parameters, it sends the feed rate override value in real time via industrial Ethernet to execute cutting. Based on the grinding wheel linear speed correction value, it adjusts the gear grinding machine spindle speed. Based on the coolant flow correction value, it adjusts the coolant pump station output flow rate. After machining is completed, it generates a second execution report containing the actual radial feed, actual grinding wheel linear speed, actual coolant flow rate, and spindle load peak value, and sends it back to the model iteration module.

[0029] The arbitration scheduling unit is connected to the CNC and precision machining buffer of each precision machining machine tool. It is used to receive buffer space request messages from each precision machining machine tool, adjudicate conflict requests according to preset arbitration rules, and allocate buffer spaces.

[0030] As a further improvement to this technical solution, the cache slot request message includes the following fields: request slot number, request machine tool equipment number, workpiece unique identifier, workpiece urgent order priority identifier, workpiece predicted finishing deformation amount, estimated remaining processing time for this workpiece, and machine tool current real-time load rate.

[0031] As a further improvement to this technical solution, the preset arbitration rules include: Rule 1: The workpiece with the highest emergency priority level will be given priority to obtain the target cache space.

[0032] Rule 2: Under the same priority conditions, the workpiece with the larger predicted value of the finishing deformation will be given priority in occupying the storage space.

[0033] Rule 3: When priority and predicted deformation are the same, the workpiece with the shorter estimated processing time shall be given priority in the allocation of storage space.

[0034] Rule 4: If the arbitration scheduling unit does not receive a response message from the target machine tool within 100 milliseconds after issuing the slot allocation instruction, the machine tool request is deemed invalid, and the slot will be reopened for competition.

[0035] As a further improvement to this technical solution, when the quality access control module routes the current workpiece to the roughing process and reallocates the subsequent cutting allowance, the reallocation is based on the measured geometric dimensions of the current workpiece, the predicted deformation amount, and the cutting capability parameters of the roughing process. A new allowance allocation scheme is generated through real-time calculation, rather than calling a fixed data table.

[0036] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0037] This invention uses a data acquisition module to collect multi-source physical signals online at the post-forging inspection station for each forging. Based on this, a deformation prediction module calls a prediction model to calculate the predicted deformation of each forging due to residual stress release during subsequent finishing. This breaks away from the traditional batch-based, extensive scheduling mode of MES / APS, allowing each forging to receive a personalized processing plan that matches its actual quality state. The collaborative decision-making module compares the predicted deformation with dual thresholds and executes a three-level hierarchical decision: normal parts are processed according to standard procedures; slightly off-center parts are issued a allowance setting value and / or holding time correction value to the forging press to compensate for deformation trends at the source; severely off-center parts trigger quality gate interception or downgrade processing and issue finishing compensation parameters, achieving a closed-loop linkage across processes from "post-process perception to pre-process compensation," reducing the finishing scrap rate at the source.

[0038] This invention upgrades the finishing buffer area from passive storage to an active temperature control actuator. When the predicted deformation exceeds a threshold, the active buffer control module issues environmental control commands to the buffer area actuator. By adjusting the opening and closing angle of the insulation cover, the fan speed, and the stacking density, it actively regulates the workpiece cooling rate, intervening in the deformation trajectory of the forging before it enters the finishing process, thus expanding the controllable deformation time window. Simultaneously, the arbitration scheduling unit efficiently adjudicates buffer space conflicts based on a four-level priority rule, ensuring the rational allocation of limited buffer resources and avoiding production line congestion or deadlock.

[0039] This invention receives execution reports from the process execution module via a model iteration module. It uses the deviation between the measured accuracy data of the finished product after finishing and the predicted deformation as feedback samples to continuously update the internal weights of the deformation prediction model. This closed-loop iteration mechanism enables the system's prediction accuracy to continuously improve with each production batch, effectively overcoming the model degradation problem caused by the traditional "one-time training, lifetime use" approach. It allows the system to adaptively evolve with changes in production line equipment status and material batch replacements, significantly improving the long-term operational stability of the production line and the consistency of the finished product. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0041] Figure 1 This is a schematic diagram of the system structure connection of the present invention. Detailed Implementation

[0042] 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.

[0043] Example: Please refer to Figure 1 As shown, an adaptive scheduling and collaborative control system for a gear multi-process forging and finishing production line is provided, including: a data acquisition module, set at the post-forging inspection station, used to acquire the measured physical signals of the current forging billet and perform local signal processing to generate a structured forging billet state feature dataset and output it.

[0044] In one specific embodiment, the forging billet state feature dataset includes at least the following fields: workpiece unique identifier, measured value of forging billet outer diameter, measured value of forging billet inner hole, runout value of tooth billet reference surface end face, characteristic value of ultrasonic echo shrinkage and porosity probability, characteristic value of infrared temperature gradient variance after forging cooling, characteristic value of temperature time-domain attenuation rate, peak pressure of final forging press, holding time of final forging, and historical statistical average of residual stress of the current batch.

[0045] The deformation prediction module, located in the edge industrial server, is connected to the data acquisition module. It is used to receive the forging billet state feature dataset, the batch information of the current workpiece, and the finished product accuracy level required by the order. It calls the deformation prediction model to calculate the predicted deformation of the current forging billet due to the release of residual stress during the subsequent finishing process.

[0046] The collaborative decision-making module, located in the edge industrial server, communicates with the deformation prediction module and is used to compare the predicted deformation amount with preset first threshold A and second threshold B, and generate corresponding process adjustment instruction packages based on the comparison results.

[0047] In one specific embodiment, the first threshold A and the second threshold B are dynamic thresholds, which are generated in real time by superimposing a preset safety offset with the statistical average of the residual stress-deformation of the historical forging billets in the current processing batch as the benchmark value; the A and B threshold values ​​of the current batch are automatically refreshed once after processing N workpieces, where N is an integer greater than or equal to 1.

[0048] For example: , ,in This represents the statistical average of residual stress-deformation for the current batch. The standard deviation of residual stress-deformation for this batch is given. , Preset safety factor (preferred) =0.8, =1.5).

[0049] In one specific embodiment, the specific logic of the collaborative decision-making module generating the corresponding process adjustment instruction package based on the comparison result is as follows: when the predicted deformation amount is ≤ A, an instruction to maintain the standard forging allowance parameter is generated, the current workpiece is marked as a normal part and transferred to the finishing buffer area.

[0050] When A < the predicted deformation amount ≤ B, an instruction package containing the adjusted target forging allowance setting value and / or forging holding time correction value is generated and sent to the forging press CNC to control the final forging pressure stop position of the subsequent workpieces to be forged in this batch to increase the single-sided forging allowance, and / or adjust the holding time of the forging press; the current workpiece is transferred to the finishing buffer area as a normal part.

[0051] When the predicted deformation is greater than B, the quality gate module is triggered to perform an interception judgment: if the interception is successful, a reflow command is generated to return the current workpiece to the re-inspection station; if the interception is unsuccessful, a high-risk label is marked for the current workpiece, and the instruction package includes recommended parameters for finishing feed compensation, grinding wheel linear speed correction value and coolant flow rate correction value, reducing the cutting feed rate of the current workpiece in finishing, lowering the grinding wheel linear speed and increasing the coolant flow rate.

[0052] In one specific embodiment, the process adjustment instruction package includes a target forging allowance setting value, a forging holding time correction value, a risk marker label, recommended parameters for finishing feed compensation, a grinding wheel linear speed correction value, and a coolant flow rate correction value.

[0053] The active buffer control module is communicatively connected to the collaborative decision-making module and the execution mechanism of the finishing buffer area. When the predicted deformation exceeds the preset buffer intervention threshold, it sends an environmental control command to the execution mechanism of the buffer area to adjust the cooling rate of the workpiece in the buffer area.

[0054] The cache intervention threshold is less than the first threshold A.

[0055] The process execution module is communicatively connected to the collaborative decision-making module and the finishing buffer area, and is set in the CNC of the forging press and / or the CNC of the finishing machine tool. It is used to receive and parse the process adjustment instruction package, execute the forging end adjustment action and / or finishing end compensation action in the process adjustment instruction package, and adjudicate and allocate buffer space conflict requests of the finishing machine tool according to the preset arbitration rules.

[0056] In one specific embodiment, the process execution module includes: a margin adjustment execution unit, which is set in the CNC of the forging press and is communicatively connected to the collaborative decision module. It is used to receive and parse the target forging margin setting value and the forging holding time correction value in the process adjustment instruction package, control the pressing stop position of the main oil cylinder according to the target forging margin setting value to adjust the single-sided forging margin of the subsequent workpieces to be forged in this batch, and / or control the holding time of the forging press according to the forging holding time correction value, and generate a first execution report containing the actual final forging pressing amount and the actual holding time after forging is completed, and send it back to the model iteration module.

[0057] The compensation execution unit, located in the CNC of the finishing machine tool, is communicatively connected to the collaborative decision-making module. It receives and parses the recommended finishing feed compensation parameters, grinding wheel linear speed correction value, and coolant flow correction value in the process adjustment instruction package. Based on the recommended finishing feed compensation parameters, it sends the feed rate override value in real time via industrial Ethernet to execute cutting. Based on the grinding wheel linear speed correction value, it adjusts the gear grinding machine spindle speed. Based on the coolant flow correction value, it adjusts the coolant pump station output flow rate. After machining is completed, it generates a second execution report containing the actual radial feed, actual grinding wheel linear speed, actual coolant flow rate, and spindle load peak value, and sends it back to the model iteration module.

[0058] The arbitration scheduling unit is connected to the CNC and precision machining buffer of each precision machining machine tool. It is used to receive buffer space request messages from each precision machining machine tool, adjudicate conflict requests according to preset arbitration rules, and allocate buffer spaces.

[0059] In one specific embodiment, the cache slot request message includes the following fields: request slot number, request machine tool equipment number, workpiece unique identifier, workpiece emergency order priority identifier, workpiece predicted finishing deformation amount, estimated remaining processing time for this workpiece, and machine tool current real-time load rate.

[0060] In one specific embodiment, the preset arbitration rules include: Rule 1: The workpiece with the highest emergency priority level will be given priority to obtain the target cache space.

[0061] Rule 2: Under the same priority conditions, the workpiece with the larger predicted value of the finishing deformation will be given priority in occupying the storage space.

[0062] Rule 3: When priority and predicted deformation are the same, the workpiece with the shorter estimated processing time shall be given priority in the allocation of storage space.

[0063] Rule 4: If the arbitration scheduling unit does not receive a response message from the target machine tool within 100 milliseconds after issuing the slot allocation instruction, the machine tool request is deemed invalid, and the slot will be reopened for competition.

[0064] The quality access control module is communicatively connected to the collaborative decision-making module and the process execution module. It is used to intercept the current workpiece from flowing downstream and generate a return command when the predicted deformation exceeds the preset rework judgment threshold, so as to return the current workpiece to the re-inspection station. It is also used to route the current workpiece to the roughing process and reallocate the subsequent cutting allowance when the predicted deformation exceeds the preset downgrade judgment threshold but the interception is not triggered.

[0065] In one specific embodiment, when the quality access control module routes the current workpiece to the roughing operation and reallocates the subsequent cutting allowance, the reallocation is based on the measured geometric dimensions of the current workpiece, the predicted deformation amount, and the cutting capability parameters of the roughing operation. A new allowance allocation scheme is generated through real-time calculation, rather than calling a fixed data table.

[0066] The model iteration module, located in the edge industrial server, is communicatively connected to the process execution module. It is used to receive the execution report containing actual processing parameters returned by the process execution module, and to use the deviation between the measured data of the finished product accuracy after finishing and the predicted deformation amount as feedback samples to iteratively update the internal weights of the deformation prediction model.

[0067] In another specific embodiment, the input shaft gear of a certain type of heavy-duty vehicle transmission (module 3.5, number of teeth 23, tooth width 42mm, material 20CrMnTiH) is used as the object, and it is implemented on a typical gear forging-finishing production line. The production line includes: a medium-frequency induction heating furnace (heating temperature 1150℃±20℃), a 1600-ton hot die forging press (including pre-forging and final forging stations), a post-forging constant temperature stacking rack (equipped with a heat insulation cover and variable frequency cooling fan), CNC lathes, gear hobbing machines, gear grinding machines, and corresponding AGV transfer systems.

[0068] The precision requirements for the finished gears are: tooth profile accuracy GB / T 10095-6 grade, tooth direction accuracy grade 6, and radial runout of the gear ring ≤0.063mm. The standard value for single-sided forging allowance during the finishing stage is 0.5mm. The measured residual stress on the surface after gear forging is approximately 285~326MPa. The finishing buffer area has a total of 12 slots.

[0069] II. The configuration of the comparison scheme is: traditional MES / APS scheduling scheme

[0070] (a) Control methods

[0071] A general-purpose MES system is used for batch-level production scheduling. The forging and finishing sections have independent production plans: the forging section operates on a "first-in, first-out" principle, with 100 pieces per batch forged and then transferred to a buffer area; the finishing section processes materials sequentially from the buffer area according to the work orders issued by the MES system. Finishing cutting parameters are uniformly set for each batch (turning feed 0.25mm / r, hobbing feed 2.0mm / r, grinding feed 0.015mm / stroke), without piece-by-piece adjustments. The buffer area serves only as a temporary storage location, and the cooling process is not actively intervened in.

[0072] (II) Experimental Results

[0073] In the continuous production of a batch of 100 gears:

[0074] Deformation prediction and compensation capability: The system cannot predict the residual stress release deformation of each forging blank individually. After slight wear of the die in the forging section, the final forging temperature of forging blanks 31 to 70 fluctuated significantly (1120–1180℃), resulting in a significant increase in the dispersion of residual stress in the blanks. Due to the lack of individual sensing and prediction methods, the finishing section cannot identify high-risk workpieces in advance.

[0075] Actual deformation: The distribution of radial runout test results for the gear ring after finishing is as follows:

[0076] Deformation range Number of workpieces percentage ≤0.063mm (Pass) 62 items 62% 0.064~0.080mm 22 items 22% 0.081~0.120mm 16 items 16%

[0077] The maximum radial runout reached 0.118mm, far exceeding the acceptable level of 0.063mm.

[0078] Scrap rate and efficiency loss: Of the 16 out-of-tolerance workpieces, 12 were scrapped because the deformation was too large to be salvaged by adjusting the cutting parameters; 4 needed to be returned to the forging section to increase the allowance and then processed again. The batch scrap rate was 16%.

[0079] Production line coordination efficiency: When the finishing process is stopped for changeover (the gear hobbing machine takes about 45 minutes to change the tool), the forging section cannot know the downstream status and continues to produce at the original pace, resulting in 28 forging billets accumulating in the buffer area, congestion in the AGV transfer system, and an overall production line efficiency (OEE) of only 67.3%.

[0080] System Implementation Scheme of the Invention

[0081] (I) System Deployment

[0082] Deploying the system of this invention at the post-forging inspection station includes:

[0083] Data acquisition modules: 3D laser scanner (detects outer diameter, inner hole, and end face runout), infrared thermal imager (acquires cooling temperature gradient sequence from 0 to 120 seconds after forging and calculates temperature time-domain attenuation rate), ultrasonic flaw detector (detects internal shrinkage cavities and porosity), and press data interface (acquires peak pressure and holding time during final forging).

[0084] Deformation prediction module: Deployed on an edge industrial server, it adopts a Gaussian process regression (GPR) model, taking 10 dimensions of features as input, including the measured outer diameter of the forging billet, the measured inner diameter, the end face runout, the ultrasonic echo probability characteristic value, the temperature gradient variance, the temperature time-domain attenuation rate, the peak downward pressure, the holding time, the statistical mean of batch residual stress, and the order accuracy level. The output is the predicted value of tooth profile deviation and the predicted value of tooth direction deviation.

[0085] Collaborative decision-making module: Preset dynamic thresholds A and B. Based on the statistical mean of residual stress-deformation in the current batch, a safety offset is added to generate a value in real time, and the module is refreshed every 5 pieces completed.

[0086] Active buffer control module: The buffer intervention threshold is set to 0.7A. When the predicted deformation exceeds this threshold but does not exceed A, a control command is sent to the buffer insulation cover and the variable frequency fan.

[0087] Process execution module: The allowance adjustment execution unit is set in the CNC of the forging press, the compensation execution unit is set in the CNC of the finishing machine tool, and the arbitration scheduling unit is deployed in the buffer industrial control computer.

[0088] Quality access control module and model iteration module: deployed on edge industrial servers.

[0089] (II) System Operation Process

[0090] Taking the actual operation of 100 gears from the same batch as an example:

[0091] 1. Pieces 1-30 (Stability stage of forging)

[0092] The data acquisition module collects physical signals from each forging billet. The deformation prediction module calculates and predicts the deformation distribution to be between 0.032 and 0.048 mm, all below the dynamic threshold A (currently A = 0.052 mm). The collaborative decision-making module classifies it as a normal part and proceeds according to standard process flow without intervention. After finishing, the measured radial runout of the gear ring is between 0.035 and 0.052 mm, all of which are qualified.

[0093] 2. Item No. 31 (Forging Fluctuation Trigger)

[0094] During the final forging of the 31st forging billet, the final forging temperature was too high (1175℃) due to temperature fluctuations in the die. The infrared thermal imaging unit detected an abnormal temperature time-domain attenuation rate for this piece. The batch average value is 0.019s.-1 The predicted deformation output by the deformation prediction module is 0.061 mm.

[0095] The collaborative decision-making module compares 0.061mm with the current dynamic threshold (A=0.052mm, B=0.078mm): A<0.061≤B, and determines it as "lightly off-center part".

[0096] Action Execution: Upon receiving a command packet containing "target forging allowance setting value 0.65mm" (standard value 0.5mm increased by 0.15mm), the allowance adjustment unit controls the main hydraulic cylinder to move its downward stop position backward during the forging cycle of the 32nd and subsequent workpieces to be forged, increasing the single-sided forging allowance from 0.5mm to 0.65mm. Simultaneously, the active buffer control module issues environmental control commands to existing workpieces in the buffer area, reducing the cooling fan speed from 1200rpm to 800rpm, decreasing the cooling rate to slow down the residual stress release process.

[0097] Results: The predicted deformation of pieces 32 to 70 fell back to the range of 0.042 to 0.055 mm.

[0098] 3. Item 71 (Severe Deviation Triggered)

[0099] Due to abnormal temperature uniformity in the heating furnace, the final forging temperature of the 71st forging billet was only 1130℃, lower than the lower limit of the process. The data acquisition module detected an abnormal increase in the peak pressure during final forging (from the baseline value of 1850kN to 2120kN) and a significant increase in the variance of the infrared temperature gradient (from the baseline value of 42℃² to 89℃²). The deformation prediction module output a predicted deformation of 0.095mm, exceeding B=0.078mm.

[0100] The collaborative decision-making module triggers the quality access control: Based on the assessment that the part has not yet reached an irreparable state (the measured geometric dimensions are still within the machinable range), no interception is performed; it is marked as a "high-risk tag," and three compensation parameters are sent to the compensation execution unit:

[0101] Finishing feed rate Override value: 0.75 (standard value 1.0, reduced by 25%)

[0102] Grinding wheel linear velocity correction value: 22m / s (standard value 28m / s, reduced by 21%)

[0103] Coolant flow rate correction: 18L / min (standard value 12L / min increased by 50%)

[0104] At the same time, the arbitration scheduling unit received the buffer bay request message from the finishing machine tool. According to the four-level arbitration rules (urgent orders take priority → those with larger predicted deformation amount take priority → those with shorter processing time take priority → re-competition after 100ms timeout), the workpiece was preferentially allocated to bay number 3, which is closest to the gear grinding machine.

[0105] Result: The measured radial runout of the gear ring after finishing was 0.071mm, which was slightly higher than the acceptable line of 0.063mm. However, it successfully avoided direct scrapping (under the traditional method, the actual radial runout corresponding to the predicted deformation of 0.095mm was about 0.105mm, which would inevitably lead to scrapping). After being transferred to the roughing process, the allowance was redistributed, and the radial runout of 0.068mm was finally downgraded to a defective product.

[0106] 4. Model Iterative Update

[0107] After every 5 finished products are completed, the model iteration module uses the deviation between the measured accuracy data of the finished product and the predicted deformation as feedback samples to iteratively update the deformation prediction model weights. By the time the 100th product is completed, the average prediction deviation of the model in the 31st to 70th product range has decreased from the initial 0.009mm to 0.004mm.

[0108] IV. Comparison Results of the Invention with Traditional Solutions

[0109] Comparison indicators Traditional MES / APS solutions This invention system Improvement range Average radial runout of the gear ring after finishing 0.071mm 0.049mm ↓31.0% Number of workpieces with radial runout exceeding tolerance (>0.063mm) 38 pieces (16 of which are >0.080mm) 6 pieces (all within the range of 0.064–0.071 mm) ↓84.2% Batch scrap rate 16% 0% (6 items were downgraded for use, none were scrapped) ↓100% Number of reworked / scrapped parts due to deformation exceeding tolerance 16 scrapped pieces + 4 reworked pieces 6 items were downgraded for use, and 0 items were scrapped. ↓70% Peak cache backlog 28 items 9 items ↓67.9% Overall Equipment Effectiveness (OEE) 67.3% 84.7% ↑25.9% Deformation prediction model prediction bias (for the 100th piece) Not applicable (no predictive model) ≤0.004mm —

[0110] This embodiment fully verifies the core innovative value of the system of the present invention:

[0111] (i) Traditional solutions assume that the quality of forging billets in the same batch is consistent, and cannot identify the residual stress anomaly caused by the final forging temperature fluctuation in the 31st piece. This invention collects 10-dimensional physical signals for each piece through a data acquisition module and calculates and predicts the deformation amount for each piece through a deformation prediction module. Risk identification and intervention are completed before the 31st piece enters the finishing process, reducing the average radial runout of the subsequent 40 pieces from 0.071 mm in the traditional solution to 0.049 mm, a reduction of 31.0%.

[0112] (ii) In the slight deviation range of parts 31 to 70 (A < predicted deformation ≤ B), the system uses the allowance adjustment unit to increase the single-sided allowance of subsequent forgings from 0.5 mm to 0.65 mm, compensating for the deformation trend from the source. In the severe deviation of part 71 (predicted deformation > B), the system uses three compensation parameters of the compensation execution unit (feed rate reduced by 25%, grinding wheel linear speed reduced by 21%, and coolant flow rate increased by 50%) to reduce the actual radial runout from the predicted 0.095 mm to 0.071 mm. The three-level grading strategy avoids the problem of a 16% scrap rate caused by the traditional "one-size-fits-all" uniform process.

[0113] (iii) The model iteration module continuously updates the prediction model weights at a frequency of 5 pieces, bringing the prediction error down from 0.009mm to 0.004mm. This capability is crucial in long-term mass production—as molds wear out and material batches change, the scrap rate of traditional solutions gradually increases, while the prediction accuracy and decision quality of the system of this invention will be continuously optimized.

[0114] (iv) The active cache control module reduced the backlog in the cache area from 28 items to 9 items by adjusting the cooling rate; the four-level arbitration rules of the arbitration scheduling unit ensured the efficient allocation of cache slots; the overall production line efficiency increased from 67.3% to 84.7%, an improvement of 25.9%.

[0115] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention.

Claims

1. An adaptive scheduling and collaborative control system for a multi-process gear forging and finishing production line, characterized in that, include: The data acquisition module is used to acquire the measured physical signals of the current forging billet and perform local signal processing to generate and output a structured forging billet state feature dataset. The deformation prediction module is used to receive the forging billet state feature dataset, the batch information of the current workpiece and the finished product accuracy level required by the order, and call the deformation prediction model to calculate the predicted deformation of the current forging billet due to the release of residual stress during the subsequent finishing process. The collaborative decision-making module is used to compare the predicted deformation with preset first threshold A and second threshold B, and generate a corresponding process adjustment instruction package based on the comparison result. An active buffer control module is used to send an environmental control command to the actuator in the buffer zone when the predicted deformation exceeds a preset buffer intervention threshold, thereby adjusting the cooling rate of the workpiece in the buffer zone. The process execution module is used to receive and parse the process adjustment instruction package, execute the forging end adjustment action and / or finishing end compensation action in the process adjustment instruction package, and adjudicate and allocate the buffer slot conflict request of the finishing machine tool according to the preset arbitration rules. The quality access control module is used to intercept the current workpiece from flowing downstream and generate a return command when the predicted deformation exceeds the preset rework judgment threshold, so as to return the current workpiece to the re-inspection station. And when the predicted deformation exceeds a preset degradation judgment threshold but no interception is triggered, the current workpiece is routed to the roughing operation and the subsequent cutting allowance is reallocated. The model iteration module is used to receive the execution report containing actual processing parameters returned by the process execution module, and use the deviation between the measured data of the finished product accuracy after finishing and the predicted deformation as feedback samples to iteratively update the internal weights of the deformation prediction model.

2. The adaptive scheduling and collaborative control system for a multi-process gear forging and finishing production line according to claim 1, characterized in that: The forging billet state feature dataset contains at least the following fields: The unique identifier of the workpiece, the measured value of the outer diameter of the forging billet, the measured value of the inner hole of the forging billet, the runout value of the end face of the tooth blank reference surface, the characteristic value of the probability of shrinkage and porosity of ultrasonic echo, the characteristic value of the variance of the infrared temperature gradient after cooling after forging, the characteristic value of the temperature time-domain attenuation rate, the peak value of the pressure under the final forging press, the holding time of the final forging, and the historical statistical average value of the residual stress of the current batch.

3. The adaptive scheduling and collaborative control system for a multi-process gear forging and finishing production line according to claim 1, characterized in that: The first threshold A and the second threshold B are dynamic thresholds, which are generated in real time by superimposing a preset safety offset with the statistical average of the residual stress-deformation of the historical forging billets in the current processing batch as the benchmark value; the A and B threshold values ​​of the current batch are automatically refreshed once after processing N workpieces, where N is an integer greater than or equal to 1.

4. The adaptive scheduling and collaborative control system for a multi-process gear forging and finishing production line according to claim 1, characterized in that: The specific logic of the collaborative decision-making module in generating the corresponding process adjustment instruction package based on the comparison results is as follows: When the predicted deformation is ≤ A, an instruction is generated to maintain the standard forging allowance parameters, the current workpiece is marked as a normal part and transferred to the finishing buffer area; When A < the predicted deformation amount ≤ B, an instruction package containing the adjusted target forging allowance setting value and / or forging holding time correction value is generated and sent to the forging press CNC to control the final forging pressure stop position of the subsequent workpieces to be forged in this batch to increase the single-sided forging allowance and / or adjust the holding time of the forging press. The current workpiece is transferred to the finishing buffer area as a normal part; When the predicted deformation is greater than B, the quality gate module is triggered to perform an interception judgment: if the interception is successful, a reflow command is generated to return the current workpiece to the re-inspection station; if the interception is unsuccessful, a high-risk label is marked for the current workpiece, and the instruction package includes recommended parameters for finishing feed compensation, grinding wheel linear speed correction value and coolant flow rate correction value, reducing the cutting feed rate of the current workpiece in finishing, lowering the grinding wheel linear speed and increasing the coolant flow rate.

5. The adaptive scheduling and collaborative control system for a multi-process gear forging and finishing production line according to claim 1, characterized in that: The process adjustment instruction package includes the target forging allowance setting value, forging holding time correction value, risk label, recommended parameters for finishing feed compensation, grinding wheel linear speed correction value, and coolant flow rate correction value.

6. The adaptive scheduling and collaborative control system for a multi-process gear forging and finishing production line according to claim 1, characterized in that: The process execution module includes: The allowance adjustment execution unit is used to receive and parse the target forging allowance setting value and the forging holding time correction value in the process adjustment instruction package, control the pressing stop position of the main oil cylinder according to the target forging allowance setting value to adjust the single-sided forging allowance of the subsequent workpieces to be forged in this batch, and / or control the holding time of the forging press according to the forging holding time correction value, and generate a first execution report containing the actual final forging pressing amount and the actual holding time after forging is completed, and send it back to the model iteration module; The compensation execution unit is used to receive and parse the recommended parameters for finishing feed compensation, the grinding wheel linear speed correction value, and the coolant flow rate correction value in the process adjustment instruction package. Based on the recommended parameters for finishing feed compensation, it sends the feed rate value in real time via industrial Ethernet to perform cutting. Based on the grinding wheel linear speed correction value, it adjusts the spindle speed of the gear grinding machine. Based on the coolant flow rate correction value, it adjusts the output flow rate of the coolant pump station. After the machining is completed, it generates a second execution report containing the actual radial feed, actual grinding wheel linear speed, actual coolant flow rate, and peak spindle load, and sends it back to the model iteration module. The arbitration scheduling unit is used to receive cache space request messages from each precision machining machine tool, adjudicate conflict requests according to preset arbitration rules, and allocate cache spaces.

7. The adaptive scheduling and collaborative control system for a multi-process gear forging and finishing production line according to claim 1, characterized in that: The cache slot request message includes the following fields: requested slot number, requested machine tool equipment number, workpiece unique identifier, workpiece urgent order priority identifier, workpiece predicted finishing deformation amount, estimated remaining processing time for this workpiece, and machine tool current real-time load rate.

8. The adaptive scheduling and collaborative control system for a multi-process gear forging and finishing production line according to claim 1, characterized in that: The preset arbitration rules include: Rule 1: The workpiece with the highest emergency priority level will be given priority in obtaining the target cache slot; Rule 2: Under the same priority conditions, the workpiece with the larger predicted value of finishing deformation will be given priority in occupying the storage space; Rule 3: When priority and predicted deformation are the same, the workpiece with the shorter estimated processing time shall be given priority in the allocation of storage space; Rule 4: If the arbitration scheduling unit does not receive a response message from the target machine tool within 100 milliseconds after issuing the slot allocation instruction, the machine tool request is deemed invalid, and the slot will be reopened for competition.

9. The adaptive scheduling and collaborative control system for a multi-process gear forging and finishing production line according to claim 1, characterized in that: When the quality access control module routes the current workpiece to the roughing process and reallocates the subsequent cutting allowance, the reallocation is based on the measured geometric dimensions of the current workpiece, the predicted deformation amount, and the cutting capability parameters of the roughing process. A new allowance allocation scheme is generated through real-time calculation, rather than calling a fixed data table.