A method of finishing a delicate part

Abstract

The present application relates to the technical field of variable part processing, and particularly relates to a finish machining method for variable parts, which combines airtight detection with real-time monitoring of clamping state, guarantees the reliability and consistency of workpiece positioning, and avoids machining reference error caused by clamping deviation or loosening; in the OP20 process, the expanded clamping mode is adopted to clamp the finished inner hole and position with the end face, ensuring the machining precision of the outer circle and the other end face; by collecting full-process machining process data, a machining quality prediction model based on key characteristic parameters is established, realizing the pre-sensing of machining state and the early prediction of abnormal risk, overcoming the limitations of relying on post-detection and passive remediation in the prior art; meanwhile, when the model prediction result exceeds the preset precision range, the automatically triggered process parameter rollback mechanism can quickly restore the current parameters to the last qualified machining state, avoiding batch waste or equipment damage caused by continuous parameter deviation.

Description

Technical Field

[0001] This invention relates to the field of machining technology for volatile parts, and more particularly to a finishing method for volatile parts. Background Technology

[0002] Deformable parts (such as thin-walled parts and irregularly shaped parts) are widely used in aerospace, precision machinery, and other fields. Their complex structures and poor rigidity make them prone to deformation during machining due to clamping forces, cutting forces, and residual stresses, severely affecting machining accuracy and product quality. To control machining deformation, various clamping and positioning methods have been developed, such as using flexible fixtures, pneumatic expansion mandrels, and floating clamping systems to reduce clamping stress. Simultaneously, some solutions use airtightness testing devices to verify the reliability of clamping and positioning. Furthermore, some methods attempt to eliminate internal stress through cryogenic treatment to reduce deformation during machining.

[0003] In existing volatile parts processing technologies, the quality of parts is usually determined by inspection after processing. It is difficult to predict and proactively intervene in quality abnormalities during processing. Furthermore, when deviations occur in processing quality, there is a lack of effective adaptive adjustment and rollback mechanisms for process parameters, making it difficult to restore the processing to normal state in a timely manner. Summary of the Invention

[0004] The purpose of this invention is to provide a finishing method for volatile parts, which solves the technical problems in the prior art that it is difficult to predict and actively intervene in quality abnormalities during the processing, and that when the processing quality deviates, there is a lack of effective adaptive adjustment and rollback mechanism for process parameters, making it difficult to restore the normal processing state in a timely manner.

[0005] To achieve the above objectives, the present invention provides a finishing method for easily deformable parts, comprising:

[0006] OP10 process: Place the workpiece on the main support, auxiliary support and coarse positioning block, and use the gripper cylinder to drive the jaws to center and clamp the workpiece. Use the high-energy cylinder to rotate the jaws and press the end face of the workpiece. Use an airtightness detection device to check whether the clamping and positioning is reliable.

[0007] After clamping and positioning, the jaws are released, and the stepped surface and inner hole of the workpiece are finished.

[0008] OP20 process: The precision-machined inner hole is clamped using an expansion clamping method, and the workpiece is positioned by the end face. The outer circle and the other end face of the workpiece are then machined.

[0009] Collect and store the entire process data of OP10 and OP20 processes. Based on the stored process data, extract the key feature parameters that affect the machining accuracy of the workpiece, establish a machining quality prediction model, determine the influence weight of each process parameter on the machining quality through model analysis, and generate process parameter optimization strategies based on the influence weights to form an optimized machining scheme.

[0010] In subsequent processing, the real-time collected processing data will be input into the processing quality prediction model. When the processing result predicted by the model exceeds the preset accuracy range, it is determined that there is a risk of processing abnormality, and the process parameter rollback mechanism will be automatically triggered to roll back the current process parameters to the process parameter version that was qualified in the most recent processing before the abnormality occurred.

[0011] In the OP10 process, the airtightness detection device introduces compressed gas into the contact area between the workpiece and the positioning surface and monitors the gas pressure changes in real time. When the gas pressure is lower than the preset threshold, it determines that the clamping and positioning is unreliable and issues an alarm signal to prevent the processing equipment from starting. When the gas pressure is within the preset threshold range, it determines that the clamping and positioning is reliable and allows the processing step to proceed.

[0012] Among them, the processing quality prediction model adopts an incremental learning approach. As the number of processing batches increases, newly generated processing data is continuously added to the training dataset to iteratively update the model.

[0013] After each batch of processing is completed, the influence weight of each process parameter is recalculated based on the updated model, and the optimized processing plan corresponding to the batch is generated, forming a process parameter iteration record versioned by processing batch. Each version records the set of process parameters and processing quality results corresponding to the iteration.

[0014] The execution method of the process parameter rollback mechanism is as follows:

[0015] When the processing quality prediction model predicts processing results that exceed the preset accuracy range, the reliability of the current process parameter version is first determined.

[0016] If the most recent actual processing quality corresponding to the current version of the process parameters is qualified, then fine-tuning is performed based on the current version of the process parameters;

[0017] If the most recent actual processing quality corresponding to the current version of the process parameters is unqualified, it will automatically revert to the previous version of the process parameters;

[0018] If the previous version is still unqualified, the process will be rolled back step by step until the most recent version that passed the quality test is returned. Each rollback operation will record the reason for the rollback and the processing result after the rollback, which will be used as label data for subsequent model training.

[0019] Among them, after each incremental learning, the processing quality prediction model evaluates the confidence level of the model output and calculates the uncertainty range of the model prediction results;

[0020] When the confidence level of the model prediction result is lower than the preset confidence threshold, the current prediction result is determined to be unreliable, the automatic rollback mechanism of process parameters is not triggered, and a manual review prompt is output instead.

[0021] When the confidence level of prediction results in multiple consecutive batches is lower than the threshold, the model retraining process is automatically triggered, and the model is remodeled using all accumulated data.

[0022] The data from the processing process is preprocessed after collection before being stored and used as input to the model. The preprocessed data is stored in a structured manner according to processing batches, and batch indexes and time-series indexes are established to facilitate subsequent data retrieval and model training.

[0023] The data preprocessing steps include:

[0024] Outlier detection is performed on the collected raw data to remove invalid data points that exceed physical limits; missing data is filled by interpolation; time alignment and synchronization are performed on multi-source data from different sensors; and feature parameters with different dimensions are normalized to bring them to the same order of magnitude.

[0025] Among them, a tool wear detection device is installed on the processing equipment to monitor the wear status of the tool in real time;

[0026] When the tool wear is detected to have reached the preset wear threshold, the tool compensation value is automatically applied to correct the current machining path.

[0027] When the tool wear exceeds the threshold, a tool replacement prompt signal is issued.

[0028] Among them, a cooling and lubricating medium is continuously supplied to the cutting contact area between the tool and the workpiece through a coolant supply system;

[0029] The supply flow rate, supply pressure, and supply temperature of the cooling and lubricating medium are adjusted in real time according to the machining stage and cutting parameters; the cooling and lubricating medium is cutting fluid, compressed air, oil mist, or a combination thereof.

[0030] The process involves spraying compressed gas into the processing area using an air blowing device to remove chips and impurities accumulated on the workpiece surface, fixture positioning surface, and processing area. The air blowing device performs an air blowing cleaning operation once before each clamping and positioning and once after each processing operation. The parameters for air blowing cleaning include air blowing pressure, air blowing duration, and air blowing angle, which are adjusted according to different processing stages and workpiece structures.

[0031] This invention provides a finishing method for volatile parts. By combining airtightness testing with real-time monitoring of clamping status, it effectively ensures the reliability and consistency of workpiece positioning in the OP10 process, avoiding machining datum errors caused by clamping offset or loosening. In the OP20 process, the finished inner hole is clamped by expansion and positioned by the end face, ensuring the machining accuracy of the outer circle and the other end face. This solves the problem that volatile parts are difficult to guarantee dimensional and geometric tolerances due to insufficient rigidity or clamping deformation. By collecting data from the entire processing process, a processing quality prediction model based on key feature parameters is established, enabling pre-processing status perception and early prediction of abnormal risks. This overcomes the limitations of existing technologies that rely on post-processing detection and passive remediation. Simultaneously, when the model's prediction results exceed the preset accuracy range, an automatically triggered process parameter rollback mechanism can quickly restore the current parameters to the most recent qualified processing state, avoiding batch scrap or equipment damage caused by continuous parameter deviation. This significantly improves the stability and adaptive control capability of the processing process, effectively reducing scrap rate and rework costs. It is suitable for mass precision machining scenarios of thin-walled and easily deformable parts. Attached Figure Description

[0032] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.

[0033] Figure 1 This is a flowchart of the finishing method for the variable parts of the present invention. Detailed Implementation

[0034] The embodiments of the present invention are described in detail below. Examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, but should not be construed as limiting the present invention.

[0035] Please refer to Figure 1 , Figure 1 This is a flowchart of the finishing method for easily deformable parts according to the present invention. An embodiment of the present invention provides a finishing method for easily deformable parts, comprising:

[0036] S1, OP10 process: Place the workpiece on the main support, auxiliary support and coarse positioning block, and use the gripper cylinder to drive the chuck to center and clamp the workpiece. Use the high-energy cylinder to rotate the chuck and press the end face of the workpiece. Use the airtightness detection device to check whether the clamping and positioning is reliable.

[0037] In this specific embodiment, during the OP10 process, the airtightness detection device introduces compressed gas into the contact area between the workpiece and the positioning surface and monitors the gas pressure changes in real time. When the gas pressure is lower than a preset threshold, it determines that the clamping and positioning is unreliable and issues an alarm signal to prevent the processing equipment from starting. When the gas pressure is within the preset threshold range, it determines that the clamping and positioning is reliable and allows the processing step to proceed.

[0038] S2. After clamping and positioning, the jaws are released, and the stepped surface and inner hole of the workpiece are finished.

[0039] S3, OP20 process: The precision-machined inner hole is clamped by expansion clamping, and the workpiece is positioned by end face, and the outer circle and the other end face of the workpiece are machined;

[0040] S4. Collect and store the entire process data of OP10 and OP20 processes. Based on the stored process data, extract the key feature parameters that affect the machining accuracy of the workpiece, establish a machining quality prediction model, determine the influence weight of each process parameter on the machining quality through model analysis, and generate a process parameter optimization strategy based on the influence weight to form an optimized machining scheme.

[0041] S5. In subsequent processing, the real-time collected processing data will be input into the processing quality prediction model. When the processing result predicted by the model exceeds the preset accuracy range, it is determined that there is a risk of processing abnormality. The process parameter rollback mechanism will be automatically triggered to roll back the current process parameters to the process parameter version that was qualified in the most recent processing before the abnormality occurred.

[0042] In this specific implementation, the processing quality prediction model adopts an incremental learning approach. As the number of processing batches increases, newly generated processing data is continuously added to the training dataset to iteratively update the model.

[0043] After each batch of processing is completed, the influence weight of each process parameter is recalculated based on the updated model, and the optimized processing plan corresponding to the batch is generated, forming a process parameter iteration record versioned by processing batch. Each version records the set of process parameters and processing quality results corresponding to the iteration.

[0044] The process parameter rollback mechanism is executed as follows:

[0045] When the processing quality prediction model predicts processing results that exceed the preset accuracy range, the reliability of the current process parameter version is first determined.

[0046] If the most recent actual processing quality corresponding to the current version of the process parameters is qualified, then fine-tuning is performed based on the current version of the process parameters;

[0047] If the most recent actual processing quality corresponding to the current version of the process parameters is unqualified, it will automatically revert to the previous version of the process parameters;

[0048] If the previous version is still unqualified, the process will be rolled back step by step until the most recent version that passed the quality test is returned. Each rollback operation will record the reason for the rollback and the processing result after the rollback, which will be used as label data for subsequent model training.

[0049] Among them, after each incremental learning, the processing quality prediction model evaluates the confidence level of the model output and calculates the uncertainty range of the model prediction results;

[0050] When the confidence level of the model prediction result is lower than the preset confidence threshold, the current prediction result is determined to be unreliable, the automatic rollback mechanism of process parameters is not triggered, and a manual review prompt is output instead.

[0051] When the confidence level of prediction results in multiple consecutive batches is lower than the threshold, the model retraining process is automatically triggered, and the model is remodeled using all accumulated data.

[0052] The data from the processing process is preprocessed after collection before being stored and used as input to the model. The preprocessed data is stored in a structured manner according to processing batches, and batch indexes and time-series indexes are established to facilitate subsequent data retrieval and model training.

[0053] The data preprocessing steps include:

[0054] Outlier detection is performed on the collected raw data to remove invalid data points that exceed physical limits; missing data is filled by interpolation; time alignment and synchronization are performed on multi-source data from different sensors; and feature parameters with different dimensions are normalized to bring them to the same order of magnitude.

[0055] Furthermore, a tool wear detection device is installed on the processing equipment to monitor the wear status of the tool in real time;

[0056] When the tool wear is detected to have reached the preset wear threshold, the tool compensation value is automatically applied to correct the current machining path.

[0057] When the tool wear exceeds the threshold, a tool replacement prompt signal is issued.

[0058] Among them, a cooling and lubricating medium is continuously supplied to the cutting contact area between the tool and the workpiece through a coolant supply system;

[0059] The supply flow rate, supply pressure, and supply temperature of the cooling and lubricating medium are adjusted in real time according to the machining stage and cutting parameters; the cooling and lubricating medium is cutting fluid, compressed air, oil mist, or a combination thereof.

[0060] The process involves spraying compressed gas into the processing area using an air blowing device to remove chips and impurities accumulated on the workpiece surface, fixture positioning surface, and processing area. The air blowing device performs an air blowing cleaning operation once before each clamping and positioning and once after each processing operation. The parameters for air blowing cleaning include air blowing pressure, air blowing duration, and air blowing angle, which are adjusted according to different processing stages and workpiece structures.

[0061] Furthermore, while recording rollback operation data, it is pushed to the management terminal. The management terminal implements hierarchical management of administrator permissions based on a role-based access control algorithm. The role-based access control algorithm assigns different roles to different administrators, and each role is associated with a corresponding set of operation permissions. After receiving the rollback operation record, the management terminal determines whether the currently logged-in administrator has the right to view the record and perform subsequent processing operations based on the administrator's role permissions. The permission levels include at least: read-only permission, parameter adjustment permission, and system configuration permission. At the same time, the management terminal logs every modification operation of process parameters by the administrator, forming a complete operation traceability chain.

[0062] By combining airtightness testing with real-time monitoring of clamping status, the reliability and consistency of workpiece positioning in the OP10 process are effectively guaranteed, avoiding machining datum errors caused by clamping offset or loosening. In the OP20 process, the precision-machined inner hole is clamped using an expansion clamping method and positioned at the end face, ensuring the machining accuracy of the outer circle and the other end face. This solves the problem of difficulty in guaranteeing dimensional and geometric tolerances for easily deformable parts due to insufficient rigidity or clamping deformation. By collecting data from the entire machining process, a machining quality prediction model based on key feature parameters is established, enabling pre-processing status perception and early prediction of abnormal risks. This overcomes the limitations of existing technologies that rely on post-processing detection and passive remediation. At the same time, when the model prediction results exceed the preset accuracy range, the automatically triggered process parameter rollback mechanism can quickly restore the current parameters to the most recent qualified machining state, avoiding batch scrap or equipment damage caused by continuous parameter deviation. This significantly improves the stability and adaptive control capability of the machining process, effectively reducing scrap rate and rework costs. It is suitable for mass precision machining scenarios of thin-walled, easily deformable parts.

[0063] The above-disclosed embodiments are merely one or more preferred embodiments of this application and should not be construed as limiting the scope of this application. Those skilled in the art can understand that all or part of the processes for implementing the above embodiments and equivalent changes made in accordance with the claims of this application still fall within the scope of this application.

Claims

1. A method for finishing easily deformable parts, characterized in that, include: OP10 process: Place the workpiece on the main support, auxiliary support and coarse positioning block, and use the gripper cylinder to drive the jaws to center and clamp the workpiece. Use the high-energy cylinder to rotate the jaws and press the end face of the workpiece. Use an airtightness detection device to check whether the clamping and positioning is reliable. After clamping and positioning, the jaws are released, and the stepped surface and inner hole of the workpiece are finished. OP20 process: The precision-machined inner hole is clamped using an expansion clamping method, and the workpiece is positioned by the end face. The outer circle and the other end face of the workpiece are then machined. Collect and store the entire process data of OP10 and OP20 processes. Based on the stored process data, extract the key feature parameters that affect the machining accuracy of the workpiece, establish a machining quality prediction model, determine the influence weight of each process parameter on the machining quality through model analysis, and generate process parameter optimization strategies based on the influence weights to form an optimized machining scheme. In subsequent processing, the real-time collected processing data will be input into the processing quality prediction model. When the processing result predicted by the model exceeds the preset accuracy range, it is determined that there is a risk of processing abnormality, and the process parameter rollback mechanism will be automatically triggered to roll back the current process parameters to the process parameter version that was qualified in the most recent processing before the abnormality occurred.

2. The finishing method for easily deformable parts as described in claim 1, characterized in that, In the OP10 process, the airtightness detection device introduces compressed gas into the contact area between the workpiece and the positioning surface and monitors the gas pressure changes in real time. When the gas pressure is lower than the preset threshold, it determines that the clamping and positioning is unreliable and issues an alarm signal to prevent the processing equipment from starting. When the gas pressure is within the preset threshold range, it determines that the clamping and positioning is reliable and allows the processing step to proceed.

3. The finishing method for easily deformable parts as described in claim 2, characterized in that, The processing quality prediction model adopts an incremental learning approach. As the number of processing batches increases, newly generated processing data is continuously added to the training dataset to iteratively update the model. After each batch of processing is completed, the influence weight of each process parameter is recalculated based on the updated model, and the optimized processing plan corresponding to the batch is generated, forming a process parameter iteration record versioned by processing batch. Each version records the set of process parameters and processing quality results corresponding to the iteration.

4. The finishing method for easily deformable parts as described in claim 3, characterized in that, The process parameter rollback mechanism is executed as follows: When the processing quality prediction model predicts processing results that exceed the preset accuracy range, the reliability of the current process parameter version is first determined. If the most recent actual processing quality corresponding to the current version of the process parameters is qualified, then fine-tuning is performed based on the current version of the process parameters; If the most recent actual processing quality corresponding to the current version of the process parameters is unqualified, it will automatically revert to the previous version of the process parameters; If the previous version is still unqualified, the process will be rolled back step by step until the most recent version that passed the quality test is returned. Each rollback operation will record the reason for the rollback and the processing result after the rollback, which will be used as label data for subsequent model training.

5. The finishing method for easily deformable parts as described in claim 4, characterized in that, After each incremental learning iteration, the processing quality prediction model performs a confidence assessment on the model output and calculates the uncertainty interval of the model prediction results. When the confidence level of the model prediction result is lower than the preset confidence threshold, the current prediction result is determined to be unreliable, the automatic rollback mechanism of process parameters is not triggered, and a manual review prompt is output instead. When the confidence level of prediction results in multiple consecutive batches is lower than the threshold, the model retraining process is automatically triggered, and the model is remodeled using all accumulated data.

6. The finishing method for easily deformable parts as described in claim 5, characterized in that, After the processing data is collected, it undergoes a data preprocessing step before being stored and used as model input. The preprocessed data is stored in a structured manner according to processing batches, and batch indexes and time-series indexes are established to facilitate subsequent data retrieval and model training.

7. The finishing method for easily deformable parts as described in claim 6, characterized in that, Data preprocessing steps include: Outlier detection is performed on the collected raw data to remove invalid data points that exceed physical limits; missing data is filled by interpolation; time alignment and synchronization are performed on multi-source data from different sensors; and feature parameters with different dimensions are normalized to bring them to the same order of magnitude.

8. The finishing method for easily deformable parts as described in claim 7, characterized in that, Install a tool wear detection device on the processing equipment to monitor the wear status of the tool in real time; When the tool wear is detected to have reached the preset wear threshold, the tool compensation value is automatically applied to correct the current machining path. When the tool wear exceeds the threshold, a tool replacement prompt signal is issued.

9. The finishing method for easily deformable parts as described in claim 8, characterized in that, The coolant supply system continuously supplies cooling and lubricating medium to the cutting contact area between the tool and the workpiece. The supply flow rate, supply pressure, and supply temperature of the cooling and lubricating medium are adjusted in real time according to the machining stage and cutting parameters; the cooling and lubricating medium is cutting fluid, compressed air, oil mist, or a combination thereof.

10. The finishing method for easily deformable parts as described in claim 9, characterized in that, Compressed gas is sprayed into the processing area by an air blowing device to remove chips and impurities accumulated on the workpiece surface, fixture positioning surface and processing area. The air blowing device performs an air blowing cleaning operation once before each clamping and positioning and once after each processing. The parameters of the air blowing cleaning include air blowing pressure, air blowing duration and air blowing angle, which are adjusted according to different processing stages and workpiece structures.

Images