A high-precision hardware spring punching forming method

Abstract

The application relates to a high-precision hardware bullet stamping forming method in the field of high-end equipment manufacturing industry, and the method comprises the following steps: acquiring positioning data in a stamping process, collecting real-time displacement data of a driving system and a die through a high-precision sensor, adopting a digital signal processing technology to denoise the collected data, and obtaining a first positioning data set; adopting a finite element analysis technology to simulate a contact state of the die and material, acquiring an influence weight of die design parameters on positioning accuracy, and determining an optimized die geometric parameter data set; generating a digital model of a multi-station die according to the optimized die geometric parameter data set, verifying a process error of the die under multi-station continuity through a simulation technology, and obtaining a process error data set; acquiring stress distribution data of material in the stamping process, collecting real-time stress values through a strain sensor, adopting a gradual stress release algorithm to control the stamping process in stages, and obtaining a first stress distribution data set.

Description

TECHNICAL FIELD

[0001] The application relates to the field of high-end equipment manufacturing industry, in particular to the field of precision stamping forming intelligent control technology, and more particularly to a high-precision hardware spring stamping forming method. BACKGROUND

[0002] High-precision multi-station continuous stamping forming technology plays an important role in modern manufacturing industry, especially in the production of precision parts such as hardware springs. Its high precision and high efficiency directly affect product quality and production cost. With the increasing demand for miniaturization and high performance of electronic products, stamping forming technology needs to meet the requirements of micron-level precision and complex shape processing, becoming one of the key technologies to promote the upgrading of manufacturing industry.

[0003] However, the traditional hardware spring stamping process has significant limitations. Low positioning accuracy leads to large part size deviation, multi-process separation leads to low production efficiency, and material stress concentration during stamping process easily causes deformation or cracks. These problems limit the application of the process in the high-precision field. The core challenge lies in how to achieve high-precision positioning, effective stress management and process continuity in multi-station stamping process. The lack of positioning accuracy is due to the limitations of the drive system and die design, which makes it difficult to achieve micron-level control; material stress concentration is caused by the lack of staged release mechanism in the stamping process, leading to part deformation; multi-process separation increases the error accumulation and time cost between processes.

[0004] How to integrate high-precision positioning, progressive stress release and real-time process monitoring in multi-station continuous stamping to overcome the problems of positioning error, material deformation and low efficiency has become a key issue to realize high-precision hardware spring forming. SUMMARY

[0005] The application provides a high-precision hardware spring stamping forming method, comprising the following steps:

[0006] Obtain positioning data during stamping process, collect real-time displacement data of drive system and die through high-precision sensor, and use digital signal processing technology to denoise the collected data to obtain first positioning data set;

[0007] Extract displacement deviation features from the first positioning data set, compare the deviation value with the preset micron-level positioning threshold T_p, T_p represents the positioning accuracy threshold, and the unit is micron, if the deviation value is greater than T_p, adjust the drive system output through servo motor control algorithm to obtain second positioning data set;

[0008] For the second positioning data set, use finite element analysis technology to simulate the contact state of the die and the material, obtain the influence weight of the die design parameters on the positioning accuracy, and determine the optimized die geometric parameter data set;

[0009] According to the optimized mold geometry parameter data set, a digital model of the multi-station mold is generated, the process error between the mold under the multi-station continuity is verified through simulation technology, and a process error data set is obtained;

[0010] The error accumulation feature is extracted from the process error data set, and the error value is compared with the preset process error threshold T_e, T_e represents the process error threshold, and the unit is microns, if the error value is greater than T_e, the process parameter is adjusted through the inter-process compensation algorithm, and the optimized process parameter set is obtained;

[0011] Obtain the stress distribution data of the material in the stamping process, collect the real-time stress value through the strain sensor, and use the progressive stress release algorithm to control the stamping process in stages to obtain the first stress distribution data set;

[0012] The stress concentration region feature is extracted from the first stress distribution data set, and the stress value is compared with the preset stress threshold T_s, T_s represents the stress threshold, and the unit is Pascal, if the stress value is greater than T_s, the stress distribution is optimized by adjusting the stamping speed and mold gap parameters, and the second stress distribution data set is obtained;

[0013] For the second stress distribution data set and the optimized process parameter set, the dynamic data of the stamping process is collected by using a real-time process monitoring system, and the data stability is analyzed by using a machine learning algorithm to obtain a stamping process stability parameter set;

[0014] According to the stamping process stability parameter set, a control instruction set for the hardware spring stamping forming is generated, and the instruction is executed by a numerical control system to output a part size consistency data meeting the micron-level precision and complex shape machining requirements.

[0015] The technical scheme provided by the embodiment of the application can include the following beneficial effects:

[0016] The application discloses a high-precision hardware spring stamping forming method, which realizes accurate positioning by collecting positioning data through a high-precision sensor and performing digital signal processing in combination with a servo motor control algorithm. Finite element analysis is used to optimize mold design, and simulation technology is used to verify multi-station continuous machining precision. The stress distribution in the stamping process is controlled by using a strain sensor and a progressive stress release algorithm, and the stress state is optimized by adjusting the stamping parameters. Finally, process stability is analyzed based on a machine learning algorithm to generate control instructions to realize high-precision stamping. The application can effectively improve the size consistency and shape complexity of hardware springs, meet micron-level precision requirements, and significantly improve stamping forming quality and efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0017] Fig. 1 It is a flowchart of a high-precision hardware spring stamping forming method.

[0018] Fig. 2 A schematic diagram of a high-precision hardware bullet stamping forming method of the present application.

[0019] Fig. 3 Another schematic diagram of a high-precision hardware bullet stamping forming method of the present application. DETAILED DESCRIPTION

[0020] The technical solutions of the present application will be described clearly and completely below in conjunction with embodiments. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0021] As Figs. 1-3 , the high-precision hardware bullet stamping forming method of the present embodiment can specifically include:

[0022] In step S101, positioning data in the stamping process is obtained, real-time displacement data of the driving system and the die is collected by a high-precision sensor, and the collected data is denoised by using a digital signal processing technology to obtain a first positioning data set.

[0023] The initial displacement data set is obtained by collecting the displacement of the driving system and the die in the stamping process by a high-precision sensor. The initial displacement data set is denoised by using a digital signal processing technology to generate a first denoised displacement data set. The Fourier transform method is applied to the first displacement data set for frequency domain analysis, the key frequency components are extracted, and the periodic characteristics of the first displacement data set are determined. According to the periodic characteristics, a time series analysis model is constructed to predict the short-term change trend of the displacement of the driving system and the die, and a predicted displacement data set is obtained. If the change trend in the predicted displacement data set exceeds a preset threshold range, the data is marked with an abnormal point to generate an abnormal mark data set; if it does not exceed the threshold range, the predicted displacement data set is directly output. The abnormal points in the abnormal mark data set are verified again by comparing historical displacement data to determine whether the abnormality is a systematic deviation, and a verified displacement data set is obtained. The verified displacement data set is optimized by using a data smoothing technology to generate a final precise positioning data set.

[0024] Exemplarily, in the specific implementation of acquiring positioning data in the stamping process, first, the displacement data of the driving system and the die are collected in real time by high-precision sensors. Assuming that a laser displacement sensor is used, the measurement accuracy reaches 0.01 millimeters, and the sampling frequency is 1000 hertz, that is, 1000 data points are collected per second, covering the key displacement changes in the stamping period. For example, the dynamic change of the die displacement from 0 millimeters to 50 millimeters is recorded in the stamping pressing stage, to generate an original data set. Subsequently, the collected data is denoised. The wavelet transform algorithm in digital signal processing technology is adopted, specifically, the Daubechies wavelet basis function (db4) is selected, and the decomposition level is set to 5 layers. The noise interference is filtered by threshold processing (the soft threshold is set to 0.5 times the standard deviation) on the high-frequency noise component, while the main features of the displacement signal are retained. The signal-to-noise ratio of the processed signal is improved from the original 15 decibels to 25 decibels, and the first positioning data set is obtained. To verify the denoising effect, the frequency spectrum of the data before and after processing is analyzed. It is found that the noise frequency above 200 hertz has been effectively suppressed, while the main frequency of the displacement signal at about 10 hertz is retained, ensuring the accuracy of the data. Further, the first positioning data set is associated with the control system of the stamping equipment. The displacement data and the stamping force data (assuming that the peak force is 500 kilonewtons) are compared and analyzed by time stamping. The coupling relationship between displacement and force is analyzed. It is found that when the displacement reaches the maximum value of 50 millimeters, the stamping force approaches the peak value, and the correlation coefficient is as high as 0.95, indicating that the positioning data can provide a reliable basis for subsequent process optimization. The above processes are realized through an automatic program. The data collection is completed by a sensor interface module, and the denoising processing and analysis are realized by relying on an embedded processor to run a preset algorithm, ensuring real-time and accuracy.

[0025] In step S102, the displacement deviation feature is extracted from the first positioning data set, and the deviation value is compared with a preset micron-level positioning threshold T_p. T_p represents the positioning accuracy threshold, and the unit is microns. If the deviation value is greater than T_p, the driving system output is adjusted through a servo motor control algorithm to obtain a second positioning data set.

[0026] The displacement deviation features are extracted from the initial positioning data set, and each set of data is analyzed. Through a pre-established feature extraction model, the specific distribution of the displacement deviation is obtained. According to the displacement deviation distribution, the preset positioning threshold is compared. If the deviation value exceeds the threshold range, the corresponding deviation data point is recorded, and the data range that needs to be adjusted is determined. The servo motor control algorithm is used to optimize the parameters of the driving system. By calculating the correction amount in real time, the adjusted driving instruction set is generated. The output state of the driving system is adjusted through the driving instruction set, and the adjusted positioning data is obtained to form a new positioning data set. For the new positioning data set, the displacement deviation features are extracted again and compared with the initial deviation features to determine whether the adjusted positioning accuracy meets the preset threshold requirements. If the positioning accuracy still does not meet the preset threshold requirements, the support vector machine algorithm is used to classify the deviation data to obtain the classified deviation mode and determine the further optimized parameter direction. According to the classified deviation mode, the input parameters of the servo motor control algorithm are adjusted to generate the final driving instruction set and obtain the optimized positioning data set.

[0027] For example, when extracting displacement deviation features from the first positioning data set, assume that the initial positioning data is collected by a high-precision laser sensor to obtain a set of coordinate data, for example, the actual position of the target point is (100.5, 200.3) microns, and the expected position is (100.0, 200.0) microns. The displacement deviation is calculated as x-axis 0.5 microns and y-axis 0.3 microns, and the comprehensive deviation value is sqrt(0.5^2+0.3^2) = 0.583 microns. Then, the deviation value is compared with the preset micron-level positioning threshold T_p. Assume that T_p is set to 0.4 microns. Through the algorithm, it is judged that 0.583 microns is greater than 0.4 microns, triggering the adjustment mechanism. Then, the servo motor control algorithm is used to adjust the output of the driving system. The PID control algorithm is adopted, with the proportional coefficient Kp = 0.8, the integral coefficient Ki = 0.2, and the differential coefficient Kd = 0.1. According to the deviation value 0.583 microns, the control amount is calculated, and the adjustment instruction is output to make the motor fine-tune. The x-axis is compensated by -0.5 microns, and the y-axis is compensated by -0.3 microns. The second positioning data set is collected in real time through the sensor after adjustment, and the new position is (100.1, 200.1) microns. The new deviation is calculated as x-axis 0.1 microns and y-axis 0.1 microns, and the comprehensive deviation value is 0.141 microns. The analysis process shows that the new deviation value is less than T_p = 0.4 microns, meeting the accuracy requirements. If it still does not meet the requirements, the PID adjustment is executed in a loop until the deviation converges within the threshold. Logically, the initial deviation calculation is compared with the threshold, and then the control adjustment and result verification are performed to form a closed-loop control chain, ensuring that the positioning accuracy is gradually optimized. At the same time, through the combination of sensor data feedback and algorithm calculation, automatic processing is realized, and the system stability is improved.

[0028] In step S103, the finite element analysis technology is used to simulate the contact state of the mold and the material for the second positioning data set, to obtain the influence weight of the mold design parameters on the positioning accuracy, and to determine the optimized mold geometric parameter data set.

[0029] The contact state of the mold and the material is modeled by the finite element analysis technology, the correlation data between the mold design parameters and the positioning accuracy is obtained, and a preliminary influence weight distribution is obtained. According to the influence weight distribution, a sensitivity analysis is performed on the geometric parameters in the mold design, and a key parameter set whose influence on the positioning accuracy is greater than a preset threshold is determined. The genetic algorithm is used to iteratively optimize the key parameter set, the simulation results of each round are obtained through multiple rounds of adjustment, and the optimization direction of the parameter combination is determined. If the simulation result shows that the positioning accuracy does not reach the preset threshold, the geometric parameters are fine-tuned, new contact state data is obtained by running the finite element analysis again, and updated accuracy performance data is obtained. According to the updated accuracy performance data, the interaction between the material properties and the contact state is analyzed, potential influencing factors are extracted, and the parameter adjustment range for further optimization is determined. For the parameters in the adjustment range, the final simulation analysis is performed in combination with the influencing factors, the optimal mold geometric parameter combination is obtained through comprehensive evaluation, and the final optimized data set is obtained.

[0030] For the second positioning data set, the contact state of the mold and the material is simulated using finite element analysis technology, and the influence weight of the mold design parameters on the positioning accuracy is obtained, and finally the optimized mold geometric parameter data set is determined. The specific implementation method is as follows. First, a contact model of the mold and the material is established by a finite element analysis software (such as ANSYS). It is assumed that the mold material is steel, the elastic modulus is 210 GPa, the Poisson's ratio is 0.3, the material is aluminum alloy, the elastic modulus is 70 GPa, the Poisson's ratio is 0.33, the contact surface friction coefficient is set to 0.15, the grid division adopts tetrahedral element, and the grid size is 0.5 mm to ensure the calculation accuracy. Second, when simulating the contact state, the boundary conditions are applied, the mold is fixed, the material is subjected to a pre-tightening force of 10 N, the nonlinear contact algorithm (such as penalty function method) is used to solve the contact pressure distribution, the maximum contact stress is 25 MPa, and the displacement deviation is 0.02 mm. Then, the influence of the mold design parameters (such as mold angle and curvature radius) on the positioning accuracy is analyzed. It is assumed that the mold angle ranges from 30 to 60 degrees, and the curvature radius ranges from 5 to 15 mm. Through orthogonal experimental design, 9 groups of parameter combinations are set, the positioning error under each group of parameters is calculated, the error range is between 0.01 and 0.05 mm, and the regression analysis algorithm is used to obtain the influence weight of the mold angle on the positioning accuracy, which is 0.65, and the weight of the curvature radius is 0.35. Finally, based on the weight result, the genetic algorithm is used to optimize the mold geometric parameters, the population size is set to 50, the iteration number is set to 100, the objective function is set to minimize the positioning error (the target value is less than 0.01 mm), the mold angle is set to 45 degrees, and the curvature radius is set to 10 mm. New data set is obtained, and the positioning error is reduced to 0.008 mm through re-finite element simulation verification, which meets the design requirements. The above process is completed through software automatic calculation and algorithm iteration, forming a complete logical chain from modeling to optimization, ensuring the scientificity and consistency of the results.

[0031] In step S104, a digital model of the multi-station mold is generated according to the optimized mold geometric parameter data set, and the process error data set is obtained by simulating the mold under the multi-station continuity.

[0032] The initial data is obtained from the optimized mold geometry parameter data set, the structure characteristics of the multi-station mold are classified and processed, and a preliminary mold design parameter set is obtained. According to the preliminary mold design parameter set, a digital model of the multi-station mold is constructed, three-dimensional structure data is generated by data modeling technology, and the performance form of the digital model in the virtual environment is determined. The continuity of the three-dimensional structure data is analyzed by simulation technology, the running state of the multi-station mold under different processes is simulated, and the preliminary distribution data of the error between processes is obtained. According to the preliminary distribution data of the error between processes, the finite element analysis method is applied for verification, and it is judged whether the error exceeds the preset threshold range. If it exceeds, the digital model is adjusted locally, and the adjusted error distribution data is obtained. Through the adjusted error distribution data, combined with the process verification logic, each process of the multi-station mold is compared one by one. If the error value of a certain process is higher than the preset threshold, the parameters of the process are optimized and iterated to determine the final process error data set. According to the final process error data set, the error mapping relationship of the multi-station mold under the continuity analysis is generated, the regression analysis algorithm is used to predict the error trend, and the regularity data of the error change is obtained.

[0033] For example, based on the optimized mold geometry parameter dataset, a digital model of the multi-station mold is first generated. The optimized parameter data can be automatically imported into the computer-aided design software, for example, the punch diameter of the mold is 25.5 mm, the die gap is 0.2 mm, and the total length of the mold is 500 mm. The three-dimensional modeling algorithm is used to generate the geometry of each station, and the spatial distribution between stations is automatically calculated to ensure that the station spacing error is less than 0.01 mm. Then the model is exported in STL format for simulation. Next, the simulation technology is used to verify the process error between stations under the continuity of the mold. The finite element analysis algorithm is used to set the material parameters such as the Young's modulus of the steel plate to be 200 GPa and the Poisson's ratio to be 0.3 in the simulation software to simulate the stress distribution and deformation during the multi-station stamping process. The calculated displacement error from the first station to the second station is 0.015 mm, and the error from the second station to the third station is 0.012 mm. Through iterative analysis of 100 times, a process error dataset is generated, which includes the error value and its distribution law of each station, and the error standard deviation is 0.003 mm. Further analysis of the error dataset is performed using a statistical regression algorithm to predict the error accumulation trend. It is found that the error accumulation is mainly affected by the mold gap, and the correlation coefficient reaches 0.85, which provides data support for subsequent optimization. To form a rigorous logical relationship, the error dataset is combined with the mold life prediction model. Assuming that the mold is used 100,000 times, the error at the end of life may reach 0.05 mm, which exceeds the allowable range of 0.03 mm. Therefore, an optimization suggestion is automatically generated to adjust the gap parameter to 0.18 mm, and the simulation is verified again to ensure that the error is controlled within 0.02 mm, forming a closed-loop optimization logic.

[0034] In step S105, the error accumulation feature is extracted from the process error dataset, and the error value is compared with the preset process error threshold T_e. T_e represents the process error threshold, and the unit is microns. If the error value is greater than T_e, the process parameter is adjusted through the inter-process compensation algorithm to obtain the optimized process parameter set.

[0035] The process error data is obtained from a storage system, and cleaning and formatting are performed on the process error data to obtain a structured error data set. According to the structured error data set, a feature extraction method is used to identify and quantify the error accumulation characteristics, and the distribution pattern of error accumulation is determined. The distribution pattern of error accumulation is compared with a preset process error threshold. If a certain group of error values exceeds the preset process error threshold range, it is marked as an abnormal data group, and the detailed features of the abnormal data group are obtained. Through the detailed features of the abnormal data group, a pre-established compensation algorithm model is called to adjust the process parameters corresponding to the abnormal data group to obtain a preliminary adjusted parameter set. According to the preliminary adjusted parameter set, historical process data is combined for simulation verification to determine whether the parameter adjustment meets the expected range. If it does not meet the expected range, the parameter set is fine-tuned again to determine the final adjustment parameters. The final adjustment parameters are used to configure and update the process system, and updated process running data is obtained to determine whether the running data is stable within the preset process error threshold range.

[0036] Exemplary, the specific implementation method of extracting error accumulation features from the process error data set and subsequent processing is as follows: assuming that we have a process error data set containing error values of 10 processes, in microns, which are 2.5, 3.1, 1.8, 4.2, 2.9, 3.7, 2.1, 5.0, 3.3, and 2.6. First, the error values are subjected to cumulative feature extraction by the data preprocessing module, and the cumulative error value of each process is calculated using the cumulative sum algorithm, for example, the cumulative value of the first process is 2.5, the second is 2.5+3.1=5.6, the third is 5.6+1.8=7.4, and so on, and finally the cumulative error sequence is [2.5, 5.6, 7.4, 11.6, 14.5, 18.2, 20.3, 25.3, 28.6, 31.2]. Then, the cumulative error values are compared with the preset process error threshold T_e=20.0 microns, and it is found through the comparison algorithm that the cumulative error of the 7th process 20.3 has exceeded the threshold, and the cumulative errors of the 8th to 10th processes also continue to exceed the threshold, and the system automatically marks these processes as objects to be adjusted. For the processes exceeding the threshold, the inter-process compensation algorithm is called to optimize the parameters, assuming that the compensation algorithm is based on error proportion adjustment, the formula is adjustment amount=(cumulative error-T_e)×0.5, the adjustment amount of the 7th process is (20.3-20.0)×0.5=0.15 microns, the system feeds back this adjustment amount to the process parameter control module, automatically updates the processing parameters of the 7th process, such as reducing the processing speed or adjusting the tool angle, and performs linked adjustment on the subsequent processes to ensure that the cumulative errors of the 8th to 10th processes gradually return below the threshold. Finally, the optimized parameter set is verified by simulation, and the cumulative error of the 10th process after adjustment is reduced to 19.8 microns, which is lower than the threshold, the system records the optimization result and generates a report, and completes the automatic processing of the whole process.

[0037] In step S106, the stress distribution data of the material in the stamping process is obtained, the real-time stress value is collected by the strain sensor, the gradual stress release algorithm is used to control the stamping process in stages, and the first stress distribution data set is obtained.

[0038] The initial stress data set is obtained by collecting material stress in the stamping process in real time through a strain sensor. According to the initial stress data set, the stamping process is divided into stages by using a progressive stress release algorithm, and stress control parameters of each stage are determined. For the stress control parameters of each stage, the change trend of the material stress is monitored in real time to obtain dynamic stress distribution information. If the dynamic stress distribution information exceeds a preset threshold range, the stamping process is intervened by adjusting the release control strategy to obtain an adjusted stress data set. According to the adjusted stress data set, the uniformity of stress distribution is analyzed to determine whether there is a local stress concentration area. The data of the local stress concentration area is processed again, the control parameters are optimized in combination with the progressive stress release algorithm, and a final first stress distribution data set is obtained. If the first stress distribution data set still has an abnormal area, the data of the abnormal area is marked to generate reference information for subsequent processing.

[0039] For example, when acquiring the stress distribution data of the material in the stamping process, high-precision strain sensors can be installed at key positions of the stamping die, such as sensors with a range of 0.01 to 500 MPa, and the sampling frequency is set to 1000 Hz to ensure real-time capture of small stress changes during the stamping process. The data is directly uploaded to the cloud server for storage and processing through a wireless transmission module. Then, for the collected real-time stress values, a progressive stress release algorithm is used for analysis. The specific method is to divide the stamping process into three stages: initial contact, plastic deformation, and final forming. The stress threshold of each stage is set to 50 MPa, 200 MPa, and 350 MPa, respectively. The stress data collected by the sensor is segmented and fitted by the algorithm, and the stress change rate of each stage is calculated, such as the stress change rate of the initial contact stage is 2.5 MPa / s, which is used as the basis to determine whether the stamping speed or pressure needs to be adjusted. Subsequently, based on the above stage control strategy, the system automatically generates a first stress distribution data set. The data set is indexed by time stamp, recording the stress value and corresponding position at each time point (accurate to milliseconds), such as at time point t = 0.1 s, the stress value at the center of the die is 48.3 MPa. Through two-dimensional visualization analysis of the data set, it can be found that the non-uniformity of stress distribution is concentrated in the edge area of the die, and the maximum stress value reaches 380.2 MPa, which is 10% higher than the safety threshold. The system automatically triggers the warning mechanism and optimizes the stamping parameters, such as reducing the stamping speed to 0.5 m / s to reduce the edge stress concentration. The above process forms a closed-loop logic through sensor data acquisition, algorithm analysis, and system automation control, ensuring the accuracy of the stress distribution data and the controllability of the stamping process. At the same time, it is associated with subsequent material performance analysis services, and the data set can be used as input to predict the fatigue life of the material, such as through the relationship model between stress value and cycle number to calculate the durability of the material under a certain stress as 10000 cycles, thereby providing data support for process improvement.

[0040] In step S107, the stress concentration region features are extracted from the first stress distribution data set, and the stress value is compared with the preset stress threshold T_s, T_s represents the stress threshold, unit is Pascal, if the stress value is greater than T_s, the stress distribution is optimized by adjusting the stamping speed and die gap parameters, and the second stress distribution data set is obtained.

[0041] The characteristics of the stress concentration area are extracted from the initial stress distribution data, and the stress values of the stress concentration area are quantitatively processed by using a feature extraction technique to obtain a preliminary stress distribution characteristic map. According to the preliminary stress distribution characteristic map, the stress values of each stress concentration area are compared with a preset threshold value. If the stress value of a certain area is greater than the preset threshold value, the position and the over-threshold amplitude of the area are recorded, and the area range that needs to be optimized is determined. After obtaining the area range that needs to be optimized, the stamping speed and the die gap parameters are preliminarily adjusted, and a pre-established finite element analysis model is used to simulate the adjusted parameters to obtain simulated stress distribution data. By using the simulated stress distribution data, it is determined whether the stress values of each area exceed the preset threshold value. If there are still areas exceeding the preset threshold value, the stamping speed and the die gap parameters are further fine-tuned to generate new simulated stress distribution data. According to the new simulated stress distribution data, a support vector machine algorithm is used to classify and evaluate the parameter adjustment effect to determine the optimal parameter combination scheme. By using the optimal parameter combination scheme, the initial stress distribution data is updated to generate a final optimized stress distribution data set.

[0042] For example, when extracting the stress concentration area characteristics from the first stress distribution data set, the stress distribution of the stamping part model can be calculated by using a finite element analysis software. It is assumed that the initial data set contains stress values of 10,000 grid points, with a unit of pascal and a range of 50,000 to 5,000,000. A clustering algorithm such as K-means is used to classify the grid points with higher stress values as concentration areas, and the K value is set to 5. After iterative calculation, the cluster with the highest average stress value is extracted as the target area, and the average value is assumed to be 3,000,000 pascals. Then, the stress values of each grid point are compared with a preset stress threshold T_s (set to 2,000,000 pascals), and a loop algorithm is used to traverse the data set. If the stress value of a certain point is greater than T_s, it is marked as an over-threshold point. The statistical results show that about 20% of the points exceed the standard, and the maximum value reaches 4,500,000 pascals. To optimize the stress distribution, the over-threshold points are simulated and optimized by adjusting the stamping speed and the die gap parameters. The initial stamping speed is set to 2.5 meters per second, and the die gap is 0.5 millimeters. Based on the gradient descent algorithm, the parameters are adjusted, and the speed is reduced by 0.1 meters per second and the gap is increased by 0.05 millimeters each iteration. After 10 simulations, it is found that when the speed is reduced to 1.8 meters per second and the gap is increased to 0.8 millimeters, the stress distribution tends to be uniform. Finally, a second stress distribution data set is generated, the stress values are recalculated, the proportion of over-threshold points is reduced to 5%, and the maximum value is reduced to 2,200,000 pascals, verifying the optimization effect. Through data comparison and analysis, the optimized stress distribution is more in line with the design requirements, and it can be associated with the subsequent fatigue life prediction module to input the stress data and calculate the durability of the stamping part under cyclic loading, ensuring product reliability and forming a complete technical closed loop.

[0043] In step S108, dynamic data of the stamping process is collected by the real-time process monitoring system for the second stress distribution data set and the optimized process parameter set, the data stability is analyzed by the machine learning algorithm to obtain a set of stamping process stability parameters.

[0044] The dynamic monitoring data is obtained from the sensor network in the stamping process to obtain an initial data set. For the initial data set, a filtering technique is used to remove noise and abnormal points to obtain a cleaned data set. According to the cleaned data set, stress distribution features are extracted, the fluctuation range is analyzed by a statistical method, and a characteristic stress distribution description is determined. According to the characteristic stress distribution description, combined with the influence of process parameters, a support vector machine algorithm is used to classify the data stability to determine the preliminary range of stability parameters. If the preliminary range of stability parameters exceeds the preset threshold, the dynamic monitoring data is analyzed again, combined with the change trend, to obtain an adjusted set of stability parameters. Through the adjusted set of stability parameters, parameter optimization is performed in real time for the stamping process to determine the optimized process parameter configuration. According to the optimized process parameter configuration, the process monitoring control logic is updated to determine whether the stamping process meets the stability requirements to obtain a final set of operating parameters.

[0045] For example, for the processing of the second stress distribution data set and the optimized process parameter set, first, the dynamic data in the stamping process is collected by the real-time process monitoring system. Assuming that the system collects the pressure value, displacement value and temperature value of the stamping equipment at a frequency of 100 times per second, a data set containing 100,000 records is obtained, in which the pressure value ranges from 500 to 2000 Newton, the displacement value ranges from 0.5 to 5.0 millimeters, and the temperature value ranges from 20 to 80 degrees Celsius. Then, these data are input into the machine learning algorithm for stability analysis. The support vector machine (SVM) algorithm is selected for data classification, the kernel function is set to radial basis function (RBF), the optimal parameter C value is determined to be 1.0 and the gamma value is determined to be 0.1 through cross-validation, and the classification accuracy is 92.5% after training the model, indicating that the model has strong recognition ability for data volatility. Subsequently, the stability parameter set of the stamping process is analyzed, the standard deviation of the pressure value is calculated to be 15.3 Newton, the standard deviation of the displacement value is calculated to be 0.2 millimeters, the standard deviation of the temperature value is calculated to be 2.7 degrees Celsius, and the proportion of abnormal points output by the SVM model is 3.2%. The stability parameter set of the stamping process is comprehensively judged, in which the pressure stability index is set to 0.85 (full score 1.0), the displacement stability index is 0.90, and the temperature stability index is 0.88. These parameters are further compared with the optimized process parameter set, and it is found that the pressure stability needs to be improved, which can be optimized by adjusting the stamping speed from 50 times / minute to 45 times / minute. The system automatically generates adjustment instructions and feeds back to the equipment control module to form a closed-loop control logic, ensuring that the subsequent production process stability is improved to a target value of 0.90 or more. Through the complete process of data collection, algorithm analysis and parameter optimization, the dynamic monitoring and improvement of the stamping process are realized, and the logic forms a closed loop from data to analysis to optimization, which is closely related to the process parameter adjustment business, ensuring the feasibility and actual effect of the technology implementation.

[0046] Step S109, according to the stability parameter set of the stamping process, the control instruction set of the hardware shell fragment stamping forming is generated, and the instruction is executed through the numerical control system to output the part size consistency data meeting the micron-level precision and complex shape processing requirements.

[0047] The stability parameter data in the stamping process is acquired to generate an initial stability parameter set. For the initial stability parameter set, a preset stamping forming rule is adopted to generate a corresponding control instruction set. The control instruction set is loaded to a numerical control system to drive the stamping equipment to run, and real-time feedback data in the stamping process is acquired. It is judged whether the real-time feedback data conforms to a preset standard. If there is a deviation between the real-time feedback data and the preset standard, the parameters in the control instruction set are adjusted to generate a corrected instruction set. According to the corrected instruction set, the machining precision data in the stamping process is continuously monitored to acquire dynamic change data of the part size. The support vector machine algorithm is adopted to classify and process the dynamic change data of the part size to judge whether the machining consistency reaches a preset precision target. If the classification result shows that part of the part size does not reach the preset precision target, the execution record of the stability parameter and the control instruction set is traced back to determine the deviation source information.

[0048] For example, in the process of generating and executing control instructions for stamping metal fragments, first, through the analysis of the parameter set of the stamping process stability, key parameters such as stamping force, die gap and material thickness are extracted. Assuming that the stamping force is 5000 Newton, the die gap is 0.02 mm, and the material thickness is 0.5 mm, the algorithm is used to evaluate the stability of these parameters, and the standard deviation formula is used to calculate the stamping force fluctuation value. Assuming that the stamping force data in 10 tests is 4998, 5002, 5000, 4999, 5001, 5003, 4997, 5004, 5000, 4996 Newton, the standard deviation is 2.6 Newton, and the stability is determined to be within the acceptable range (standard deviation less than 5 Newton). Subsequently, based on these parameters, a set of control instructions is generated, and the stamping force, die gap and other parameters are converted into executable code of the numerical control system through programming, for example, setting the stamping speed to 300 times per minute, the stamping depth error is controlled within 0.001 mm, and the stamping force is adjusted in real time through the feedback algorithm. If the deviation exceeds 0.5 Newton, it is automatically compensated to the target value of 5000 Newton. Next, the numerical control system executes the instructions to drive the stamping equipment to run, and real-time processing data such as stamping depth and part size is collected, and the size data after each stamping is recorded by sensors. Assuming that 100 parts are collected, the size target value is 10.000 mm, and the actual measurement value range is 9.998 to 10.002 mm, the average value is 10.0003 mm, and the variance is 0.0004, indicating that the size consistency meets the micron-level precision requirement. Finally, the part size consistency data is output, a statistical report is generated, and a size distribution chart and consistency index are automatically generated by data analysis software, with an index value of 0.95 (full score is 1), indicating that the processing result meets the complex shape processing requirement. If the consistency index is less than 0.9, the system automatically triggers the parameter optimization algorithm to adjust the stamping force to 5050 Newton to improve the stability, forming a closed-loop control logic to ensure continuous optimization of subsequent processing precision.

[0049] The specific embodiments of the present application are described above. It should be understood that the present application is not limited to the above specific embodiments, and those skilled in the art can make various changes or modifications within the scope of the claims, which does not affect the essential content of the present application. The embodiments of the present application and the features in the embodiments can be arbitrarily combined without conflict.

Claims

1. A high-precision metal spring stamping forming method, characterized in that, The method includes the following steps: Step S101: Obtain positioning data during the stamping process. Real-time displacement data of the drive system and mold are collected by high-precision sensors. Digital signal processing technology is used to denoise the collected data to obtain the first positioning dataset. Step S102: Extract displacement deviation features from the first positioning dataset, and compare the deviation value with the preset micron-level positioning threshold T_p, where T_p represents the positioning accuracy threshold in micrometers. If the deviation value is greater than T_p, adjust the output of the drive system through the servo motor control algorithm to obtain the second positioning dataset. Step S103: For the second positioning dataset, use finite element analysis technology to simulate the contact state between the mold and the material, obtain the influence weight of the mold design parameters on the positioning accuracy, and determine the optimized mold geometric parameter dataset. Step S104: Based on the optimized mold geometric parameter dataset, generate a digital model of the multi-station mold, and verify the inter-process error of the mold under multi-station continuity through simulation technology to obtain the process error dataset. Step S105: Extract the cumulative error feature from the process error dataset, and compare the error value with the preset process error threshold T_e, where T_e represents the process error threshold in micrometers. If the error value is greater than T_e, adjust the process parameters through the inter-process compensation algorithm to obtain an optimized process parameter set. Step S106: Obtain stress distribution data of the material during the stamping process. Collect real-time stress values ​​through strain sensors and use a progressive stress relief algorithm to control the stamping process in stages to obtain the first stress distribution dataset. Step S107: Extract stress concentration region features from the first stress distribution dataset, and compare the stress value with the preset stress threshold T_s, where T_s represents the stress threshold in Pascals. If the stress value is greater than T_s, optimize the stress distribution by adjusting the stamping speed and die clearance parameters to obtain the second stress distribution dataset. Step S108: For the second stress distribution dataset and the optimized process parameter set, the dynamic data of the stamping process is collected by the real-time process monitoring system, and the data stability is analyzed by machine learning algorithm to obtain the stamping process stability parameter set; Step S109: Based on the set of stability parameters of the stamping process, generate a set of control instructions for the stamping of metal spring sheets, execute the instructions through the CNC system, and output part size consistency data that meets the requirements of micron-level precision and complex shape processing.

2. The high-precision metal spring stamping method according to claim 1, characterized in that, Step S101 includes: The initial displacement dataset is obtained by real-time data acquisition of the drive system and die displacement during the stamping process using high-precision sensors. The initial displacement dataset is denoised using digital signal processing techniques to generate a denoised first displacement dataset. For the first displacement dataset, the Fourier transform method is applied to perform frequency domain analysis, extract key frequency components, and determine the periodic characteristics of the first displacement dataset. Based on the periodicity characteristics, a time series analysis model is constructed to predict the short-term trend of the displacement of the drive system and the mold, and a predicted displacement dataset is obtained. If the trend of change in the predicted displacement dataset exceeds the preset threshold range, then the data is marked as an anomaly and an anomaly-marked dataset is generated. If the value does not exceed the threshold range, the predicted displacement dataset is output directly. For the aforementioned anomaly-marked dataset, the anomalies are verified a second time by comparing them with historical displacement data to determine whether the anomalies are systematic deviations, thus obtaining the verified displacement dataset. The verified displacement dataset is optimized using data smoothing techniques to generate the final precise positioning dataset.

3. The high-precision metal spring stamping method according to claim 1, characterized in that, Step S102 includes: Displacement deviation features are extracted from the initial positioning dataset, and each set of data is analyzed. The specific distribution of the displacement deviation is obtained through a pre-established feature extraction model. Based on the displacement deviation distribution, a comparison is made with a preset positioning threshold. If the deviation value exceeds the threshold range, the corresponding deviation data point is recorded to determine the data range that needs to be adjusted. The parameters of the drive system are optimized by using a servo motor control algorithm. The adjusted drive instruction set is generated by calculating the deviation correction amount in real time. The output state of the drive system is adjusted by the drive instruction set, and the adjusted positioning data is obtained to form a new positioning dataset. For the new positioning dataset, the displacement deviation features are extracted again and compared with the initial deviation features to determine whether the adjusted positioning accuracy meets the preset threshold requirements. If the positioning accuracy still does not meet the preset threshold requirement, the deviation data is classified using the support vector machine algorithm to obtain the classified deviation pattern and determine the direction of further optimization parameters. Based on the categorized deviation patterns, the input parameters of the servo motor control algorithm are adjusted to generate the final drive instruction set, resulting in an optimized positioning dataset.

4. A high-precision metal spring stamping method according to any one of claims 1-3, characterized in that, Step S103 includes: The contact state between the mold and the material is modeled using finite element analysis technology to obtain the correlation data between the mold design parameters and the positioning accuracy, and to obtain the preliminary influence weight distribution. Based on the aforementioned influence weight distribution, a sensitivity analysis is performed on the geometric parameters in the mold design to determine the set of key parameters whose impact on positioning accuracy exceeds a preset threshold. A genetic algorithm is used to iteratively optimize the set of key parameters. The simulation results of each round are obtained through multiple rounds of adjustments to determine the optimization direction of the parameter combination. If the simulation results show that the positioning accuracy does not reach the preset threshold, the geometric parameters are fine-tuned, and new contact state data is obtained by running the finite element analysis again to obtain updated accuracy performance data. Based on the updated accuracy performance data, the interaction between material properties and contact state is analyzed to extract potential influencing factors and determine the parameter adjustment range for further optimization. For the parameters within the adjustment range, a final simulation analysis is conducted in conjunction with the influencing factors. The optimal combination of mold geometric parameters is obtained through comprehensive evaluation, resulting in the final optimized dataset.

5. A high-precision metal spring stamping method according to any one of claims 1-3, characterized in that, Step S104 includes: Initial data is obtained from the optimized mold geometry parameter dataset, and the structural characteristics of multi-station molds are classified to obtain a preliminary set of mold design parameters. Based on the preliminary set of mold design parameters, a digital model of a multi-station mold is constructed, and three-dimensional structural data is generated using data modeling technology to determine the representation of the digital model in a virtual environment. The three-dimensional structural data were analyzed using simulation technology to simulate the operation of a multi-station mold under different processes, and preliminary distribution data of inter-process errors were obtained. The preliminary distribution data of the inter-process error is verified by the finite element analysis method to determine whether the error exceeds the preset threshold range. If it does, the digital model is locally adjusted to obtain the adjusted error distribution data. Using the adjusted error distribution data and combined with the process verification logic, each process of the multi-station mold is compared one by one. If the error value of a certain process is higher than the preset threshold, the parameters of that process are optimized and iterated to determine the final process error dataset. Based on the final process error dataset, the error mapping relationship of multi-station molds under continuous analysis is generated, and the error trend is predicted by regression analysis algorithm to obtain regularity data of error changes.

6. A high-precision metal spring stamping method according to any one of claims 1-3, characterized in that, Step S105 includes: Process error data is obtained from the storage system, and the process error data is cleaned and formatted to obtain a structured error dataset; Based on the structured error dataset, feature extraction methods are used to identify and quantify the error accumulation features, and to determine the distribution pattern of error accumulation. The distribution pattern of the accumulated error is compared with a preset process error threshold. If a certain group of error values ​​exceeds the preset process error threshold range, it is marked as an abnormal data group, and the detailed characteristics of the abnormal data group are obtained. By using the detailed characteristics of the abnormal data group, a pre-established compensation algorithm model is invoked to adjust the process parameters corresponding to the abnormal data group, resulting in a preliminarily adjusted parameter set. Based on the initially adjusted parameter set, simulation verification is performed using historical process data to determine whether the parameter adjustment meets the expected range. If it does not meet the expected range, the parameter set is fine-tuned a second time to determine the final adjustment parameters. The process system is configured and updated using the final adjustment parameters, and the updated process operation data is obtained. It is then determined whether the operation data is stable within the preset process error threshold range.

7. A high-precision metal spring stamping method according to any one of claims 1-3, characterized in that, Step S106 includes: The material stress during the stamping process is collected in real time by strain sensors to obtain an initial stress data set; Based on the initial stress data set, a progressive stress relief algorithm is used to divide the stamping process into stages and determine the stress control parameters for each stage. For the stress control parameters at each stage, the changing trend of material stress is monitored in real time to obtain dynamic stress distribution information; If the dynamic stress distribution information exceeds the preset threshold range, the stamping process is intervened by adjusting the release control strategy to obtain the adjusted stress data set. Based on the adjusted stress data set, the uniformity of stress distribution is analyzed to determine whether there are local stress concentration areas. The data from the local stress concentration area are processed a second time, and the control parameters are optimized by combining a progressive stress release algorithm to obtain the final first stress distribution dataset. If there are still abnormal regions in the first stress distribution dataset, the data in the abnormal regions are marked to generate reference information for subsequent processing.

8. A high-precision metal spring stamping method according to any one of claims 1-3, characterized in that, Step S107 includes: Features of stress concentration regions are extracted from the initial stress distribution data, and the stress values ​​of the stress concentration regions are quantified using feature extraction techniques to obtain a preliminary stress distribution feature map. Based on the preliminary stress distribution feature map, the stress value of each stress concentration region is compared with a preset threshold. If the stress value of a certain region is greater than the preset threshold, the location and the magnitude of the exceedance of the threshold of the region are recorded to determine the range of regions that need to be optimized. After obtaining the area to be optimized, the stamping speed and die clearance parameters are initially adjusted, and the adjusted parameters are simulated using a pre-established finite element analysis model to obtain simulated stress distribution data. Using the simulated stress distribution data, it is determined whether the stress value in each region exceeds the preset threshold. If there are still regions that exceed the preset threshold, the stamping speed and die clearance parameters are further fine-tuned to generate new simulated stress distribution data. Based on the new simulated stress distribution data, the support vector machine algorithm is used to classify and evaluate the effect of parameter adjustment, and determine the optimal parameter combination scheme. The initial stress distribution data is updated using the optimal parameter combination scheme to generate the final optimized stress distribution dataset.

9. A high-precision metal spring stamping method according to any one of claims 1-3, characterized in that, Step S108 includes: Dynamic monitoring data is acquired from the sensor network during the stamping process to obtain an initial data set; For the initial dataset, filtering techniques are used to remove noise and outliers, resulting in a cleaned dataset. Based on the cleaned data set, stress distribution characteristics are extracted, fluctuation range is analyzed using statistical methods, and a characteristic stress distribution description is determined. Based on the described characteristic stress distribution and considering the influence of process parameters, a support vector machine algorithm is used to classify the data stability and determine the preliminary range of stability parameters. If the initial range of the stability parameters exceeds the preset threshold, the dynamic monitoring data is analyzed a second time, and the adjusted stability parameter set is obtained by combining the changing trend. Using the adjusted stability parameter set, parameters are optimized based on real-time feedback from the stamping process to determine the optimized process parameter configuration. Based on the optimized process parameter configuration, update the process monitoring and control logic, determine whether the stamping process meets the stability requirements, and obtain the final operating parameter set.

Images