A multi-process composite stamping forming die control method and system

Abstract

This application provides a control method and system for multi-stage composite stamping forming dies, applied in the field of die manufacturing technology. By synchronously acquiring multi-source operational status data streams during the multi-stage stamping process and extracting associated characteristic parameters reflecting physical coordination relationships to construct a standard operating mode benchmark, the method then calculates the deviation of the current stamping cycle in real time and identifies the deviation type. Finally, it adaptively compensates and adjusts the motion control parameters or process load parameters of the stamping actuator. This method effectively solves the problem in existing technologies where persistent, low-amplitude deviations caused by differences in the microscopic properties of raw materials and accumulated wear of die components are difficult to identify and compensate for. It can accurately capture subtle, cumulative deviations during the stamping process and perform targeted compensation based on the deviation type, thereby avoiding batch defects in products and significantly improving product yield and production stability.

Description

Technical Field

[0001] This application relates to the field of mold manufacturing technology, and in particular to a method and system for controlling multi-process composite stamping molds. Background Technology

[0002] In modern industrial production, multi-stage composite stamping forming die systems are key equipment for manufacturing precision parts. The core of their control method lies in precisely defining the sequence and execution logic of each process. This logic is based on preset ideal material properties and stable equipment operation to ensure the coordinated operation of each process.

[0003] However, challenges exist in actual production: even if raw materials pass routine inspections, slight differences in microstructure, internal stress, and other deep-seated characteristics may exist between batches. During continuous high-speed stamping, these differences cause the actual springback of the material to deviate slightly from the preset compensation value of the control system. This continuous, minute deviation gradually affects the positioning reference of subsequent processes and the stress distribution inside the die.

[0004] At this point, the real-time position and pressure data collected by the sensors inside the mold, although still within the lenient alarm threshold, have stably deviated from the historical data trajectory of good products (e.g., slightly higher pressure peaks, slightly off-target position feedback). Simultaneously, the long-term non-ideal stress distribution causes uneven, minute wear on the mold's guiding components (such as guide pillars and bushings), increasing the clearance and causing slight wobble in the slider during high-speed operation. The combined effect of these two factors results in a double disturbance to the execution accuracy of the closed-loop control system: target position drift and physical instability of the execution platform.

[0005] Ultimately, the accumulation of these minute deviations leads to batch defects in the products (such as burrs and angular deviations), resulting in a decrease in yield. Since the root cause of the problem is multi-source, low-amplitude, and interacting deviations, rather than a single equipment failure, existing control methods that mainly target explicit anomalies cannot effectively identify and compensate for them. There is an urgent need for new solutions that can handle this complex and cumulative problem. Summary of the Invention

[0006] In view of the shortcomings of the prior art, this application provides a control method and system for multi-process composite stamping forming dies, which aims to solve the technical problem that in the actual production of multi-process composite stamping forming dies, due to the differences in the micro-characteristics of raw materials and the cumulative wear of die components, the existing control methods are difficult to effectively identify and compensate for the continuous, low-amplitude deviations caused by the wear of the raw materials, which ultimately leads to batch defects and a decrease in the yield rate of the products.

[0007] In a first aspect, a method for controlling a multi-stage composite stamping die, the method comprising the following steps:

[0008] S1: Synchronously acquire multi-source operating status data streams during the multi-process stamping forming process;

[0009] S2: Extract the correlation feature parameters reflecting the physical coordination relationship between the multi-source operating status data streams to construct a standard operating mode benchmark that characterizes the optimal operating condition;

[0010] S3: Extract the associated feature parameters of the current stamping cycle in real time and generate a real-time operation mode vector;

[0011] S4: Calculate the overall deviation of the real-time operation mode vector relative to the standard operation mode benchmark;

[0012] S5: Determine whether the overall deviation meets the preset continuous deviation condition, and identify the type of characteristic component that causes the deviation when it does.

[0013] S6: Based on the identified characteristic component type, adaptive compensation adjustment is performed on the motion control parameters or process load parameters of the stamping actuator.

[0014] Furthermore, step S1 includes:

[0015] S11: Obtain the real-time position information of the stamping actuator through a position detection element at a sampling frequency of not less than 10kHz;

[0016] S12: Instantaneous load data during the stamping process is acquired synchronously through pressure detection elements;

[0017] S13: Acquire the initial pose offset data of the sheet metal to be stamped through a non-contact displacement sensor;

[0018] S14: Use a high-speed data acquisition module to perform timestamp alignment processing on the real-time location information, the instantaneous load data, and the initial pose offset data to generate the multi-source operating status data stream.

[0019] Furthermore, in step S2, the associated characteristic parameters include at least one of the following: the time difference between the pressure peak and the displacement bottom dead center; the instantaneous slope ratio between the pressure signal and the position signal; and the oscillation characteristics of the position signal within a preset frequency band.

[0020] Furthermore, step S2 includes:

[0021] S21: Under the preset optimal operating conditions, record the multi-source operating status data stream for a preset number of consecutive cycles;

[0022] S22: Perform statistical analysis on the recorded multi-source operating status data streams and extract the correlation feature parameters that reflect the collaborative characteristics between the data streams;

[0023] S23: Construct a multi-dimensional standard pattern vector based on the associated feature parameters, which serves as the benchmark for the standard operating mode.

[0024] Furthermore, step S3 includes:

[0025] S31: Perform digital signal processing on the multi-source operating state data stream to extract the dynamic characteristics of the multi-source operating state data stream in the frequency domain or time-frequency domain;

[0026] S32: The digital signal processing includes Fast Fourier Transform or Wavelet Transform.

[0027] Furthermore, step S4 includes:

[0028] S41: Calculate the Euclidean distance or Mahalanobis distance between the real-time operation mode vector and the standard operation mode reference, and use the calculation result as the comprehensive deviation.

[0029] Furthermore, step S5 includes:

[0030] S51: When the overall deviation exceeds the warning threshold in a consecutive preset number of stamping cycles, it is determined that the continuous deviation condition is met;

[0031] S52: Identify the type of the feature component based on the deviation weight of each feature component in the real-time operation mode vector, wherein the type of the feature component includes material property deviation and mechanical clearance deviation.

[0032] Furthermore, step S6 includes:

[0033] S61: When the characteristic component type is the material property deviation, adjust the blank holder force in the process load parameters;

[0034] S62: When the characteristic component type is the mechanical clearance deviation, adjust the stamping depth or slider instantaneous speed curve in the motion control parameters.

[0035] Furthermore, step S61 includes:

[0036] S611: If the feature component in the real-time operation mode vector that reflects the time difference between the pressure peak and the displacement bottom dead center continues to increase, or the feature component that reflects the instantaneous slope ratio between the pressure signal and the position signal continues to decrease, then it is determined that the springback of the sheet metal increases.

[0037] S612: Based on the increase in the springback of the sheet metal, increase the blank holder force and fine-tune the stamping depth of the subsequent forming process;

[0038] Step S62 includes:

[0039] S621: If the characteristic component reflecting the oscillation amplitude of the position signal in the preset frequency band in the real-time operation mode vector continues to increase, or the characteristic component reflecting the energy density of a specific frequency component continues to increase, then it is determined that the stamping actuator is shaking.

[0040] S622: The instantaneous speed curve of the slider is changed by adjusting the drive current or valve opening, or a compensating torque opposite to the sway direction is applied when the stamping actuator descends to near the bottom dead center, in order to suppress or attenuate the sway.

[0041] Secondly, a multi-stage composite stamping die control system is provided, the system being used to implement the steps of any of the methods described above, the system comprising:

[0042] The data acquisition module is used to synchronously acquire multi-source operating status data streams during the multi-process stamping forming process;

[0043] The benchmark construction module is used to extract the correlation feature parameters reflecting the physical coordination relationship between the multi-source operating status data streams, so as to construct a standard operating mode benchmark that characterizes the optimal operating condition.

[0044] The vector generation module is used to extract the associated feature parameters of the current stamping cycle in real time and generate a real-time operation mode vector.

[0045] The deviation calculation module is used to calculate the overall deviation of the real-time operation mode vector relative to the standard operation mode benchmark;

[0046] The deviation identification module is used to determine whether the comprehensive deviation meets the preset continuous deviation condition, and to identify the type of feature component that causes the deviation when it does.

[0047] The adaptive compensation module is used to adaptively compensate and adjust the motion control parameters or process load parameters of the stamping actuator according to the identified feature component type.

[0048] Beneficial Effects: This application proposes a multi-process composite stamping die control method and system. By synchronously acquiring multi-source operational status data streams during the multi-process stamping process and extracting associated characteristic parameters reflecting physical coordination relationships to construct a standard operating mode benchmark, the method then calculates the deviation of the current stamping cycle in real time and identifies the deviation type. Finally, it adaptively compensates and adjusts the motion control parameters or process load parameters of the stamping actuator. This method effectively solves the problem in existing technologies where persistent, low-amplitude deviations caused by differences in the microscopic properties of raw materials and accumulated wear of die components are difficult to identify and compensate. By establishing a dynamic operating mode benchmark and real-time deviation analysis, this application can accurately capture subtle, cumulative deviations during the stamping process and perform targeted compensation based on the deviation type, thereby avoiding batch defects and significantly improving product yield and production stability. Attached Figure Description

[0049] Figure 1 This is a flowchart of a multi-process composite stamping die control method proposed in this application.

[0050] Figure 2 This is a structural diagram of a multi-process composite stamping die control system proposed in this application.

[0051] Figure 3 This is a schematic diagram of a multi-process composite stamping die control system proposed in this application.

[0052] Labeling Explanation: 201, Data Acquisition Module; 202, Benchmark Construction Module; 203, Vector Generation Module; 204, Deviation Calculation Module; 205, Deviation Identification Module; 206, Adaptive Compensation Module. Detailed Implementation

[0053] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and marked in the accompanying drawings can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0054] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0055] Traditional stamping processes face a dual dilemma. On the one hand, due to the continuous deviation of material springback characteristics, the forming target itself, which the control system needs to precisely control, drifts slightly. On the other hand, due to slide wobbling caused by wear of guide components, the operating platform itself, which performs precise control, becomes unstable. These two effects interact and superimpose, causing a cumulative decline in the forming quality of the final product. For example, slight burrs appear on the hole edges, minor deviations in bending angles, and a slight decrease in surface flatness. Initially, these problems are often insufficient to trigger traditional alarm mechanisms based on fixed thresholds set by a single sensor reading, because each deviation is very small. The control program continues to judge that the mold is operating normally, but in reality, the entire production process has quietly deviated from the optimal operating conditions, continuously producing a large number of substandard products. Only when problems are discovered by manual sampling or the scrap rate increases significantly is the machine forced to stop for investigation, resulting in huge economic losses and wasted time.

[0056] To resolve this issue, please refer to... Figure 1 This application proposes a method for controlling multi-stage composite stamping forming dies, the method comprising the following steps:

[0057] S1: Synchronously acquire multi-source operating status data streams during the multi-process stamping forming process;

[0058] S2: Extract the correlation feature parameters that reflect the physical coordination relationship between the multi-source operating status data streams in order to construct a standard operating mode benchmark that characterizes the optimal operating conditions;

[0059] S3: Extract the relevant feature parameters of the current stamping cycle in real time and generate a real-time operation mode vector;

[0060] S4: Calculate the overall deviation of the real-time operating mode vector from the standard operating mode baseline;

[0061] S5: Determine whether the overall deviation meets the preset continuous deviation conditions, and identify the type of characteristic component that causes the deviation if it does.

[0062] S6: Based on the identified characteristic component type, adaptive compensation and adjustment are performed on the motion control parameters or process load parameters of the stamping actuator.

[0063] This application constructs a standard operating mode benchmark by introducing the synchronous acquisition of multi-source operating status data streams and the extraction of associated feature parameters, and monitors deviations in the stamping process in real time. By identifying the types of deviations, adaptive compensation and adjustment of the motion control parameters or process load parameters of the stamping actuator are achieved, thereby effectively solving the shortcomings of traditional control methods in handling multi-source, low-amplitude, and cumulative deviations, and significantly improving product yield and production stability.

[0064] Specifically, in step S1, the multi-source operating status data stream can utilize various sensors installed on the stamping equipment to collect data in real time. Specifically, position sensors can acquire real-time position information of the stamping actuator; pressure sensors can acquire instantaneous load data during the stamping process; and displacement sensors can acquire initial pose offset data of the sheet metal to be stamped. These sensors need to have a sufficiently high sampling frequency to capture rapid dynamic changes during the stamping process.

[0065] In step S2, the associated feature parameters refer to key indicators extracted from the multi-source operating status data stream that reflect the physical coordination relationship between different data streams. These include, but are not limited to, the time difference between the pressure peak and the displacement bottom dead center, the instantaneous slope ratio between the pressure signal and the position signal, and the oscillation characteristics of the position signal within a preset frequency band. To construct a standard operating mode benchmark, multi-source operating status data streams for multiple consecutive stamping cycles can be recorded under optimal mold conditions. Then, statistical analysis is performed on these recorded data streams to extract the aforementioned associated feature parameters, and a multi-dimensional standard mode vector is constructed based on these parameters. This standard mode vector serves as the standard operating mode benchmark for subsequent deviation calculations.

[0066] In step S3, the real-time acquired multi-source operating status data stream can be rapidly processed using a digital signal processor (DSP) or a field-programmable gate array (FPGA). Specifically, the data stream can undergo preprocessing operations such as digital filtering and noise reduction, and then the associated characteristic parameters of the same type defined in step S2 can be calculated. For example, the specific parameter values ​​corresponding to the time difference between the pressure peak and the displacement bottom dead center of the current stamping cycle, the specific parameter values ​​corresponding to the instantaneous slope ratio between the pressure signal and the position signal, and the specific parameter values ​​corresponding to the oscillation characteristics of the position signal within a preset frequency band can be calculated in real time. Combining these real-time calculated associated characteristic parameters forms the real-time operating mode vector.

[0067] In step S4, the standard operating mode baseline refers to a reference model constructed by statistically analyzing the associated characteristic parameters under optimal die operating conditions. This baseline represents the ideal stamping process behavior and is used for subsequent deviation calculations.

[0068] The real-time operation mode vector refers to the vector composed of the associated feature parameters extracted in real time during the current stamping cycle. This vector reflects the actual operating status of the current stamping process.

[0069] The overall deviation refers to the degree of deviation of the real-time operating mode vector from the standard operating mode baseline. This deviation can quantify the difference between the current stamping process and the ideal state.

[0070] Specifically, Euclidean distance or Mahalanobis distance can be used to calculate the distance between the real-time operating mode vector and the standard operating mode baseline. Euclidean distance is an intuitive distance metric, suitable for situations where the feature components are independent. Mahalanobis distance, on the other hand, considers the correlation between feature components and can more accurately reflect the similarity of multi-dimensional data points. The calculated distance value is the overall deviation; the larger the value, the greater the deviation of the current stamping process from the optimal operating condition.

[0071] In step S5, the persistent deviation condition refers to the state where the overall deviation exceeds a preset threshold in multiple consecutive stamping cycles. This indicates a systematic and cumulative problem in the stamping process, rather than an occasional fluctuation. For example, it can be set that the persistent deviation condition is met when the overall deviation exceeds a warning threshold in a preset number of consecutive stamping cycles. Once the persistent deviation condition is met, it is necessary to further identify the type of characteristic component causing the deviation. This can be achieved by analyzing the deviation weights of each characteristic component in the real-time operation mode vector. For example, if the characteristic component reflecting the time difference between the pressure peak and the displacement bottom dead center deviates significantly, it may indicate a material property deviation; if the characteristic component reflecting the oscillation amplitude of the position signal within a preset frequency band deviates significantly, it may indicate a mechanical clearance deviation.

[0072] In step S6, the characteristic component type refers to the root cause of the overall deviation, such as material property deviation or mechanical clearance deviation.

[0073] A stamping actuator is a mechanical component responsible for performing stamping actions, such as a slider or a drive system.

[0074] Motion control parameters refer to the parameters that control the motion trajectory and speed of the stamping actuator, such as stamping depth and instantaneous speed curve of the slider.

[0075] Process load parameters refer to parameters that affect material deformation and stress during stamping, such as blank holder force.

[0076] Adaptive compensation adjustment refers to automatically adjusting motion control parameters or process load parameters based on the identified deviation type, so as to restore the stamping process to the optimal working condition.

[0077] When the characteristic component is identified as a material property deviation, the blank holder force in the process load parameters can be adjusted. If the material springback increases, the blank holder force can be appropriately increased to suppress springback. When the characteristic component is identified as a mechanical clearance deviation, the stamping depth or the instantaneous speed curve of the slide block in the motion control parameters can be adjusted. For example, if slide block wobbling exists, the instantaneous speed curve of the slide block can be changed by adjusting the drive current or valve opening, or a compensating torque opposite to the direction of wobbling can be applied when the stamping actuator descends near the bottom dead center to suppress or attenuate the wobbling.

[0078] The above method enables real-time identification and compensation of persistent springback deviations caused by microscopic differences in the characteristics of metal sheet batches, as well as the resulting changes in internal stress distribution in the mold and the subtle slide movement caused by progressive wear of guide components. This ensures high-precision molding quality and yield. Early warning and intervention before microscopic deviations accumulate and lead to macroscopic quality problems effectively prevent scrap, improve production efficiency and equipment utilization, and achieve refined management of mold operation.

[0079] Specifically, in order to achieve synchronous acquisition of multi-source operational status data streams, step S1 includes:

[0080] S11: Obtain the real-time position information of the stamping actuator through the position detection element at a sampling frequency of not less than 10kHz;

[0081] S12: Instantaneous load data during the stamping process is acquired synchronously through pressure detection elements;

[0082] S13: Acquire the initial pose offset data of the sheet metal to be stamped through a non-contact displacement sensor;

[0083] S14: Use the high-speed data acquisition module to perform timestamp alignment processing on real-time location information, instantaneous load data and initial pose offset data to generate a multi-source operating status data stream.

[0084] In stamping, the actuator typically refers to the slider that drives the upper die in reciprocating motion. High-precision position detection elements are used, such as optical encoders or magnetostrictive displacement sensors mounted on the side or top of the slider, with resolution down to the micrometer level. Using a high sampling frequency of at least 10 kHz is crucial because the subtle wobbling of the slider caused by wear on the guide components is essentially a high-frequency, low-amplitude vibration. Only a sufficiently high sampling frequency can accurately capture this dynamic detail, providing a reliable data foundation for subsequent oscillation characteristic analysis. If the sampling frequency is too low, this weak wobbling signal will be drowned out or distorted, making it impossible to effectively identify mechanical clearance deviations.

[0085] Pressure sensing elements, such as piezoelectric or strain gauge pressure sensors, are installed in critical stress areas of the mold, such as the support surface of the lower die or near the punch of the upper die. These sensors operate at the same sampling frequency as the position sensing elements, ensuring a precise temporal correspondence between the force and displacement signals. This synchronization is a prerequisite for analyzing their synergy; for example, calculating the time difference between the peak punching force and the slider reaching the bottom dead center, or analyzing the ratio of the pressure signal's rising slope to the position signal's falling slope, all depend on the precise alignment of the two data streams at every point in time.

[0086] At the initial positioning station where the sheet metal is fed into the die, a laser displacement sensor or machine vision sensor is installed to measure the minute positional and orientational offsets of the sheet metal edge or specific reference holes in a non-contact manner. The purpose of this step is to eliminate or compensate for errors introduced by inaccurate initial positioning of the sheet metal when analyzing deviations during the stamping process, so that subsequent pattern deviation analysis can focus more on problems caused by changes in material properties and die conditions.

[0087] Finally, a high-speed data acquisition module is used to perform timestamp alignment processing on real-time position information, instantaneous load data, and initial pose offset data to generate a multi-source operational status data stream. All analog or digital signals acquired by the sensors are fed into a high-speed data acquisition module based on a field-programmable gate array (FPGA) or digital signal processor (DSP). The core task of this module is to assign a high-precision timestamp to each data point from different channels, ensuring that all data streams are placed on a unified time reference. After timestamp alignment processing, these originally independent signal streams are merged into a unified, multi-dimensional time-series data set, i.e., a multi-source operational status data stream.

[0088] In constructing a standard operating mode benchmark representing optimal working conditions, the extracted correlation characteristic parameters are not isolated sensor readings, but rather dynamic indicators that profoundly reflect the inherent synergistic relationship between physical quantities such as force, displacement, and time during the stamping process. These parameters can depict the health status of the die from a higher dimension. Specifically, the correlation characteristic parameters may include at least one of the following: the time difference between the pressure peak and the displacement bottom dead center; the instantaneous slope ratio between the pressure signal and the position signal; and the oscillation characteristics of the position signal within a preset frequency band.

[0089] In the stamping process, the plastic deformation of the sheet metal reaches its maximum when the slide reaches its lowest point of travel, i.e., the bottom dead center. The moment when the stamping force reaches its peak is closely related to the material's deformation resistance, hardening behavior, and the geometry of the die cavity. For an ideal, stable sheet metal, under fixed process parameters, the time difference between the peak pressure and the slide reaching the bottom dead center should be a stable value. However, when a new batch of sheet metal experiences increased springback due to microstructural differences, it means the material has stronger elastic recovery capabilities, usually accompanied by higher deformation resistance. This causes the stamping force to reach its peak more quickly, resulting in a continuous, subtle change in the time difference between the peak pressure and the bottom dead center, such as a continuous increase. By monitoring this time difference, the drift in the material's springback characteristics can be indirectly detected.

[0090] At the instant the slider makes contact with the sheet metal and begins to apply pressure, the pressure signal rises rapidly, while the position signal continues to decline. At this moment, the time derivative of the pressure signal, i.e., the pressure rise slope, reflects the loading rate; the time derivative of the position signal, i.e., the position fall slope, reflects the instantaneous velocity of the slider. The ratio of these two values, in a physical sense, approximates the contact stiffness of the system at that instant. When the hardness or yield strength of the material changes slightly, its resistance to deformation also changes, directly resulting in different pressure rise rates at the same slider speed. For example, if the material hardens and the springback increases, the pressure signal rise slope will be relatively larger at the instant of contact, causing a continuous decrease in the instantaneous slope ratio. Therefore, this parameter can also serve as an effective feature for diagnosing material property deviations. This characteristic parameter reflects the equivalent dynamic stiffness of the material at the instant of contact during the stamping process. Specifically, starting from the slider's initial position when it contacts the sheet metal, five consecutive sampling points are taken to calculate the pressure change rate by calculating the derivative of the load signal with respect to time, and simultaneously, the instantaneous velocity of the slider is calculated by calculating the derivative of the position signal with respect to time. The instantaneous slope ratio is the absolute value of the pressure change rate divided by the instantaneous velocity of the slider.

[0091] The oscillation characteristics of position signals within a preset frequency band are indicators specifically used to diagnose the mechanical condition of molds. Typically, the guide components of precision molds have extremely small clearances, and the slider's trajectory is smooth and stable during high-speed movement. In the frequency domain, the position signal's energy is mainly concentrated in the low-frequency range related to the stamping frequency. As production progresses, guide components such as guide pillars and guide sleeves experience progressive wear, increasing the clearance. During high-speed descent, the slider may experience slight lateral swaying or vibration due to uneven force or inertial effects. This mechanical swaying is superimposed on the slider's main motion as high-frequency, low-amplitude disturbances. This is reflected in the data collected by the position sensor as abnormal energy peaks or oscillation amplitudes within a specific high-frequency band, such as 100 to 200 Hz. By performing spectral analysis on the position signal and extracting characteristics such as oscillation amplitude and energy density within these preset frequency bands, the degree of slider swaying can be quantified, thus enabling early warning of mechanical clearance deviations.

[0092] Furthermore, step S2 includes:

[0093] S21: Under the preset optimal operating conditions, record a multi-source operating status data stream for a preset number of consecutive cycles;

[0094] S22: Perform statistical analysis on the recorded multi-source operational status data streams and extract correlation feature parameters that reflect the collaborative characteristics between data streams;

[0095] S23: Construct a multi-dimensional standard pattern vector based on the associated feature parameters, which serves as the benchmark for the standard operating mode.

[0096] The preset optimal operating condition refers to a situation where the mold has just been installed and debugged, all components are in their best condition, and production is carried out using standard batches of metal sheets that have undergone rigorous inspection and whose performance fully meets design requirements. Simultaneously, the quality of the produced products has been precisely measured and confirmed as fully qualified, superior-grade products. Under these conditions, a data recording program is initiated to continuously record, for example, a thousand stamping cycles, all multi-source operational status data streams. Selecting to continuously record multiple cycles is to obtain statistically significant data, eliminate potential random fluctuations in individual cycles, and ensure that the collected data fully reflects the true state of the mold under stable operation.

[0097] The collected raw data is imported into the processing unit, where specialized algorithms perform in-depth data mining. For example, for each stamping cycle, time-domain features such as the time difference between the pressure peak and the bottom dead center, and the instantaneous slope ratio between the pressure signal and the position signal are calculated. Simultaneously, a Fast Fourier Transform is performed on the position signal to analyze its energy distribution across different frequency bands, extracting the average oscillation amplitude within specific high-frequency bands. Statistical analysis is then performed on all characteristic parameters calculated across these thousand cycles, determining the mean, standard deviation, and distribution range statistics for each characteristic parameter.

[0098] Finally, the average values ​​of all the feature parameters obtained from the above statistical analysis are combined into a multi-dimensional vector. For example, a standard pattern vector can be represented as:

[0099] ,in, This represents the average time difference between the pressure peak and the displacement bottom dead center in the standard mode vector. This represents the instantaneous slope ratio between the pressure signal and the position signal in the standard pattern vector; This represents the oscillation characteristics of the position signal within a preset frequency band in the standard mode vector. Mathematically, this vector defines a point in a high-dimensional feature space, representing the healthiest operating state of the mold. Simultaneously, the standard deviation of each feature parameter is recorded for normalization processing during subsequent deviation calculations. This set, containing the mean vector and standard deviation information, together constitutes the standard operating mode benchmark, becoming the gold standard for state comparison in all subsequent real-time production processes.

[0100] Furthermore, step S3 includes:

[0101] S31: Perform digital signal processing on the multi-source operating status data stream to extract the dynamic characteristics of the multi-source operating status data stream in the frequency domain or time-frequency domain;

[0102] S32: Digital signal processing includes Fast Fourier Transform or Wavelet Transform.

[0103] In practical implementations, digital signal processing techniques employed include Fast Fourier Transform (FFT) or Wavelet Transform. FFT is a highly effective tool for extracting oscillatory features from position signals. By performing FFT on a segment of position signal data within a stamping cycle, it can be transformed from the time domain to the frequency domain, yielding a spectrum. This spectrum clearly shows the energy distribution of the signal at different frequencies. By examining a preset, specific high-frequency band associated with mechanical sway, such as 100 to 200 Hz, the energy peak value within that band or its average energy density can be directly read, thereby quantifying the degree of slider sway.

[0104] In some more complex applications, slider swaying may not be continuous, but rather occur instantaneously at a specific stage of the stroke. For such non-stationary signals, wavelet transform offers more powerful analytical capabilities. Wavelet transform not only provides frequency information but also temporal localization information of frequency components, enabling analysis of the specific time point at which vibration of which frequency occurred. This is extremely helpful for more accurately diagnosing the cause and location of swaying. Through these efficient digital signal processing algorithms, all necessary associated characteristic parameters can be rapidly calculated within a very short time after the end of each stamping cycle and combined into the current real-time operating mode vector, preparing for subsequent deviation calculations.

[0105] Once the real-time operating mode vector representing the current stamping cycle state is obtained, the next crucial step is to objectively and accurately quantify its deviation from the standard operating mode baseline. This quantification is not simply a matter of checking each item for deviations, but rather calculating a comprehensive deviation that reflects the overall pattern differences. Specifically, step S4 includes:

[0106] S41: Calculate the Euclidean or Mahalanobis distance between the real-time operating mode vector and the standard operating mode baseline, and use the calculation result as the comprehensive deviation.

[0107] In a specific embodiment, it is assumed that both the standard mode vector and the real-time mode vector are three-dimensional, and their components are the time difference, slope ratio, and oscillation amplitude, respectively. Calculating the Euclidean distance is equivalent to calculating the straight-line distance between two points corresponding to these two vectors in three-dimensional space. The calculation formula can be expressed as:

[0108] ,in, The Euclidean distance between the two vectors; This is the time difference between the pressure peak and the displacement bottom dead center in the real-time mode vector. is the time difference between the pressure peak and the displacement bottom dead center in the standard mode vector; is the instantaneous slope ratio between the pressure signal and the position signal in the real-time mode vector; This represents the instantaneous slope ratio between the pressure signal and the position signal in the standard pattern vector; This refers to the oscillation characteristics of the position signal in the real-time mode vector within a preset frequency band. The oscillation characteristics of the position signal in the quasi-mode vector within a preset frequency band are used. Euclidean distance intuitively reflects the geometric differences between the two vectors and is simple and fast to calculate.

[0109] However, in some cases, the dimensions and fluctuation ranges of different feature components can vary significantly, and there may be correlations between the components. For example, the unit of time difference is milliseconds, while the unit of oscillation amplitude is micrometers. Directly using Euclidean distance may result in feature components with large fluctuation ranges receiving excessive weight in the distance calculation. To address this issue, Mahalanobis distance can be used. Mahalanobis distance considers the covariance between each feature component during calculation, and is a distance metric that eliminates the influence of dimensions and correlations. By calculating Mahalanobis distance, the statistical significance of real-time model vectors deviating from the standard model baseline can be assessed more accurately.

[0110] The introduction of Mahalanobis distance involves linearly transforming and scaling the feature space using the covariance matrix from the standard operating mode benchmark. Specifically, the difference vector between the real-time operating mode vector and the standard operating mode vector is first extracted, and then this difference is used to perform a quadratic operation with the inverse covariance matrix obtained from the standard benchmark during the statistical phase. This approach is equivalent to measuring deviations within a space that considers the fluctuation weights of each feature component, automatically suppressing noisy but low-information feature components and highlighting subtle mode shifts that play a crucial early warning role in physical coordination.

[0111] The calculated overall deviation is merely an immediate warning signal. More importantly, it's crucial to determine whether this deviation is persistent and to uncover its underlying causes, thereby avoiding overreactions to occasional, harmless system fluctuations. Therefore, step S5 further includes:

[0112] S51: When the overall deviation exceeds the warning threshold in a consecutive preset number of stamping cycles, the continuous deviation condition is determined to be met.

[0113] S52: Identify the type of feature component based on the deviation weight of each feature component in the real-time operation mode vector. The feature component types include material property deviation and mechanical clearance deviation.

[0114] First, a rule can be set: a deviation warning is only triggered when the overall deviation D exceeds a preset warning threshold T_warning for 100 consecutive stamping cycles. This continuity judgment logic is key to effectively filtering out random factors such as occasional burrs, electromagnetic interference, or single-batch material defects that occur during production. Only when the deviation exists stably and continuously does it indicate that it is likely a true state drift caused by systemic factors, such as changes in material batch characteristics or gradual wear of the die.

[0115] Once the persistent deviation condition is confirmed, the next step is to analyze the root cause of the deviation in depth. Specifically, based on the deviation weights of each characteristic component in the real-time operation mode vector, the types of characteristic components causing the deviation are identified. These characteristic component types are mainly divided into two categories: material property deviation and mechanical clearance deviation. The identification method is to analyze the contribution of each component constituting the overall deviation. For example, when calculating the overall deviation, the degree to which each characteristic component deviates from its standard value can be calculated separately, for example... ,in, These are the feature components in the real-time running mode vector; These are the characteristic components in the standard mode vector. By comparing the deviation contributions of each component, it's possible to determine which feature(s) dominate the overall mode deviation. If the time difference between the pressure peak and the displacement bottom dead center, and the instantaneous slope ratio of the pressure and position signals have the largest deviation contributions, then the deviation type can be identified as a material property deviation. Conversely, if the oscillation amplitude of the position signal within a preset frequency band has the largest deviation contribution, then the deviation type can be identified as a mechanical clearance deviation. This diagnostic mechanism allows subsequent compensation adjustments to be targeted and precise.

[0116] Once the root cause of the deviation is identified, targeted adaptive compensation adjustments can be implemented. This is a precise, closed-loop correction process designed to guide the mold's operating state back to optimal conditions. Further, step S6 includes:

[0117] S61: When the characteristic component type is material property deviation, adjust the blank holder force in the process load parameters;

[0118] S62: When the characteristic component type is mechanical backlash deviation, adjust the stamping depth or instantaneous speed curve of the slider in the motion control parameters.

[0119] This logic of differentiation and adjustment is based on a deep understanding of the physical process of stamping. The core of material property deviation lies in the change in the springback behavior of the sheet metal. In stamping, especially in the deep drawing process, the blank holder force is a key process load parameter for controlling sheet metal flow, preventing wrinkling, and regulating springback. Therefore, by adjusting the magnitude of the blank holder force, problems caused by changes in the material's inherent properties can be directly intervened in and compensated for.

[0120] Mechanical backlash deviation stems from instability in the motion trajectory of the stamping actuator. The stamping depth and instantaneous slide velocity curves are control parameters that directly define the actuator's motion behavior. By fine-tuning these motion control parameters, the negative impacts caused by mechanical sway can be compensated for or suppressed. This differentiated adjustment strategy based on diagnostic results ensures the targeted effectiveness of the compensation measures.

[0121] To make the compensation adjustment more precise and intelligent, this method further refines the specific adjustment logic under different deviation types. Specifically, step S61 includes:

[0122] S611: If the feature component in the real-time operation mode vector that reflects the time difference between the pressure peak and the displacement bottom dead center continues to increase, or the feature component that reflects the instantaneous slope ratio between the pressure signal and the position signal continues to decrease, then it is determined that the springback of the sheet metal has increased.

[0123] S612: Increase the blank holder force according to the increase in the springback of the sheet metal, and fine-tune the stamping depth of the subsequent forming process.

[0124] The adjustment range here is not constant, but rather related to the degree of deviation. For example, a mapping model can be established to map the overall deviation or the deviation value of a specific characteristic component to the percentage increase in blank holder force and the fine adjustment of stamping depth. This quantitative adjustment can avoid overcompensation and form a closed-loop feedback control by continuously monitoring the adjusted operating mode vector until the mode deviation returns to the normal range.

[0125] As a specific implementation method, the mapping relationship model can be established by training a regression model based on historical data. First, during the mold debugging and initial production stages, different degrees of material property deviation or mechanical clearance deviation are intentionally introduced, and the multi-source operating status data streams and corresponding comprehensive deviation or specific feature component deviation values ​​are recorded. Simultaneously, the percentage increase in blank holder force and the fine-tuning amount of stamping depth applied to restore the mold's operating state to its optimal condition are recorded. This data constitutes the training dataset. Then, a regression algorithm, such as support vector regression or multinomial regression, is selected as the mapping relationship model. The comprehensive deviation or the deviation value of specific feature components is used as the model's input features, and the percentage increase in blank holder force and the fine-tuning amount of stamping depth are used as the model's output targets. The regression model is trained using the training dataset to optimize the model's parameters, enabling it to accurately learn the mapping relationship from deviation values ​​to compensation amounts.

[0126] When dealing with mechanical clearance deviations, it is also necessary to confirm the existence of wobble based on the changes in characteristic components. Step S62 includes:

[0127] S621: If the characteristic component reflecting the oscillation amplitude of the position signal in the preset frequency band in the real-time operation mode vector continues to increase, or the characteristic component reflecting the energy density of a specific frequency component continues to increase, then it is determined that the stamping actuator is shaking.

[0128] S622: By adjusting the drive current or valve opening to change the instantaneous speed curve of the slider, or by applying a compensating torque opposite to the direction of swaying when the stamping actuator is descending and approaching the bottom dead center, swaying can be suppressed or attenuated.

[0129] If the stamping actuator exhibits wobbling, various compensation strategies can be employed to address this wobbling. One strategy involves adjusting the drive current or valve opening to alter the instantaneous speed curve of the slider. For instance, by analyzing which stage of the stroke the wobbling primarily occurs in, the instantaneous speed or acceleration of the slider can be appropriately reduced at that stage to decrease dynamic impact and inertial forces, thereby passively suppressing the wobbling.

[0130] In another, more proactive embodiment, particularly on presses directly driven by servo motors, a compensating torque opposite to the direction of sway can be applied as the press actuator descends near the bottom dead center to actively suppress or attenuate sway. Specifically, the instantaneous frequency and phase of the sway are extracted by real-time high-frequency filtering and analysis of the position signal. Then, the servo motor driver is controlled to simultaneously output the torque required for the main motion and superimpose a small compensating torque opposite to the sway. This compensating torque acts like an invisible hand, counteracting the disturbance caused by the sway in real time, thus achieving active damping control of the sway. This model-based and real-time feedback-driven proactive compensation strategy can more efficiently improve the dynamic stability of the die.

[0131] The compensation adjustment employs a gradual closed-loop strategy based on incremental updates. After identifying the type of characteristic component, the controller does not directly jump to the target parameter. Instead, it queries a preset sensitivity mapping table based on the magnitude of the overall deviation to obtain an initial compensation increment. In subsequent stamping cycles, the changing trend of the associated characteristic parameters is continuously tracked. If the overall deviation begins to decrease, the current compensation slope is maintained and adjustments continue; if the rate of decrease in deviation is slower than expected or a rebound occurs, the proportional-integral-differential algorithm is used to correct the compensation increment in real time. This logic ensures a smooth transition in the adjustment process when dealing with low-amplitude, long-cycle deviations such as differences in the microstructure of metal sheets, avoiding severe mechanical vibrations in the mold or sheet cracking caused by a single over-adjustment.

[0132] Please refer to Figure 2 , Figure 3 This application also provides a multi-stage composite stamping die control system, which is used to implement the steps of any of the above methods, and the system includes:

[0133] Data acquisition module 201 is used to synchronously acquire multi-source operating status data streams during the multi-process stamping forming process;

[0134] The benchmark construction module 202 is used to extract the correlation feature parameters reflecting the physical coordination relationship between the multi-source operating status data streams in order to construct a standard operating mode benchmark that characterizes the optimal operating condition.

[0135] The vector generation module 203 is used to extract the associated feature parameters of the current stamping cycle in real time and generate a real-time operation mode vector.

[0136] Deviation calculation module 204 is used to calculate the comprehensive deviation of the real-time operation mode vector from the standard operation mode baseline;

[0137] The deviation identification module 205 is used to determine whether the comprehensive deviation meets the preset continuous deviation conditions, and to identify the type of feature component that causes the deviation when the conditions are met.

[0138] The adaptive compensation module 206 is used to adaptively compensate and adjust the motion control parameters or process load parameters of the stamping actuator according to the identified feature component type.

[0139] Specifically, the data acquisition module 201 can integrate multiple sensors, such as position detection elements, pressure detection elements, and non-contact displacement sensors, to acquire real-time position information of the stamping actuator, instantaneous load data during the stamping process, and initial pose offset data of the sheet metal to be stamped. To ensure data time synchronization and high accuracy, the data acquisition module 201 typically employs a high-speed data acquisition module and performs timestamp alignment processing on the acquired multi-source data to generate a multi-source operating status data stream for subsequent analysis.

[0140] The benchmark construction module 202 extracts correlation feature parameters that characterize the physical coordination relationship between data streams by statistically analyzing continuous multi-cycle, multi-source operating status data streams recorded under preset optimal operating conditions. These correlation feature parameters may include the time difference between the pressure peak and the displacement bottom dead center, the instantaneous slope ratio between the pressure signal and the position signal, and the oscillation characteristics of the position signal within a preset frequency band. Based on these parameters, the benchmark construction module can construct a multi-dimensional standard pattern vector as a standard operating mode benchmark for subsequent real-time monitoring and deviation judgment.

[0141] In practical applications, the vector generation module 203 performs digital signal processing on the multi-source operating status data stream collected in the current stamping cycle. For example, it extracts dynamic features in the frequency domain or time-frequency domain through methods such as fast Fourier transform or wavelet transform, and extracts real-time feature values ​​corresponding to the associated feature parameters from them, thereby generating a real-time operating mode vector that represents the current operating status.

[0142] Furthermore, the deviation calculation module 204 calculates the distance between the real-time operating mode vector and the standard operating mode reference, such as Euclidean distance or Mahalanobis distance, and uses the calculation result as the comprehensive deviation. This comprehensive deviation intuitively reflects the degree of deviation between the current stamping cycle's operating mode and the ideal mode.

[0143] Furthermore, the deviation identification module 205 continuously monitors the overall deviation. When it exceeds the warning threshold for a preset number of consecutive stamping cycles, it is determined that the continuous deviation condition is met. Based on this, the deviation identification module identifies the type of feature component causing the deviation, such as material property deviation or mechanical clearance deviation, according to the deviation weight of each feature component in the real-time operation mode vector.

[0144] Finally, the adaptive compensation module 206 performs precise compensation adjustments based on the identified causes of deviation. When the characteristic component type is identified as material property deviation, the module adjusts the blank holder force in the process load parameters, for example, by increasing the blank holder force to compensate for the increase in sheet springback. When the characteristic component type is identified as mechanical clearance deviation, the module adjusts the stamping depth or slider instantaneous speed curve in the motion control parameters, for example, by adjusting the drive current or valve opening to change the slider instantaneous speed curve, or by applying a compensation torque to suppress the wobbling of the stamping actuator.

[0145] This application's system visualizes each logical step in the multi-process composite stamping die control method as a functional module, thereby achieving real-time, closed-loop, and adaptive control of the complex stamping process. The data acquisition module 201, acting as the system's perception layer, ensures comprehensive and high-precision capture of the stamping process state, providing a reliable data foundation for subsequent analysis. The benchmark construction module 202, through offline learning and modeling, sets ideal operating standards for the system, enabling it to perform state assessments based on evidence. The vector generation module 203 and the deviation calculation module 204 work together to form the core of the system's real-time monitoring and diagnosis, quickly identifying the difference between the current operating condition and the ideal operating condition. More importantly, the deviation identification module 205 can further analyze and distinguish the specific physical causes of the deviation, such as changes in material properties or mechanical structural problems, providing a basis for subsequent precise compensation decisions. Ultimately, the adaptive compensation module 206, as the system's execution layer, can adjust the motion control parameters or process load parameters of the stamping actuator in a targeted manner according to the specific problems identified, thereby redirecting the stamping process to the optimal working condition and effectively solving the problem that traditional methods are difficult to respond in real time and intervene precisely in practical applications.

[0146] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for controlling a multi-stage composite stamping die, characterized in that, The method includes the following steps: S1: Synchronously acquire multi-source operating status data streams during the multi-process stamping forming process; S2: Extract the correlation feature parameters reflecting the physical coordination relationship between the multi-source operating status data streams to construct a standard operating mode benchmark that characterizes the optimal operating condition; S3: Extract the associated feature parameters of the current stamping cycle in real time and generate a real-time operation mode vector; S4: Calculate the overall deviation of the real-time operation mode vector relative to the standard operation mode benchmark; S5: Determine whether the overall deviation meets the preset continuous deviation condition, and identify the type of characteristic component that causes the deviation when it does. The specific steps include: S51: When the overall deviation exceeds the warning threshold in a consecutive preset number of stamping cycles, it is determined that the continuous deviation condition is met; S52: Identify the type of the feature component based on the deviation weight of each feature component in the real-time operation mode vector, wherein the type of the feature component includes material property deviation and mechanical clearance deviation; S6: Based on the identified characteristic component type, adaptive compensation adjustment is performed on the motion control parameters or process load parameters of the stamping actuator; specific steps include: S61: When the characteristic component type is the material property deviation, adjust the blank holder force in the process load parameters; the specific steps include: S611: If the characteristic component in the real-time operation mode vector reflecting the time difference between the pressure peak and the displacement bottom dead center continues to increase, or the characteristic component reflecting the instantaneous slope ratio between the pressure signal and the position signal continues to decrease, then it is determined that the sheet springback amount has increased; S612: According to the degree of increase in the sheet springback amount, increase the blank holder force and fine-tune the stamping depth of the subsequent forming process; S62: When the characteristic component type is the mechanical backlash deviation, adjust the stamping depth or slider instantaneous speed curve in the motion control parameters; the specific steps include, S621: if the characteristic component reflecting the oscillation amplitude of the position signal in the preset frequency band in the real-time operation mode vector continues to increase, or the characteristic component reflecting the energy density of a specific frequency component continues to increase, then it is determined that the stamping actuator has swaying; S622: by adjusting the drive current or valve opening to change the slider instantaneous speed curve, or by applying a compensating torque opposite to the swaying direction when the stamping actuator descends and approaches the bottom dead center, to suppress or attenuate the swaying.

2. The method for controlling a multi-process composite stamping die according to claim 1, characterized in that, Step S1 includes: S11: Obtain the real-time position information of the stamping actuator through a position detection element at a sampling frequency of not less than 10kHz; S12: Instantaneous load data during the stamping process is acquired synchronously through pressure detection elements; S13: Acquire the initial pose offset data of the sheet metal to be stamped through a non-contact displacement sensor; S14: Use a high-speed data acquisition module to perform timestamp alignment processing on the real-time location information, the instantaneous load data, and the initial pose offset data to generate the multi-source operating status data stream.

3. The method for controlling a multi-process composite stamping die according to claim 1, characterized in that, In step S2, the associated characteristic parameters include at least one of the following: the time difference between the pressure peak and the displacement bottom dead center; the instantaneous slope ratio between the pressure signal and the position signal; and the oscillation characteristics of the position signal within a preset frequency band.

4. The method for controlling a multi-process composite stamping die according to claim 3, characterized in that, Step S2 includes: S21: Under the preset optimal operating conditions, record the multi-source operating status data stream for a preset number of consecutive cycles; S22: Perform statistical analysis on the recorded multi-source operating status data streams and extract the correlation feature parameters that reflect the collaborative characteristics between the data streams; S23: Construct a multi-dimensional standard pattern vector based on the associated feature parameters, which serves as the benchmark for the standard operating mode.

5. The method for controlling a multi-process composite stamping die according to claim 1, characterized in that, Step S3 includes: S31: Perform digital signal processing on the multi-source operating status data stream to extract the dynamic features of the multi-source operating status data stream in the frequency domain or time-frequency domain; the digital signal processing includes fast Fourier transform or wavelet transform.

6. The method for controlling a multi-process composite stamping die according to claim 1, characterized in that, Step S4 includes: S41: Calculate the Euclidean distance or Mahalanobis distance between the real-time operation mode vector and the standard operation mode reference, and use the calculation result as the comprehensive deviation.

7. A multi-process composite stamping forming die control system, characterized in that, The system is used to implement the steps of the method according to any one of claims 1-6, and the system includes: The data acquisition module is used to synchronously acquire multi-source operating status data streams during the multi-process stamping forming process; The benchmark construction module is used to extract the correlation feature parameters reflecting the physical coordination relationship between the multi-source operating status data streams, so as to construct a standard operating mode benchmark that characterizes the optimal operating condition. The vector generation module is used to extract the associated feature parameters of the current stamping cycle in real time and generate a real-time operation mode vector. The deviation calculation module is used to calculate the overall deviation of the real-time operation mode vector relative to the standard operation mode benchmark; The deviation identification module is used to determine whether the comprehensive deviation meets the preset continuous deviation condition, and to identify the type of feature component that causes the deviation when it does. The adaptive compensation module is used to adaptively compensate and adjust the motion control parameters or process load parameters of the stamping actuator according to the identified feature component type.

Images