A method for automatic compensation and closed-loop control of forming error of precision stamping parts

Abstract

The present application provides a kind of precision stamping part forming error automatic compensation and closed loop control method, it is related to stamping automation control technical field.The method includes: control equipment continuous stamping, reads stamping part appearance size deviation parameter and executes closed loop compensation to feeding mechanism.Multiple back stamping cycle is divided into monitoring time window, and the alternative frequency that the difference value direction of deviation parameter occurs positive and negative jump is counted;The frequency is compared with threshold value, and decoupling system is in high-frequency oscillation or same direction accumulation state;If it is high-frequency oscillation state, then block step distance adjustment instruction and send pressure delay instruction to absorb microcosmic material stress;If it is same direction accumulation state, then drift slope is fitted, utilizes dynamic damping coefficient to implement amplitude modulation and executes closed loop compensation accordingly.The present application realizes the classification decoupling control of compound error, and improves the long-term consistency of forming size.

Description

Technical Field

[0001] This invention relates to the field of stamping automation control technology, and in particular to a method for automatic compensation and closed-loop control of forming errors in precision stamped parts. Background Technology

[0002] Precision metal stamping manufacturing typically employs multi-station progressive dies in conjunction with a servo feeding system for mass production. To ensure that the dimensions of the formed products remain stable within a given tolerance range, vision measurement equipment is commonly deployed on the production line to acquire dimensional deviation data of the stamped parts after demolding. After receiving this feedback data, the control system calculates the reverse feed step adjustment and sends it to the feeding actuator, thereby achieving closed-loop correction of dimensional deviations. This feedback-based online compensation mechanism can address some of the process errors that occur during manufacturing.

[0003] Under long-term continuous stamping conditions, the physical causes of dimensional deviations in stamped parts exhibit a diverse and intertwined nature. The stamping die endures prolonged mechanical loads and frictional heat, leading to thermal expansion accumulation in its internal guiding components and gradual, yielding physical wear on the cutting edge. These factors cause a continuous, cumulative drift in the dimensions of the stamped part along a single dimension. Simultaneously, at the moment the metal strip completes plastic deformation and escapes the die's constraint, the residual uneven stress within the material's microcrystalline lattice is released in a concentrated manner, causing short-term elastic micro-vibrations in the strip, resulting in high-frequency fluctuations in the measured dimensions.

[0004] Current closed-loop control strategies for dimensions tend to directly convert transiently acquired dimensional offset values ​​into step adjustment commands. When faced with high-frequency elastic jumps caused by the release of internal material stress, if the controller immediately outputs a reverse correction displacement to the servo mechanism, the compensation excitation applied by the mechanical side is prone to superimposed at the same frequency with the physical vibrations of the material itself, potentially inducing instability in the feeding position control loop. When dealing with unidirectional cumulative displacement caused by mold heating or wear, conventional step adjustment modes using fixed gain parameters are difficult to adapt to the subtle dynamic fluctuations in the hardness of different batches of raw materials. When material hardness changes, the original compensation span often causes overshoot in the feed adjustment or a slower convergence speed. The industry needs to find an automatic compensation control scheme that can accommodate multiple complex error causes. Summary of the Invention

[0005] The purpose of this invention is to provide an automatic compensation and closed-loop control method for forming errors of precision stamped parts, so as to solve the technical problems of existing stamping closed-loop feedback mechanisms, which lack decoupling from the physical causes of errors, leading to instability of the feeding position control loop due to direct step compensation based on transient high-frequency elastic fluctuations, and the use of fixed parameters to deal with unidirectional cumulative deviations, which easily leads to feed adjustment overshoot.

[0006] This invention provides an automatic compensation and closed-loop control method for forming errors of precision stamped parts, comprising the following steps:

[0007] The stamping equipment is controlled to deliver the strip material into the mold according to the preset step distance to perform multi-station continuous stamping operation;

[0008] At the re-upsetting station after stamping is completed, the dimensional deviation parameters of the stamped part are read.

[0009] The feeding mechanism performs closed-loop compensation based on the aforementioned dimensional deviation parameters.

[0010] The method is characterized by further comprising the following control process:

[0011] The continuously set multiple stamping cycles are divided into monitoring time windows, and multiple sets of the external dimension deviation parameters within the monitoring time windows are collected.

[0012] Calculate the direction of the difference between the dimensional deviation parameters corresponding to two adjacent stamping cycles, and count the frequency of alternation between positive and negative jumps in the direction of the difference within the monitoring time window;

[0013] The alternation frequency is compared with a preset state determination threshold to decouple and determine whether the current mold system is in a high-frequency oscillation state or a unidirectional accumulation state.

[0014] When the decoupling determines that the current mold system is in the high-frequency oscillation state, the step adjustment command issued to the feeding mechanism is blocked, and a pressure holding delay command is sent to the re-upsetting station to absorb micro-material stress by extending the dwell time of the punch at the bottom dead center position.

[0015] When the decoupling determines that the current mold system is in the same direction accumulation state, a drift slope is generated by fitting multiple sets of external dimension deviation parameters, a pre-configured dynamic damping coefficient is extracted to perform a reduction modulation operation on the drift slope, and the closed-loop compensation action is performed on the feeding mechanism based on the compensation amount after the reduction modulation operation.

[0016] Optionally, the controlled stamping equipment conveys the strip material into the mold in a step-by-step manner according to a preset step distance to perform multi-station continuous stamping operations, including:

[0017] The strip has micro-tube wall characteristics, and for the strip, a multi-step micro-deformation bending operation is carried out using an integrated rolling forming structure;

[0018] Obtain the pre-calculated springback compensation angle and import the springback compensation angle into the mold control system as a correction reference;

[0019] Using the aforementioned correction reference combined with a high-precision guide sleeve assembly and a floating stripping assembly, a high-pressure shaping process is applied to the formed hollow outer wall after the multi-step micro-deformation bending operation to ensure that the pipe wall's roundness and coaxiality parameters meet the requirements.

[0020] Optionally, the controlled stamping equipment, which feeds the strip material into the mold in a step-by-step manner according to a preset step distance to perform multi-station continuous stamping operations, further includes:

[0021] Stamping is performed using a pre-precision oil-cutting and optical grinding blade;

[0022] A clearance adjustment block is embedded inside the blanking die cavity;

[0023] The gap adjustment block is used to perform dynamic leveling operation on the punching blade to reduce the gap offset error generated during the punching process.

[0024] Optionally, the re-upsetting station after stamping completion, which reads the dimensional deviation parameters of the stamped part, includes:

[0025] The surface contour image of the completed stamping part is captured by a high-speed vision detection module.

[0026] Feature point matching and edge fitting are performed on the captured surface contour image;

[0027] Extract the height deviation, flatness deviation, and terminal spacing deviation values ​​from the data obtained after fitting.

[0028] The height deviation value, the flatness deviation value, and the terminal spacing deviation value are collectively defined as the external dimensional deviation parameter.

[0029] Optionally, the step of calculating the direction of the difference between the dimensional deviation parameters corresponding to two adjacent stamping cycles, and counting the frequency of alternations of the direction of the difference in positive and negative changes within the monitoring time window, includes:

[0030] All dimensional deviation parameters within the monitoring time window are arranged in chronological order.

[0031] Extract the mathematical sign of the difference between the back row parameters and the front row parameters as the direction of the difference;

[0032] The total number of times the difference direction changes from positive to negative and from negative to positive is accumulated, and the total number of occurrences is denoted as the alternation frequency.

[0033] Optionally, comparing the alternation frequency with a pre-set state determination threshold to decouple and determine whether the current mold system is in a high-frequency oscillation state or a unidirectional accumulation state includes:

[0034] When the alternation frequency is greater than the state determination threshold, the material is confirmed to be in the random elastic fluctuation stage, and the high-frequency oscillation state result is output.

[0035] When the alternation frequency is less than or equal to the state determination threshold, and the absolute difference between the first and last parameters within the monitoring time window crosses the safety tolerance limit, continuous thermal expansion or mechanical wear is confirmed, and the same-direction cumulative state result is output.

[0036] Optionally, sending a pressure-holding delay command to the re-upsetting station, utilizing the extended dwell time of the punch at the bottom dead center to absorb microscopic material stress, includes:

[0037] Extract the maximum deviation from the extreme peak value under the high-frequency oscillation state;

[0038] Based on a pre-set library of material plastic rheology curves, the time required for stress release that matches the maximum deviation from the extreme peak value is calculated in reverse.

[0039] The time required for stress release is encapsulated in the pressure holding delay command and sent to the pressure drive servo unit to keep the punch in the lowest shaping position.

[0040] Optionally, the step of generating a drift slope based on fitting multiple sets of the aforementioned dimensional deviation parameters, and extracting a pre-configured dynamic damping coefficient to perform amplitude reduction modulation on the drift slope, includes:

[0041] A linear regression operation is performed on the multiple sets of dimensional deviation parameters to obtain the drift slope characterizing the unidirectional offset rate;

[0042] Read the dynamic damping coefficient from the system register, the dynamic damping coefficient having a floating-point value between 0 and 2;

[0043] The attenuation bias is obtained by multiplying the newly measured dimensional deviation parameter by the dynamic damping coefficient.

[0044] The attenuation bias and the expected development variable corresponding to the drift slope are summed to generate the compensation amount after the amplitude reduction modulation operation.

[0045] Optionally, the dynamic damping coefficient is an adjustable control parameter, and the closed-loop compensation action further includes:

[0046] Periodically obtain the hardness fluctuation ratio of newly cut strip batches;

[0047] When the hardness fluctuation ratio shows an increasing trend, the value of the dynamic damping coefficient is reduced according to the preset adjustment span.

[0048] When the hardness fluctuation ratio shows a decreasing trend and tends to level off, the value corresponding to the dynamic damping coefficient is amplified according to the preset adjustment span.

[0049] Optionally, after performing the closed-loop compensation action on the feeding mechanism based on the compensation amount after the amplitude reduction modulation operation, the method further includes:

[0050] Within the immediate subsequent monitoring time window, the latest dimensional deviation parameters of the newly stamped part are reread;

[0051] Verify whether the latest dimensional deviation parameter converges and falls within the set acceptable reference range;

[0052] If the mold fails to fall into the qualified benchmark range, it is determined that the mold has suffered irreversible physical wear and damage, triggering the emergency stop control logic for the entire line and illuminating the audible and visual alarm device.

[0053] The present invention has achieved the following beneficial effects:

[0054] This invention, by systematically monitoring the alternating frequency of positive and negative jumps in the dimensional deviation parameters within a time window, decouples the current stamping die system from a high-frequency oscillation state caused by the release of micro-stress in the material strip to a state of unidirectional accumulation caused by thermal expansion and mechanical wear. This provides a control basis for subsequent differentiated intervention. When the system is determined to be in a high-frequency oscillation state, this method actively blocks the step adjustment command to the feeding mechanism and sends a pressure holding delay command to the re-upsetting station to extend the dwell time of the punch at the bottom dead center. It relies on physical ballast action to absorb the residual lattice stress inside the material strip, avoiding the divergence of the servo position control loop caused by direct feedback of transient error commands. When the system is determined to be in a unidirectional accumulation state, this method uses the collected deviation parameters to fit and generate a drift slope, extracts the dynamic damping coefficient to perform a reduction modulation operation on the step compensation amount, and adaptively updates the damping coefficient based on the hardness fluctuation ratio of the new batch of material strip. This dynamic closed-loop control strategy based on physical state classification alleviates the overshoot adjustment that is prone to occur when using fixed compensation parameters to deal with different batches of materials. It makes the correction action of the servo feeding module smoother and helps to maintain the dimensional control accuracy and overall system stability during long-cycle operation of multi-station stamping production lines.

[0055] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0056] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0057] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0058] Figure 1 This is the main flowchart of the automatic compensation and closed-loop control method for forming errors of precision stamping parts provided in the embodiments of the present invention;

[0059] Figure 2 This is a flowchart of a method for reading dimensional deviation parameters provided in an embodiment of the present invention;

[0060] Figure 3 A flowchart illustrating the method for determining the direction and alternation frequency of statistical dimensional deviation parameters provided in this embodiment of the invention;

[0061] Figure 4 This is a flowchart of a method for sending a pressure holding delay command in a high-frequency oscillation state, provided by an embodiment of the present invention.

[0062] Figure 5 This is a flowchart of a method for generating compensation amounts based on drift slope and dynamic damping coefficient, provided in an embodiment of the present invention.

[0063] Figure 6 The flowchart illustrates the adaptive adjustment and safety verification closed-loop of the dynamic damping coefficient provided in this embodiment of the invention. Detailed Implementation

[0064] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0065] This embodiment provides an automatic compensation and closed-loop control method for forming errors of precision stamped parts, such as... Figure 1 As shown, the execution steps are as follows:

[0066] Step S10: Control the stamping equipment to feed the strip material into the mold according to the preset step distance to perform multi-station continuous stamping operation;

[0067] Step S20: At the re-upsetting station after stamping is completed, read the dimensional deviation parameters of the stamped part.

[0068] Step S30: Perform closed-loop compensation on the feeding mechanism based on the external dimensional deviation parameters.

[0069] Furthermore, the method also includes the following control process:

[0070] Step S40: Divide the continuously set multiple stamping cycles into monitoring time windows, and collect multiple sets of the external dimension deviation parameters within the monitoring time windows;

[0071] Step S50: Calculate the direction of the difference between the external dimensional deviation parameters corresponding to two adjacent stamping cycles, and count the frequency of alternation of the positive and negative jumps of the difference direction within the monitoring time window;

[0072] Step S60: Compare the alternation frequency with a preset state determination threshold, and decouple to determine whether the current mold system is in a high-frequency oscillation state or a unidirectional accumulation state;

[0073] Step S70: When the decoupling shows that the current mold system is in the high-frequency oscillation state, block the step adjustment command issued to the feeding mechanism and send a pressure holding delay command to the re-upsetting station to absorb micro-material stress by extending the dwell time of the punch at the bottom dead center position.

[0074] Step S80: When the decoupling shows that the current mold system is in the same direction accumulation state, a drift slope is generated by fitting multiple sets of external dimension deviation parameters, a pre-configured dynamic damping coefficient is extracted to perform a reduction modulation operation on the drift slope, and the closed-loop compensation action is performed on the feeding mechanism based on the compensation amount after the reduction modulation operation.

[0075] The stamping system includes a central logic controller, a servo feeding module, a high-speed stamping press, and progressive stamping dies. The central logic controller and the servo feeding module are interconnected via an industrial communication bus. The servo feeding module is internally equipped with a high-precision rotary encoder and traction rollers. Regarding step S10, the central logic controller sets a theoretical step length constant according to the layout drawing and converts the theoretical step length constant into a position pulse signal sequence. The specific mathematical conversion association rule is as follows: Let the theoretical step length constant be... The outer diameter of the reference solid of the traction roller is The rated physical resolution pulse constant generated by a high-precision rotary encoder per revolution is (Set value) (pulse / cycle), total number of pulses generated Satisfy the independent equation: In the formula, This is the theoretical step length constant, in millimeters (mm). ); The reference solid outer diameter of the traction roller, in millimeters. ); Pi is a constant, dimensionless, and its values ​​range from [value 1] to [value 2]. ; The rated physical resolution pulse constant, in units of pulses per revolution ( ); The total number of pulses generated is a unitless positive integer. The servo feeding module receives the position pulse signal sequence, drives the traction roller to rotate, and relies on surface static friction to drive the strip material to slide along the feeding guide groove into the inner cavity of the progressive stamping die. The spindle crank of the high-speed stamping machine is connected to a slider mechanism. When the spindle crank drives the slider mechanism to the stripping phase interval that is separated from the strip surface, the central logic controller triggers the feeding enable signal, and the servo feeding module completes the material conveying action of the specified length within the stripping phase interval. The kinematic quantization judgment limit of the stripping phase interval is defined as: with the top dead point of the slider spindle crank as... With the reference zero position fixed, the rotation span of the main spindle crank is at... to The mechanical stroke range. Within this range, the perpendicular orthogonal physical distance between the physical bottom surface of the punch and the upper surface of the strip is greater than the nominal thickness of the strip. times.

[0076] The strip material has micro-fine tube wall characteristics. For the strip material, the stamping equipment is controlled to step-feed the strip material into the mold according to a preset step distance to perform multi-station continuous stamping operation, including: using an integrated rolling forming structure to perform multi-step micro-deformation bending operation; obtaining a pre-calculated springback compensation angle and importing the springback compensation angle into the mold control system as a correction reference; using the correction reference in combination with a high-precision guide sleeve assembly and a floating stripping assembly, performing strong pressure shaping on the formed hollow outer wall after performing the multi-step micro-deformation bending operation to ensure that the tube wall roundness and coaxiality parameters meet the requirements.

[0077] Specifically, the integrated roll forming structure is arranged in the middle section of the progressive stamping die, and multiple sets of concave cavities and punches with decreasing radii of curvature are connected in series along the material conveying direction. The decreasing radii of curvature follow an arithmetic sequence distribution rule: the total number of progressive bending stations is set to... The initial bend radius is The target forming inner fillet radius constant is Then the permutation sequence of the th radius of curvature of the work station Based on algebraic equations Extract and determine. In the formula, The total number of progressive bending stations is greater than [a certain value]. A unitless positive integer; Represents the initial bend radius, in millimeters (mm). ); Represents the constant of the inner fillet radius of the target forming, in millimeters (mm). ); The variable representing the unitless positive integer index of the workstation arrangement sequence, with a value range of... ; Representing the The radius of curvature of the workstation, in millimeters (mm). Here, the aforementioned arithmetic progression distribution rule of the radius of curvature is based on the law of volume invariance and the principle of strain hardening suppression in the bending forming of metallic materials. By making the radius of curvature between adjacent stations decrease linearly, it is ensured that the relative bending plastic strain increment occurring in each pass is controlled within the allowable elongation threshold range of the material. The implementation of this formula effectively avoids micro-cracks on the outer side of the tube wall and wrinkling defects on the inner side caused by a single severe deformation, ensuring continuous stress release and smooth transition of forming load in the multi-step micro-deformation bending process.

[0078] As the strip passes through the integrated roll forming structure, it is subjected to a gradient of bending deflection, and the cross-section continuously converges towards the target arc shape. The pre-calculated springback compensation angle is obtained and imported into the mold control system as a correction benchmark. This includes: using elastoplastic finite element analysis software to input the elastic modulus and yield strength constant of the strip batch. The aforementioned acquisition of the elastic modulus and yield strength constant involves: cutting standard tensile specimens from the same batch of strip, performing uniaxial tension at a constant displacement rate using an electronic universal tensile testing machine, recording the stress-strain curve, and extracting the slope value of the linear elastic stage as the elastic modulus. The stress scalar corresponding to the residual plastic strain is used as the yield strength constant. The elastic modulus is... The unit is gigapascal (GPa). The yield strength constant); The unit is megapascal (MPa). A three-dimensional digital model containing a punch, die, and strip mesh is constructed. A virtual stamping load is applied within the model, and an unloading calculation process is executed. The springback angle vector after virtual strip demolding is extracted and converted into a negative geometric offset parameter, defined as the springback compensation angle. The specific extraction and determination rule for the springback angle vector is as follows: extract the normal tangent equations of the two end faces of the pipe wall opening of the virtual model before and after the unloading calculation, calculate the magnitude of the angle displacement difference between the two normal tangents in the polar coordinate projection system, and define this magnitude as the springback compensation angle. In the formula The unit is radians ( The mold control system reads the springback compensation angle and, by driving the bottom dead center adjustment mechanism, sets an additional downward offset value for the vertical downward depth limit coordinate of the subsequent forming punch, forming the correction reference. The additional downward offset value... The derivation formula for spatial geometric transformation is as follows: In the formula This represents the physical rotational reference radius of the current forming concave cavity. The extracted rebound compensation angle is in radians (rad). The unit is millimeters (mm) ); The unit is millimeters (mm) ); This is the operator for the trigonometric cosine function.

[0079] Furthermore, during the high-pressure forming process, as the upper die holder descends, the high-precision guide sleeve assembly fixed to the bottom of the upper die holder preferentially slides against the guide post on the lower die holder's table surface to constrain lateral sway. As the upper die holder continues to descend, the floating stripping assembly surrounding the punch, under the thrust of its internal pre-tensioned spring, contacts the non-deformable reference strip on the strip surface. The floating stripping assembly applies a clamping force, clamping the strip to the lower die reference plane and suppressing lateral deformation and longitudinal slippage. The clamping force generates this clamping effect. The quantitative calculation equation model is as follows: In the formula The set edge pressure safety empirical coefficient (with a defined value range of ) to ), To input the yield strength of the strip, The physical thickness of the material. is the orthogonal projection length of the perimeter of the contact pressure edge assembly. Where, The blank holder force generated is represented by the unit Newton (N). ); These are unitless empirical constants; The unit is megapascal (MPa). ); Represents the physical thickness of the material strip, in millimeters. ); The orthogonal projection length of the circumference of the contact edge assembly is represented in millimeters. Within the physical boundaries constructed by the floating stripping assembly, the main punch, carrying an additional interference-force stroke corresponding to the corrected reference, applies a load to contact and compress the outer wall of the formed hollow section. The compressive load induces deep plastic flow in the metal lattice of the tube wall cross-section. After the die opens for stripping, the additional interference-force stroke offsets the springback, and the formed micro-tube wall remains stationary at the closed angle. By covering elastic deformation with plastic flow, the macroscopic morphology tolerance of the tube wall converges within the set coordinate zone.

[0080] Furthermore, the controlled stamping equipment, which feeds the strip material into the mold according to a preset step distance to perform multi-station continuous stamping operations, also includes: using a pre-treated cutting edge that has undergone precision oil cutting combined with optical polishing to perform stamping; embedding a gap adjustment block inside the die cavity; and using the gap adjustment block to perform dynamic leveling of the cutting edge to reduce gap offset errors generated during stamping. The precision oil cutting process uses hydrocarbon synthetic insulating oil as the electrical discharge machining fluid medium. The physical characteristic parameters of the hydrocarbon synthetic insulating oil are defined as follows: the kinematic viscosity of the medium is within... to Between, the dielectric breakdown voltage threshold is set to be not less than A constant was maintained. A micro-wire electrode was used to perform pulsed discharge erosion on the mold steel. The extreme boundary of the control electrical parameters for performing the pulsed discharge erosion was limited to: the peak pulse discharge current was maintained at... to The range, the width of a single discharge pulse is set to to The interval, with the single-sided discharge erosion gap constant controlled within, is... to After the electrical discharge cutting (EDC) process, the cutting edge is transferred to an optical projection grinding machine to perform the optical grinding process. A grinding wheel carrying diamond abrasive grains is driven to perform smooth polishing along the direction of the material cutting force. The mechanical cutting feed parameters for this smooth polishing are set as follows: a grit size of [grit size code missing]. to The ultra-fine diamond abrasive wheel has a single feed depth scalar limit that is controlled by hardware. to Arithmetic mean roughness of the surface profile of the target working surface The convergence shutdown threshold is set to .

[0081] Inside the mounting groove on the side of the blanking die cavity, a clearance adjusting block with a wedge-shaped surface is slidably fitted in parallel. The physical geometric angle constant between the wedge-shaped surface of the clearance adjusting block and the horizontal longitudinal displacement axis is... Manufacturing for to The wedge surface of the gap adjustment block is in contact with the side force reference surface of the blanking cutter insert. A servo micro-adjustment screw mechanism is connected to the end of the gap adjustment block. When the control system determines that the blanking gap has a micrometer-level displacement, it sends an adjustment pulse to the servo micro-adjustment screw mechanism. The servo micro-adjustment screw mechanism pushes the gap adjustment block to slide longitudinally along the mounting groove. Based on the mechanical displacement scaling ratio of the wedge surface, the longitudinal sliding amount is converted into a normal pushing force applied to the back of the blanking cutter insert. The mathematical correlation for this longitudinal pushing to normal displacement mapping conversion is: In the formula To perform the normal jacking compensation displacement output for knife edge correction, The longitudinal slippage generated by the servo fine-tuning screw drive. The aforementioned angle of inclination. In the formula, The unit is millimeters (mm) ); The unit is millimeters (mm) ); The constant of the physical geometric angle between the inclined wedge surface and the longitudinal displacement axis is expressed in degrees ( ). ); The value is the trigonometric tangent function. Using the aforementioned transformation displacement, the deviated cutting edge is moved to the coordinate system of the center of symmetry, completing the dynamic leveling operation and restoring the balance of the double-sided punching gap.

[0082] For step S20, such as Figure 2 As shown, the re-upsetting station after stamping completion reads the dimensional deviation parameters of the stamped part, including: capturing the surface contour image of the processed stamped part using a high-speed vision detection module; performing feature point matching and edge fitting processing on the captured surface contour image; extracting the height deviation value, flatness deviation value, and terminal spacing deviation value contained in the data obtained after fitting processing; and comprehensively defining the height deviation value, flatness deviation value, and terminal spacing deviation value as the dimensional deviation parameter. The re-upsetting station is deployed at the end section of the progressive stamping die's physical length. Because it is at the end of the forming action, the reading process avoids the transient vibrations generated at the stretching and rolling nodes. The high-speed vision detection module is fixedly mounted above the exit area of ​​the re-upsetting station and contains an imaging device and a dual telecentric optical lens group. The imaging device uses an effective photosensitive pixel array of not less than 10,000 pixels Sensor. The optical parameters of the dual telecentric optical lens group are calibrated as follows: full field of view telecentricity. Working panoramic depth range System optical distortion rate index When the spindle crank angle sensor feedback slider crosses the bottom dead center and reaches the material removal stationary phase window, the central logic controller sends a hardware trigger pulse to the high-speed vision detection module. The explicit logical boundary for outputting the hardware trigger pulse is: reading the spindle crank angle sensor value, when the angle value falls within... When the physical dead zone is stationary and the slider's downward velocity vector returns to zero, a rising edge trigger pulse is output, and its electrical signal response delay time is limited to... Within this range, the light source component emits light, and the imaging device simultaneously exposes the image, acquiring the surface contour image and storing it within the image cache matrix.

[0083] After extracting the image cache matrix data, the image processing kernel uses a smoothing operator and a bilateral filtering function to suppress shot noise. Within the underlying computational system, the smoothing operator calls... A two-dimensional Gaussian convolution kernel matrix in the pixel dimension; the bilateral filtering function includes the following distribution weight parameters: spatial domain Gaussian kernel standard deviation. Fixed Assignment Constant Pixel grayscale value range Gaussian kernel standard deviation Fixed Assignment Constant Both standard deviation parameters are dimensionless. Step information of the edge gradient is preserved. Subsequently, feature point matching and edge fitting processing are performed. The system loads a pre-stored standard geometric topology reference map. An image retrieval algorithm is applied to retrieve a cluster of localization reference points mapped to the standard geometric topology reference map in the surface contour image. The image retrieval algorithm calls normalized cross-correlation (...). The matching model's internal extraction and judgment equation logic is as follows: It calculates the real-time sliding pixel window matrix and the reference atlas matrix... Inner product coefficients, extracting cross-relationship values. The set of peak projection coordinates is directly output as the positioning reference point cluster. Specifically... The algebraic equation for calculating the inner product is: In the formula This is a scalar value representing the grayscale intensity within a real-time sliding pixel window. For the reference spectral template, the corresponding grayscale intensity scalar. and These are the arithmetic mean of the corresponding matrices, and all the variables mentioned above are values. to Unitless real numbers; The normalized cross-correlation inner product coefficient is represented by a unitless real scalar. A unitless positive integer index representing a local pixel point within the sliding pixel window and the reference map matrix; A summation operator is used to iterate over all pixels. The affine transformation transition matrix, including translation and rotation constants, is calculated using the least squares method. The mapping model of the affine transformation transition matrix is ​​established as follows: and The coordinates of non-collinear scattered points are extracted from the data link and substituted into the objective function of minimizing the sum of residuals. The elements of the rotation and scaling matrix are solved by Gaussian elimination pseudo-inverse operation. to and translation constant , In the formula, and These are the x and y coordinates of the points in the original image before transformation, in pixels. ); and These are the x and y coordinates of the target point in the reference coordinate system after affine transformation, in pixels. ); Construct a unitless two-dimensional rotation and scaling matrix Internal elements; and They respectively constitute the translation constant vector The horizontal and vertical translation constant elements, in pixels ( ).

[0084] In the objective function, Representing the A column vector of extracted measured point coordinates; The first in the reference atlas A column vector of corresponding reference point coordinates; This represents the two-dimensional rotation and scaling matrix constructed earlier; This represents the translation constant vector generated alongside the translation. The total number of points involved in the calculation, a unitless positive integer; The unitless positive integer index parameter representing the point of operation is limited to a range of values. ; This indicates the Euclidean distance norm operation on a vector. Represents the minimization function operator; This represents the summation operator. Spatial inversion is performed through affine transformation to calibrate the image coordinate system to a horizontal orthogonal reference coordinate system. Within the calibrated image region, the region of interest (ROI) where the target measurement boundary is located is defined. The geometric spatial definition of the ROI is as follows: using the center vector topology line extracted from the initial identification of the outer boundary as a reference, extend equidistantly inwards and outwards along the normal direction of the boundary curve for a length of... A rectangular pixel envelope of pixels is defined. The pixel gray-level gradient response amplitude is extracted along the boundary normal. After locating the gray-level abrupt change peak point, the surrounding neighboring pixel data is extracted, and a continuous gray-level distribution equation is constructed using a curve approximation model.

[0085] The curve approximation model specifically utilizes a quadratic polynomial fitting curve function equation framework: In the formula Normal coordinates The gray intensity at that point is a real number, and the coefficient is... Perform using neighborhood discrete pixel data The result is obtained through nonlinear least squares iterative solution. In the formula, Represents the real number of pixel grayscale intensity, with values ​​ranging from 0 to 1. to Unitless parameters between them; This represents a one-dimensional position coordinate variable distributed along the normal, in pixels. ); and These represent the constant coefficients of the quadratic and linear terms obtained from the aforementioned continuous gray-scale distribution equation; coefficients The unit is ,coefficient The unit is ,coefficient Unitless. Solve for the extreme points of the derivative of the equation to extract the sub-pixel coordinate point set. The specific sub-pixel precision coordinate analytical logic is as follows: For the fitted equation... Execute about space variables The first derivative operation, let the fundamental derivative... The sub-pixel precise coordinates of the one-dimensional extremum center point are obtained as follows: In the formula, This represents the scalar value of the exact center coordinates of the single-dimensional sub-pixel, in pixels. The coefficients in this quadratic polynomial continuous gray-scale distribution equation , , The specific solution approach is as follows: The underlying hardware system uses the initially located pixel-level grayscale transition peak as the center, extracts the grayscale response values ​​and position coordinates of 3 to 5 consecutive discrete pixels in its normal neighborhood, and constructs an overdetermined system of linear algebraic equations. The computational kernel then calls the Gauss-Newton iteration method or the least squares matrix pseudo-inverse to solve the problem and approximate the above coefficients. The sub-pixel coordinate point set is connected and fitted to reconstruct a continuous edge boundary vector model.

[0086] Further, the system identifies the outer vector boundary of the top edge of the stamped part and calculates the vertical normal distance from a specific group of points on the outer vector boundary to the fitted straight line of the bottom horizontal reference. After calculating the average of all vertical normal distance values, the height deviation value with positive and negative signs is obtained. The height deviation value... The algebraic scalar derivation formula is defined as follows: In the parameters, Extract the corresponding discrete sequence of ordinates of the point set for the outer boundary. This represents the total number of sampling points. is the scalar of the ordinate of the horizontally fitted straight line at the bottom. The absolute constant of the theoretical nominal height specified for the product drawings. Where, This indicates the numerical value of the height deviation, in millimeters (mm). ); The pixel equivalent conversion factor for the optical system, expressed in millimeters per pixel (mm). ); the pixel equivalent conversion coefficient The method for acquiring and dynamically determining the calibration parameters is as follows: During the equipment initialization and calibration phase, a high-precision alumina ceramic calibration plate with known physical spacing characteristics is placed aligned with the working focal plane of the optical lens (its surface is printed with a standard checkerboard pattern or dot array, and its manufacturing tolerance is specified). The visual detection module captures the image of the calibration board and extracts the pixel coordinates of the feature point array. The central processing unit compares the difference between the actual known relative distance in physical space and the pixel span of the corresponding feature points in the image, and uses a multi-point averaging or least squares fitting algorithm to obtain the average pixel equivalent scaling factor. This compensates for and avoids the inherent optical distortion and magnification errors in industrial lens processing or assembly from a physical perspective. This represents the vertical coordinates of the generated outer boundary points, in pixels. ); This represents the total number of boundary sampling points, a unitless positive integer. The unitless positive integer sequence index representing the boundary extraction point set, whose value boundaries are limited to: ; The unit is pixels ( ); The unit is millimeters ( ); This is an accumulation operator. A discrete point cloud set representing the upper surface topography is selected, and a local smooth reference surface is fitted in three-dimensional space using a matrix factorization algorithm. The deployed matrix factorization algorithm calls singular value decomposition (SVD) at its underlying level. The kernel processing and matrix execution path are limited to: calculating the arithmetic mean of the point cloud set in the discrete coordinate system to perform decentralization, generating a third-order covariance matrix; and then performing the following on the covariance matrix. The singular value decomposition operation extracts the feature column vector corresponding to the smallest eigenvalue, and directly assigns this vector as the spatial unit normal vector parameter of the local smoothing reference surface. The orthogonal deviation distance from the local smoothing reference surface to all coordinate points within the point cloud set is calculated, and the flatness deviation value is output.

[0087] Independent orthogonal deviation distance The arithmetic extraction model is as follows: In the formula These are the equation coefficients for the locally smoothed reference surface. The flatness deviation value... The extraction formula is , which is the algebraic sum of the absolute values ​​of the extreme values ​​of the deviation distances in the positive and negative normal directions. In the formula, Indicates the first The orthogonal deviation distance of each three-dimensional point from the reference plane, in millimeters. ); These represent the points in the input point cloud set, respectively. The three-dimensional physical projection coordinate parameters of a spatial point, with the physical unit being millimeters. ); here A unitless positive integer index variable representing discrete spatial points in the input point cloud set; The normal vector constituting the local smooth reference surface has no unit constant. For the constant term of the plane equation, the unit is millimeters (mm). ); This indicates the generated flatness deviation value, in millimeters (mm). ); and These represent operators for finding the maximum and minimum values, respectively. This represents the operator for calculating absolute values. Extract the equations of the symmetrical central axes of the adjacent end bodies. Calculate the spatial translation distance parameter between the two symmetrical central axes, subtract the theoretical nominal spacing value, and obtain the deviation value of the terminal spacing. Obtain the measured spatial translation distance parameter. The data derivation process is as follows: In the formula and respectively through The pixel coordinates of the corresponding normal reference points on the two centerlines to be measured, extracted by linear transformation. Where, and The physical unit is pixel ( ); This represents the measured distance parameter, in millimeters (mm). ); The conversion factor for the optical system is expressed in millimeters per pixel. The terminal spacing deviation value is equivalent to Directly subtract the given spacing nominal constant (unit: The predetermined data communication frame format follows a specific byte address alignment distribution: the total length is fixed. bytes, including those occupying The first synchronization frame header of the bytes accounts for The byte stores a clock-synchronized long integer global timestamp, followed by three bytes, each occupying a specific space. byte width Single-precision floating-point memory segments respectively load the deviation constants for height, flatness, and spacing, with the tail reserved. byte padding Cyclic redundancy check (CR) is used for bus verification. The height deviation, flatness deviation, and terminal spacing deviation are combined into the overall dimensional deviation parameter and then fed into the control processing bus.

[0088] Regarding step S40, the continuously set multiple stamping cycles are divided into monitoring time windows, and multiple sets of the external dimensional deviation parameters within the monitoring time windows are collected. The purpose of the control action is to provide a continuous analysis basis with a time dimension for discrete sampled data. The central logic controller plans a queue buffer space of fixed length within the non-volatile memory area. The number of stamping cycles is defined. The constant model of the correlation function is established as follows: In the formula, the denominator For the single-cycle operation time of the servo spindle system, the molecule To determine the heat conduction time constant for achieving steady-state heat transfer based on the volume and thermal conductivity of the mold base itself, The sampling scaling constant is set to (as follows) ), This is the underlying function operator for rounding down to the nearest integer. In the formula, The number of stamping cycles in the calculated output is represented as a unitless discrete positive integer. The unit is seconds ( ); The unit is seconds ( ); This is a unitless scalar constant. The latest collected dimensional deviation parameters, accompanied by timestamps, are continuously pushed into the head address of the queue cache space. When new data is pushed in, the memory controller schedules the process, performing a postshift operation on the entire existing historical data block towards the tail address of the queue. End parameters exceeding the time span capacity are marked as discarded and released by the system. Relying on continuous synchronized data enqueue and dequeue operations, the system precisely maintains the analysis sample library at its rated capacity, continuously collecting multiple sets of dimensional deviation parameters within the monitoring time window.

[0089] For step S50, such as Figure 3 As shown, the calculation of the direction of the difference between the dimensional deviation parameters corresponding to two adjacent stamping cycles, and the statistical analysis of the frequency of alternating positive and negative jumps in the direction of the difference within the monitoring time window, includes: arranging all dimensional deviation parameters within the monitoring time window in chronological order; before extracting the polarity sign operation flow, additional Boolean logic interception judgment for background white noise filtering is performed: calculating the modulus of the algebraic difference between the extracted data and the data from the previous row, requiring that the modulus of this difference be greater than a pre-locked optical measurement detection limit constant (e.g., set to...). When the difference meets the condition, subsequent directional polarity physical determination is allowed. The mathematical sign of the difference value is extracted by subtracting the front parameters from the back parameters as the difference value direction; the total number of times the difference value direction changes from positive to negative and from negative to positive is accumulated, and the total number of occurrences is calibrated as the alternation frequency. The reason why the deviation of the external dimensions cannot be decoupled from the root cause of the failure is that the deviation caused by the thermal expansion of the mold and the metal oscillation has the characteristic of numerical overlap at a certain node.

[0090] Specifically, the logic operation unit reads the data image from the queue cache space. Based on the timestamp identifier, the out-of-order data is reconstructed using a monotonically increasing time signature, arranging all dimensional deviation parameters within the monitoring time window in chronological order. The logic operation unit allocates dual index pointers and performs a progressive traversal along the reconstructed data column. The tail pointer extracts parameters from later time nodes, and the head pointer extracts parameters from adjacent time nodes. The arithmetic module executes subtraction instructions. The calculation core specifically reads the sign bit level of the subtraction result output. When the sign bit indicates the result is greater than zero, the calibrated stamping dimension is in an expanding evolution state; the positive sign is extracted and defined as the difference direction. When the sign bit indicates the result is less than zero, the calibrated stamping dimension is in a contracting evolution state; the negative sign is extracted and defined as the difference direction. When a difference of zero is encountered, the program copies the difference direction from the preceding adjacent time and assigns it to the current node. The offset floating-point array is converted into a logical linked list containing only positive and negative polarity identifiers. The state frequency statistics module sets the comparison cursor to advance step-by-step along the logical linked list. The cursor temporarily stores the polarity state of the previous node and performs an XOR comparison with the current node's polarity state. The underlying Boolean expression of the specific logic gate that triggers the XOR comparison extraction judgment is: Within this framework, the positive polarity Boolean assignment of the system is always specified as 0. Negative polarity Boolean assignment is always 0. , and These represent the status words read from the previous and current nodes, respectively. The result of a bitwise XOR operation... (If the output is high), the hardware determines and captures the single-circuit polarity space reversal inflection point. In the formula, The Boolean result of the logical operation represents the output; and For the value to be or A unitless Boolean constant; This represents the bitwise XOR operation operator at the underlying level. Each time an independent turning point is captured, the state frequency statistics module sends a trigger pulse to the accumulation unit to perform an increment calculation. When the cursor reaches the end of the linked list, the system extracts the integer result latched by the accumulation unit and marks the total number of occurrences as the alternation frequency. The alternation frequency within the algorithm system... The mathematical representation of the attribute is calculated as follows: That is, in a capacity of Within the time window span of individual samples, a Boolean impulse event causes a jump, triggering the output true value to be... The cumulative integer sum. Where, The cumulative value representing the alternation frequency of the final output is a unitless positive integer scalar. Represents the first in the time window sequence The real-time XOR logic outputs a Boolean value at each decision node; here... The unitless positive integer monotonically increasing index representing the decision node within the time window sequence, with a value range of... , A scalar representing the number of stamping cycles covered by the current monitoring time window; This represents the summation operator for the sequence of judgment results. The alternation frequency, at the algorithm layer, represents the frequency density of macroscopic reciprocating fluctuations.

[0091] Regarding step S60, comparing the alternation frequency with a pre-set state determination threshold to decouple and determine whether the current mold system is in a high-frequency oscillation state or a unidirectional accumulation state includes: when the alternation frequency is greater than the state determination threshold, confirming that the material is in a random elastic fluctuation stage and outputting the high-frequency oscillation state result; when the alternation frequency is less than or equal to the state determination threshold, and the absolute difference between the first and last parameters within the monitoring time window crosses the safety tolerance limit, confirming that continuous thermal expansion or mechanical wear has occurred and outputting the unidirectional accumulation state result. The state determination threshold is calculated based on the machine body's natural frequency and the mold base resonance characteristics under the no-load operation state of the punch press. The state determination threshold is calculated and solved. The product algebra formula is set as follows: In the set of constraint parameters of the equation, The real number of the first-order natural mechanical frequency extracted from the piezoelectric test point of the main machine tool. The actual rated stamping frequency given to the frequency converter. The set random noise suppression safety filter constant (value locked at) In the formula, The threshold value representing the state determination is a unitless positive integer. The number of stamping cycles encompassed by the monitoring time window; The unit is Hertz (Hz) ); The unit is Hertz (Hz) ); It is a unitless pure number; This is the rounding operator. The physical mechanism of the mathematical model for solving the state determination threshold lies in the frequency domain decoupling of the machine's inherent resonance period and the actual high-frequency oscillation signal of the material being transported. In the equation, the frequency ratio... The theoretical occurrence and aliasing frequency of the machine tool's inherent mechanical resonance period within a single stamping action time window were characterized. A safety filter constant for background noise random suppression was introduced. The fundamental purpose is to construct a dynamically adaptive digital low-pass discriminator. This filter constant can mask and suppress occasional false positive data jumps caused by white noise from the workshop environment's foundation vibration or high-frequency thermal noise from optical sensors. By setting this dynamic comparison threshold based on physical resonance frequency mapping, the system can possess anti-artifact capabilities, effectively stripping away and identifying intrinsic high-frequency oscillations purely caused by the release of residual stress in the metal lattice.

[0092] The numerical comparison module compares the real-time calculated alternation frequency with a preset state determination threshold. If the alternation frequency is greater than the state determination threshold, it indicates that the external dimensions are in a high-frequency bidirectional jump phase. Logical verification: After demolding, the unevenly distributed lattice stress inside the metal strip is released in a concentrated manner, causing the strip body to produce reciprocating micro-vibrations. High-frequency fluctuations do not conform to the physical progressive wear law. The system outputs the high-frequency oscillation state result accordingly. If the comparison result determines that the alternation frequency does not exceed the state determination threshold, the system proceeds to execute the span amplitude verification logic. The first and last parameters in the memory column within the monitoring time window are called, and arithmetic subtraction is performed to calculate the modulus of the value. When the modulus of the value crosses a preset safety tolerance limit, the preset safety tolerance limit... The numerical inference and calibration model is established as follows: In which the equation parameters The constant is the total allowable bandwidth constant for bidirectional tolerance variation, extracted from the target monitoring dimension as specified on the process engineering drawings. Where, This represents the preset safety tolerance limit scalar, in millimeters (mm). ); The unit is millimeters ( The system confirms that the dimensional deviation continues to extend along a fixed dimension. This unidirectional accumulation tendency is objectively attributed to the thermal expansion of the guide components caused by the heat load generated during continuous machining, or to the progressive wear of the cutting edge. The system confirms the presence of continuous thermal expansion or mechanical wear and outputs the unidirectional accumulation result.

[0093] Regarding step S70, when the decoupling determines that the current mold system is in the high-frequency oscillation state, the step adjustment command issued to the feeding mechanism is blocked, and a pressure holding delay command is sent to the re-upsetting station to absorb microscopic material stress by extending the dwell time of the punch at the bottom dead center. Under this unsteady condition, the external dimensions of the stamped part exhibit fluctuations around the reference zero position. If a displacement compensation command is issued to the servo controller based on the latest transiently detected deviation vector, the reverse torque output by the servo driver will be superimposed with the fluctuation of the metal strip at the same frequency, inducing the control position loop to diverge and become unstable. The motion control scheduling unit inside the central logic controller locks the write channel connected to the position data bus of the feeding mechanism, blocking the step adjustment command issued to the feeding mechanism. The step feed constant is anchored on the set coordinate reference, cutting off the reverse excitation source of the servo system.

[0094] Furthermore, such as Figure 4 As shown, sending a pressure-holding delay command to the re-upsetting station, utilizing the extended dwell time of the punch at the bottom dead center to absorb microscopic material stress, includes: extracting the maximum deviation from the extreme peak value under the high-frequency oscillation state; and calibrating it as the maximum deviation from the extreme peak value. The extreme value traversal judgment function command for extracting the extreme peak value is: Directly in length of The array is fully traversed to retrieve a single, independent floating-point scalar with an extremely large magnitude, and this scalar value is then assigned to the maximum deviation from the peak value. In the formula, This represents the extracted maximum deviation from the peak value, in millimeters (mm). ); These represent the elements in the dimensional deviation queue within the current monitoring time window, with the physical unit being millimeters (mm). ); Represents the low-level operator for extracting extreme values; This represents the modulus operator. Based on a pre-set library of material plastic rheology curves, the stress release time required to match the maximum deviation from the extreme peak value is calculated in reverse; the inverse calculation model is based on the generalized viscoelastic relaxation equation: Solve for the time required for stress relief. The inverse transformation algebraic equation is: . Within the formula To retrieve the viscoelastic characteristic time constant of a specific alloy in the library, The initial equivalent stress is derived by multiplying the maximum deviation from the extreme peak value by the model stiffness mapping. This represents the safety residual internal stress threshold for yielding after material stress dissipation. Multiply this calculated time variable by the base value. The scale is converted to microseconds and limited to... to Physical span range. In the formula, Represents the physical time variable of pressure holding (Unit: seconds) The residual internal stress exhibits attenuation (in megapascals). ); For the natural constant The exponential function operator with base 0; Operator for the natural logarithm function; This represents the theoretical time required for stress relief, initially derived in seconds. ); The unit is seconds ( ); The unit is megapascal (MPa). ); The unit is megapascal (MPa). ).

[0095] Under the physical boundary constraint of constant displacement (constant strain) during the pressure holding period at the bottom dead center of the punch, the elastic deformation energy inside the metal lattice gradually transforms into irreversible viscous-plastic flow, macroscopically manifested as an exponential relaxation decay of residual internal stress over time. The viscoelastic characteristic time constant is... This is an empirical physical characteristic parameter extracted through nonlinear curve fitting after obtaining scattered data from a uniaxial stress relaxation mechanical experiment conducted on the current batch of metal materials. The required holding pressure time is calculated based on this mechanical model, thus avoiding the production cycle waste (over-holding pressure) or springback dimensional failure (under-holding pressure) caused by blindly setting the time based on manual experience. The required stress release time is encapsulated in the holding pressure delay command and sent to the pressure drive servo unit, keeping the punch in the lowest fixed position. Extending the dwell time of the punch at the bottom dead center position absorbs microscopic material stress. The conversion logic for executing the dwell-at-still command is: dividing the calculated stress release time by the servo underlying interpolation operation cycle constant (set to...). Extract the total number of pulse group sequences containing integer values ​​of the action loop. Within this pulse group sequence, continuously and constantly write a constant to the servo position loop accumulation increment field. The physical power of the feed is cut off. When the mechanical movement of the punch reaches the lowest bottom dead center coordinate limit, the servo motor injects a set excitation current and outputs a torque that balances the reverse thrust of the mold spring, keeping the punch stationary at the lowest shaping position. Within the extended dwell time window, the distorted lattice dislocations inside the metal material continue to move and rearrange along the slip surface, dissipating the residual elastic internal stress and transforming it into a steady-state plastic shaping state.

[0096] Regarding step S80, when the decoupling reveals that the current mold system is in the same-direction accumulation state, a drift slope is generated based on fitting multiple sets of external dimension deviation parameters. A pre-configured dynamic damping coefficient is extracted to perform a reduction modulation operation on the drift slope, and the closed-loop compensation action is performed on the feeding mechanism based on the compensation amount after the reduction modulation operation. For example... Figure 5 As shown, the specific calculation and derivation steps include: the processing unit establishes a mapping operation space to form a discrete scatter matrix; linear regression is performed on the multiple sets of external dimension deviation parameters to obtain the drift slope characterizing the unidirectional offset rate; and the least squares method is applied to obtain the tangent offset drift slope. The specific discrete algebraic equations are set together as follows: Within the formula, This represents the calculated tangent offset slope, expressed in millimeters per second. ); The time-series timestamp of the collected samples is represented by the independent variable, in seconds. ); This indicates the deviation of the sample size from the real number of the dependent variable, in millimeters (mm). ); The arithmetic mean time constant of the set of timestamp sequences (in units of 1000 m / s). ); The arithmetic mean deviation constant of the set of deviation distance sequences (in units of 1000 m / s) ); This represents the total number of samples involved in the calculation; here... The unitless positive integer isomorphic index representing the timestamp sequence and size deviation sequence of the collected samples, with a value range of 1. ; To iterate over the summation operator, the dynamic damping coefficient is read from the system register. The dynamic damping coefficient has a floating-point value between 0 and 2. During the cold start and complete reset phase of the power-on motherboard system on the automated production line, the default initialization constant value of the dynamic damping coefficient is set to 0.50.

[0097] The attenuation bias is obtained by multiplying the newly measured dimensional deviation parameter by the dynamic damping coefficient; the attenuation bias is then summed with the expected development variable corresponding to the drift slope to generate the compensation amount after amplitude reduction modulation. The final amplitude reduction modulation composite compensation amount is then calculated. The formula for predicting physics using univariate addition in algebraic mathematics is established as follows: The equation components include: the dynamic damping coefficient parameter referenced from the address. The latest single-point static measured deviation lags real component The drift slope constant referenced in the pre-module solution And the look-ahead time interval delay constant for anticipating the transmission action duration. In the formula, This represents the compensation amount in the final synthesized output, in millimeters (mm). ); This represents the dynamic damping coefficient parameter, which is a unitless multiplier factor. The unit is millimeters ( ); The unit is millimeters per second (m / s) ); The unit is seconds ( In this additive formula, the look-ahead time interval delay constant that compensates for the time span of the expected transmission action is used. The specific measurement and synthesis method is as follows: the main control system reads and accumulates the physical transmission delay period of the communication bus, the time consumed by a single scan cycle of the programmable logic controller (PLC) internal program, and the inherent dead-zone response time of the servo mechanical actuator from power-on to generating effective torque. The sum of the above three physical delay times is used as the inherent look-ahead time interval delay constant of the system. The central logic controller converts the compensation amount after amplitude reduction modulation into an incremental drive command containing polarity sign and step number, and sends it to the feeding mechanism. The servo feeding module decodes the command and performs micro-step adjustment of the length extension and retraction of the metal strip entering the mold cavity channel, relying on physical conveying correction to offset the cumulative deformation deviation caused by thermal expansion and contraction of mechanical parts or physical wear.

[0098] Furthermore, the dynamic damping coefficient is an adjustable control parameter, such as... Figure 6 As shown, the closed-loop compensation action also includes an adaptive property feedback loop: periodically acquiring the hardness fluctuation ratio of newly cut material batches; the automated production line deploys a non-contact online eddy current hardness detection sensor at the beginning of the feeding channel. The sensor emits a high-frequency alternating magnetic field towards the material body and receives the eddy current impedance electromagnetic signal induced inside the material based on the principle of electromagnetic induction. The signal processing board performs Fourier spectrum analysis and benchmark conversion on the electromagnetic signal, converting it into a continuously distributed array of hardness measurement values ​​and uploading it to the main control network. The control program has a built-in timer interrupt service to periodically acquire the hardness fluctuation ratio of newly cut material batches. A unitless hardness fluctuation ratio is constructed. The specific physical and mathematical equation is defined as follows: In the equation parameters: This is a scalar value representing the average Vickers hardness of the measured metal during the current detection cycle, calculated and returned based on eddy current testing. To standardize the nominal Vickers hardness constant loaded into the database. Where, The real number representing the extracted, anisotropic hardness fluctuation ratio. The unit is the Vickers hardness scale unit ( ); The unit is the Vickers hardness scale unit ( The calculation module performs a difference operation on the average hardness value sequence within adjacent detection cycles.

[0099] When the difference operation result continuously shows a positive increasing value, indicating that the hardness floating ratio is rising, the value corresponding to the dynamic damping coefficient is reduced according to the preset adjustment span; the specific algebraic step overwrite control logic equation for the reduction operation is: update the damping coefficient value. Simultaneously, the code executes a threshold boundary judgment comparison sub-statement, intercepts and specifies that the coefficient value after iterative recalculation is greater than the bottom edge anti-oscillation stability limit safety constant. In the formula, This represents the newly generated dynamic damping coefficient value, which is dimensionless. This represents the original dynamic damping coefficient value saved in the previous control cycle, without units. The displacement output tracking weight of single-cycle servo compensation intervention is reduced from the algorithm gain perspective. The servo traction mechanism is constrained to perform correction using a multi-frequency, micro-step trajectory method to avoid overshoot rebound caused by harder materials. When the difference operation result reverses to a negative value and approaches a small fluctuation on the zero axis, indicating that the hardness fluctuation ratio is decreasing and tending to level off, the material's tensile deformation capacity weakens, and the rebound disturbance amplitude decreases. The specific value corresponding to the dynamic damping coefficient is amplified according to the preset adjustment span. The algebraic step increment overwrite logic of the amplification operation follows the execution of the increment equation: updating the damping coefficient value. The system sets the maximum upper limit threshold for the damping coefficient to 1.85 to prevent drive divergence. The output amplitude suppression constraint is removed, allowing the servo communication bus to output feed step size compensation commands with larger amplitudes, shortening the system closed-loop tuning and recovery time span, and accelerating the dimensional convergence process.

[0100] After the closed-loop compensation action is performed on the feeding mechanism based on the compensation amount after amplitude modulation, the safety inspection closed loop also includes: in the immediately following monitoring time window, rereading the latest dimensional deviation parameters of the newly stamped part; verifying whether the latest dimensional deviation parameters converge and fall within the set qualified benchmark range; if it is determined that they have not fallen within the qualified benchmark range, it is determined that the mold has physical wear damage, triggering the full-line emergency stop control logic and illuminating the audible and visual alarm device. After the servo traction mechanism completes the compensation intervention feedback and is ready, the metal strip material cut into the correction point completes the action sequence in sequence inside the progressive mold. In the immediately following monitoring time window, the high-speed vision detection module is scheduled by the strobe pulse, restarts the capture and algorithm contour edge extraction and analysis process, and outputs the latest dimensional deviation parameters that are completely controlled by the compensation correction constraints. The control unit delivers the extracted values ​​to the pre-set qualified benchmark range for boundary condition Boolean comparison. The qualified benchmark range is composed of the theoretical tolerance upper and lower limit constants and the reserved safety anti-collision margin.

[0101] If the logic comparison result indicates that the latest dimensional deviation parameter remains within the set acceptable reference range, it confirms that the compensation intervention has effectively eliminated the physical error source causing the drift, the closed-loop system has achieved the correction target, and the normalized processing continues to operate. If the judgment logic finds that the feedback data has not fallen within the acceptable reference range, and the parameter deviation shows a divergent phenomenon, the main control processor determines that mechanical wear damage, which is not elastic deformation, has occurred within the precision component group of the mold entity. This includes the breakage of the pressure-bearing cutting main punch, the disintegration of the high-pressure pre-tightening nitrogen spring of the floating unloading plate, or the jamming and tearing of the inner wall of the positioning guide sleeve due to dry friction. Faced with the physical failure of the hardware entity, the control underlying monitoring kernel immediately triggers a hardware interrupt request. Under the guidance of the full-line emergency stop control logic, the power manager disconnects the AC contactor contact group of the main drive circuit, physically blocking the kinetic energy output of the servo spindle motor stator. Simultaneously, a release and locking command is sent to the composite pneumatic brake control valve located on the main drive shaft, using mechanical resistance torque to stop and lock the slider linkage mechanism. The control system synchronously outputs excitation current to the audible and visual alarm device, issuing a shutdown error message. The underlying data gateway automatically encapsulates a record packet containing traced waveforms and fault code data frames, and pushes it to the execution control system server and field operation terminals to guide targeted diagnostic and replacement repair operations.

[0102] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for automatic compensation and closed-loop control of forming errors in precision stamped parts, comprising the following steps: The stamping equipment is controlled to deliver the strip material into the mold according to the preset step distance to perform multi-station continuous stamping operation; At the re-upsetting station after stamping is completed, the dimensional deviation parameters of the stamped part are read. The feeding mechanism performs closed-loop compensation based on the aforementioned dimensional deviation parameters. The method is characterized by further comprising the following control process: The continuously set multiple stamping cycles are divided into monitoring time windows, and multiple sets of the external dimension deviation parameters within the monitoring time windows are collected. Calculate the direction of the difference between the dimensional deviation parameters corresponding to two adjacent stamping cycles, and count the frequency of alternation between positive and negative jumps in the direction of the difference within the monitoring time window; The alternation frequency is compared with a preset state determination threshold to decouple and determine whether the current mold system is in a high-frequency oscillation state or a unidirectional accumulation state. When the decoupling determines that the current mold system is in the high-frequency oscillation state, the step adjustment command issued to the feeding mechanism is blocked, and a pressure holding delay command is sent to the re-upsetting station to absorb micro-material stress by extending the dwell time of the punch at the bottom dead center position. When the decoupling determines that the current mold system is in the same direction accumulation state, a drift slope is generated by fitting multiple sets of external dimension deviation parameters, a pre-configured dynamic damping coefficient is extracted to perform a reduction modulation operation on the drift slope, and the closed-loop compensation action is performed on the feeding mechanism based on the compensation amount after the reduction modulation operation.

2. The method for automatic compensation and closed-loop control of forming errors in precision stamping parts according to claim 1, characterized in that, The controlled stamping equipment delivers the strip material into the mold in a step-by-step manner according to a preset step distance to perform multi-station continuous stamping operations, including: The strip has micro-tube wall characteristics, and for the strip, a multi-step micro-deformation bending operation is carried out using an integrated rolling forming structure; Obtain the pre-calculated springback compensation angle and import the springback compensation angle into the mold control system as a correction reference; Using the aforementioned correction reference combined with a high-precision guide sleeve assembly and a floating stripping assembly, a high-pressure shaping process is applied to the formed hollow outer wall after the multi-step micro-deformation bending operation to ensure that the pipe wall's roundness and coaxiality parameters meet the requirements.

3. The method for automatic compensation and closed-loop control of forming errors in precision stamping parts according to claim 2, characterized in that, The controlled stamping equipment delivers the strip material into the mold in a step-by-step manner according to a preset step distance to perform multi-station continuous stamping operations, and also includes: Stamping is performed using a pre-precision oil-cutting and optical grinding blade; A clearance adjustment block is embedded inside the blanking die cavity; The gap adjustment block is used to perform dynamic leveling operation on the punching blade to reduce the gap offset error generated during the punching process.

4. The method for automatic compensation and closed-loop control of forming errors in precision stamping parts according to claim 1, characterized in that, The re-upsetting station after stamping formation reads the dimensional deviation parameters of the stamped part, including: The surface contour image of the completed stamping part is captured by a high-speed vision detection module. Feature point matching and edge fitting are performed on the captured surface contour image; Extract the height deviation, flatness deviation, and terminal spacing deviation values ​​from the data obtained after fitting. The height deviation value, the flatness deviation value, and the terminal spacing deviation value are collectively defined as the external dimensional deviation parameter.

5. The method for automatic compensation and closed-loop control of forming errors of precision stamped parts according to claim 1, characterized in that, The calculation of the direction of the difference between the dimensional deviation parameters corresponding to two adjacent stamping cycles, and the statistical analysis of the frequency of alternating positive and negative jumps in the direction of the difference within the monitoring time window, includes: All dimensional deviation parameters within the monitoring time window are arranged in chronological order. Extract the mathematical sign of the difference between the back row parameters and the front row parameters as the direction of the difference; The total number of times the difference direction changes from positive to negative and from negative to positive is accumulated, and the total number of occurrences is denoted as the alternation frequency.

6. The method for automatic compensation and closed-loop control of forming errors of precision stamped parts according to claim 5, characterized in that, The step of comparing the alternation frequency with a pre-set state determination threshold to decouple and determine whether the current mold system is in a high-frequency oscillation state or a unidirectional accumulation state includes: When the alternation frequency is greater than the state determination threshold, the material is confirmed to be in the random elastic fluctuation stage, and the high-frequency oscillation state result is output. When the alternation frequency is less than or equal to the state determination threshold, and the absolute difference between the first and last parameters within the monitoring time window crosses the safety tolerance limit, continuous thermal expansion or mechanical wear is confirmed, and the same-direction cumulative state result is output.

7. The method for automatic compensation and closed-loop control of forming errors of precision stamped parts according to claim 1, characterized in that, Sending a pressure holding delay command to the re-upsetting station, utilizing the extended dwell time of the punch at the bottom dead center to absorb microscopic material stress, includes: Extract the maximum deviation from the extreme peak value under the high-frequency oscillation state; Based on a pre-set library of material plastic rheology curves, the time required for stress release that matches the maximum deviation from the extreme peak value is calculated in reverse. The time required for stress release is encapsulated in the pressure holding delay command and sent to the pressure drive servo unit to keep the punch in the lowest shaping position.

8. The method for automatic compensation and closed-loop control of forming errors in precision stamping parts according to claim 1, characterized in that, The process of generating a drift slope based on fitting multiple sets of dimensional deviation parameters, and extracting a pre-configured dynamic damping coefficient to perform amplitude reduction modulation on the drift slope includes: A linear regression operation is performed on the multiple sets of dimensional deviation parameters to obtain the drift slope characterizing the unidirectional offset rate; Read the dynamic damping coefficient from the system register, the dynamic damping coefficient having a floating-point value between 0 and 2; The attenuation bias is obtained by multiplying the newly measured dimensional deviation parameter by the dynamic damping coefficient. The attenuation bias and the expected development variable corresponding to the drift slope are summed to generate the compensation amount after the amplitude reduction modulation operation.

9. The method for automatic compensation and closed-loop control of forming errors in precision stamping parts according to claim 8, characterized in that, The dynamic damping coefficient is an adjustable control parameter, and the closed-loop compensation action further includes: Periodically obtain the hardness fluctuation ratio of newly cut strip batches; When the hardness fluctuation ratio shows an increasing trend, the value of the dynamic damping coefficient is reduced according to the preset adjustment span. When the hardness fluctuation ratio shows a decreasing trend and tends to level off, the value corresponding to the dynamic damping coefficient is amplified according to the preset adjustment span.

10. The method for automatic compensation and closed-loop control of forming errors in precision stamping parts according to claim 1, characterized in that, After performing the closed-loop compensation action on the feeding mechanism based on the compensation amount after the amplitude reduction modulation operation, the method further includes: Within the immediate subsequent monitoring time window, the latest dimensional deviation parameters of the newly stamped part are reread; Verify whether the latest dimensional deviation parameter converges and falls within the set acceptable reference range; If the mold fails to fall into the qualified benchmark range, it is determined that the mold has suffered irreversible physical wear and damage, triggering the emergency stop control logic for the entire line and illuminating the audible and visual alarm device.

Images