Dotting production process of longitudinal beam U-shaped numerical control wing surface dotting production line

Abstract

The application relates to the technical field of longitudinal beam precise stamping manufacturing control, and discloses a dotting production process of a longitudinal beam U-shaped numerical control wing surface dotting production line, which comprises the following steps: acquiring longitudinal beam theoretical dotting parameters and initializing dotting serial numbers and compensation amounts, combining a theoretical cumulative position and the compensation amounts to calculate a final dotting position, completing dotting through zoning pressure of a material pressing unit and time sequence control of pre-striking locking and post-striking releasing; subsequently, servo load data of a wing surface material pressing assembly is collected to judge a cross section stable state, pitch error and cross section stable margin are calculated under the stable state, a wing surface development error memory model is constructed or updated based on the pitch error, a final compensation amount is obtained through model prediction theoretical compensation amount and amplitude limiting combined with the stable margin, and the above steps are cyclically executed until all dotting is completed. The process improves dotting pitch consistency and double-wing symmetry accuracy, and optimizes post-stage dotting stability and production batch stability.

Description

Technical Field

[0001] This application relates to the field of precision stamping manufacturing control technology for longitudinal beams, specifically a dotting production process for a CNC stamping production line for U-shaped longitudinal beams. Background Technology

[0002] Due to issues such as the difference in fiber elongation between the inner and outer sides and the neutral layer shift, the U-shaped longitudinal beam flange after roll forming exhibits nonlinear unfolding errors along its length, which easily leads to cumulative offset of the dot pitch. Simultaneously, the U-shaped cross-section experiences breathing deformations such as flange opening and rebound, and web bulging during the instant of dot impact, causing transient offset of the dot position and distortion of the pitch measurement. Furthermore, existing processes often employ integral pressing methods, which cannot specifically suppress these types of deformations.

[0003] Furthermore, the existing process directly collects the marking error data for modeling without considering the disturbance effect of cross-sectional breathing deformation, resulting in a distorted error model and a lack of reliability in subsequent error compensation strategies. Moreover, the fixed compensation range is prone to inversely triggering cross-sectional breathing deformation, forming an error amplification closed loop, which ultimately leads to a decrease in the consistency of marking in the later stage and a deterioration in the matching accuracy of the double-wing surface. At the same time, it has poor adaptability to process states such as roll pressing and roll changing, and material batch changes, making it difficult to guarantee the stability of production batches.

[0004] Currently, most related technologies employ simple pitch compensation and static benchmark control, which are no longer sufficient to meet the high-precision marking requirements of high-strength steel thin-walled longitudinal beams. Summary of the Invention

[0005] The purpose of this application is to provide a dotting production process for a CNC punching wing surface dotting production line for a longitudinal beam U-shaped beam, so as to solve the problems mentioned in the background art.

[0006] According to the first aspect of this application, a dotting production process is provided for a CNC punching wing surface dotting production line for a longitudinal beam U-shaped beam, comprising the following steps: S1: Obtain the theoretical marking parameters of the longitudinal beam to be processed, including the theoretical pitch sequence and the theoretical cumulative position sequence; S2: Set the initial dot number i=1 and set the initial compensation amount to zero; S3: Based on the theoretical cumulative position of the i-th mark in the theoretical cumulative position sequence and the current compensation amount, calculate the final mark position of the i-th mark, control the pressing unit to apply pressure to the longitudinal beam, lock the pressing unit before the punch makes the mark impact, and control the punch to make the mark at the final mark position in the locked state, and release the pressing unit after the impact is completed. S4: After the pressing unit is released, collect the servo load data of the left and right wing pressing components, calculate the load difference based on the servo load data, and compare the load difference with a preset threshold to determine whether the cross section has entered a stable state. S5: When the cross section is determined to be in a stable state, the actual position of the i-th marking point is collected, and the pitch error of the i-th marking point is calculated based on the actual marking point position and the theoretical marking point parameters. S6: Calculate the cross-sectional stability margin of the i-th marking point based on the load difference and the preset threshold; S7: Based on the pitch error of multiple points including at least the i-th point, construct or update the airfoil deployment error memory model; S8: Predict the theoretical compensation amount of the (i+1)th mark based on the wing deployment error memory model, and limit the theoretical compensation amount based on the cross-sectional stability margin of the (i)th mark to obtain the final compensation amount of the (i+1)th mark. S9: Let i = i + 1, and repeat steps S3 to S9 until all dots are marked.

[0007] Preferably, the pressure applied to the longitudinal beam by the pressure-control unit includes: setting a zoned constraint pressure according to the cross-sectional dimensions of the longitudinal beam, wherein the zoned constraint pressure includes a flange constraint pressure, a rounded corner buffer pressure, and a web stabilizing pressure, and controlling the flange pressure-controlling assembly, the rounded corner pressure-controlling assembly, and the web pressure-controlling assembly of the pressure-controlling unit to output pressure according to the flange constraint pressure, the rounded corner buffer pressure, and the web stabilizing pressure, respectively; wherein the flange constraint pressure is greater than the rounded corner buffer pressure, and the rounded corner buffer pressure is greater than or equal to the web stabilizing pressure.

[0008] Preferably, locking the pressing unit before the punch makes a dotting impact includes: determining the pre-punch locking timing parameters according to the punch position signal, and sending a locking command before the punch enters the downward phase at the time specified in the pre-punch locking timing parameters; releasing the pressing unit after the impact includes: determining the post-punch release timing parameters according to the punch position signal, and sending a release command after the punch returns to its original position at the time specified in the post-punch release timing parameters.

[0009] Preferably, the load difference is calculated according to the formula Calculation, where This refers to the servo load data of the left wing face pressure assembly. This is the servo load data for the right wing pressure assembly.

[0010] Preferably, the cross-sectional stability margin is calculated according to the formula... Calculation, where The preset threshold, The load difference is mentioned.

[0011] Preferably, the step of predicting the theoretical compensation amount of the (i+1)th marking point according to the wing deployment error memory model includes: outputting the pitch error prediction value of the (i+1)th marking point according to the wing deployment error memory model, and accumulating the pitch error prediction values ​​from the second marking point to the (i+1)th marking point and taking the negative value to obtain the theoretical compensation amount.

[0012] Preferably, the step of limiting the theoretical compensation amount based on the cross-sectional stability margin of the i-th point includes: if the cross-sectional stability margin is less than or equal to zero, the final compensation amount is zero; otherwise, the absolute value of the theoretical compensation amount is compared with the value of the cross-sectional stability margin multiplied by the preset proportional coefficient; if the absolute value of the theoretical compensation amount is greater than the product, the final compensation amount is the product multiplied by the sign of the theoretical compensation amount; otherwise, the final compensation amount is the theoretical compensation amount.

[0013] Preferably, the airfoil deployment error memory model is a Gaussian process regression model. The airfoil deployment error memory model takes the dot sequence number as input and the pitch error prediction value as output. Its kernel function is a radial basis function, and the model hyperparameters are trained based on a reliable pitch error dataset.

[0014] Preferably, the method further includes an error mutation detection and model self-update step: calculating the gradient of the pitch error between two consecutive points; if the gradient is greater than a preset mutation threshold, then weighting the historical training data of the wing deployment error memory model is reduced, and the model is retrained based on the reduced weights and the latest pitch error dataset.

[0015] In a second aspect, this application also provides a dotting production system for a longitudinal beam U-shaped CNC punching wing surface dotting production line, comprising: The parameter acquisition module is used to acquire the theoretical marking parameters of the longitudinal beam to be processed. The theoretical marking parameters include the theoretical pitch sequence and the theoretical cumulative position sequence. The initialization module is used to set the initial dot number i=1 and the initial compensation amount to zero; The dotting execution module is used to calculate the final dotting position of the i-th dotting point based on the theoretical cumulative position of the i-th dotting point in the theoretical cumulative position sequence and the current compensation amount, control the pressing unit to apply pressure to the longitudinal beam, lock the pressing unit before the punch makes the dotting impact, and control the punch to make the dotting at the final dotting position in the locked state, and release the pressing unit after the impact is completed. The cross-section state monitoring module is used to collect servo load data of the left and right wing pressing components after the pressing unit is released, calculate the load difference based on the servo load data, and compare the load difference with a preset threshold to determine whether the cross section has entered a stable state. The error acquisition module is used to acquire the actual position of the i-th marking point when the cross section enters a stable state, and to calculate the pitch error of the i-th marking point based on the actual marking point position and the theoretical marking point parameters. The margin calculation module is used to calculate the cross-sectional stability margin of the i-th point based on the load difference and the preset threshold. The model building and updating module is used to build or update the airfoil deployment error memory model based on the pitch error of multiple points, including at least the i-th point. The compensation amount prediction and limiting module is used to predict the theoretical compensation amount of the (i+1)th mark based on the wing surface deployment error memory model, and to limit the theoretical compensation amount based on the cross-sectional stability margin of the (i)th mark, so as to obtain the final compensation amount of the (i+1)th mark. The loop control module is used to set i=i+1 and repeatedly trigger the point-marking execution module, cross-section state monitoring module, error acquisition module, margin calculation module, model building and updating module, and compensation amount prediction and limiting module until all points are marked.

[0016] This application suppresses the breathing deformation of the impact transient section through partitioned stiffness constraints and temporal locking. It obtains the true airfoil deployment error through reliable sampling under stable state constraints, constructs an accurate error memory model based on a Gaussian process regression model, and achieves error compensation for coupled constraints by combining section stability margin. This forms a cross-scale coupled control system encompassing section impact propagation regulation, reliable error sampling, and stable constraint mapping compensation. This process effectively solves the problems of non-uniform deployment error, impact transient section breathing deformation, and the error amplification closed loop formed by these two issues during the CNC marking process of the U-shaped longitudinal beam airfoil. It significantly reduces the degree of section breathing deformation, ensures the stability of the error model, greatly improves the consistency of the subsequent marking and the accuracy of the double-wing symmetry matching, and simultaneously achieves adaptiveness to changes in process conditions such as roll pressing and material batch variations, improving the stability of production batches. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0018] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 A schematic diagram of the dotting production process of a CNC punching wing surface dotting production line for a longitudinal beam U-shaped structure provided in this application embodiment; Figure 2 A schematic diagram illustrating the process of calculating the final dot position and performing zone pressing and timing locking dot marking provided in this embodiment of the application; Figure 3 This is a schematic diagram of the process for collecting servo load data and determining the stability of the cross-section provided in an embodiment of this application; Figure 4 This is a schematic diagram illustrating the process of collecting actual dot positions and calculating pitch error provided in an embodiment of this application. Figure 5 This is a schematic diagram of the model structure provided in the embodiments of this application; Figure 6 This is a schematic diagram of the dotting production system of a longitudinal beam U-shaped CNC punching wing surface dotting production line provided in an embodiment of this application. Detailed Implementation

[0019] 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 some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0020] It should be noted that all user information (including but not limited to user device information, user personal information, object information corresponding to device usage data, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, device usage data, etc.) involved in all embodiments of this application are information and data authorized by the user or fully authorized by all parties.

[0021] This implementation method is suitable for the CNC marking production scenario of U-shaped longitudinal beam flanges in the precision stamping manufacturing field after automotive longitudinal beam roll forming. It can be applied to continuous marking production lines for U-shaped longitudinal beam parts, marking of high-strength steel thin-walled longitudinal beams, multi-specification mixed-line production, and marking processes for welding positioning parts with high requirements for flange pitch consistency and double-wing symmetry accuracy. As an example, the implementation conditions are a collaborative control system consisting of a CNC punching and marking equipment control system, a feeding servo system, a pressing execution unit control system, and a data processing unit. Real-time data interaction and synchronous command execution are achieved between the units. Before implementation, calibration and communication debugging of each piece of equipment must be completed to ensure the accuracy and real-time performance of parameter acquisition and command issuance. The hardware accuracy of each execution unit must match the process accuracy requirements of longitudinal beam marking.

[0022] The following detailed description, in conjunction with specific embodiments, illustrates the implementation process of the dotting production line for the U-shaped longitudinal beam CNC punching surface described in this application. It should be noted that this embodiment is merely for explaining this application and not for limiting the scope of protection of this application. Any conventional adjustments or substitutions made by those skilled in the art to the steps without departing from the concept of this application should be included within the scope of protection of this application.

[0023] like Figure 1 As shown in the figure, this application discloses a schematic diagram of the dotting production process of a CNC punching wing surface dotting production line for a longitudinal beam U-shaped beam, including the following steps: S1: Obtain the theoretical marking parameters of the longitudinal beam to be processed, including the theoretical pitch sequence and the theoretical cumulative position sequence; S2: Set the initial dot number i=1 and set the initial compensation amount to zero; S3: Based on the theoretical cumulative position of the i-th mark in the theoretical cumulative position sequence and the current compensation amount, calculate the final mark position of the i-th mark, control the pressing unit to apply pressure to the longitudinal beam, lock the pressing unit before the punch makes the mark impact, and control the punch to make the mark at the final mark position in the locked state, and release the pressing unit after the impact is completed. S4: After the pressing unit is released, collect the servo load data of the left and right wing pressing components, calculate the load difference based on the servo load data, and compare the load difference with a preset threshold to determine whether the cross section has entered a stable state. S5: When the cross section is determined to be in a stable state, the actual position of the i-th marking point is collected, and the pitch error of the i-th marking point is calculated based on the actual marking point position and the theoretical marking point parameters. S6: Calculate the cross-sectional stability margin of the i-th marking point based on the load difference and the preset threshold; S7: Based on the pitch error of multiple points including at least the i-th point, construct or update the airfoil deployment error memory model; S8: Predict the theoretical compensation amount of the (i+1)th mark based on the wing deployment error memory model, and limit the theoretical compensation amount based on the cross-sectional stability margin of the (i)th mark to obtain the final compensation amount of the (i+1)th mark. S9: Let i = i + 1, and repeat steps S3 to S9 until all dots are marked.

[0024] In some embodiments, for step S1, exemplarily, after receiving the specifications of the longitudinal beam to be processed input by the operator, the data processing unit retrieves the theoretical marking parameters matching the specifications from the pre-stored process database. The theoretical marking parameters include a theoretical pitch sequence and a theoretical cumulative position sequence. The theoretical pitch sequence is the set of theoretical distances between adjacent marking points along the length direction of the longitudinal beam flange, and the theoretical cumulative position sequence is the set of theoretical cumulative distances between each marking point on the longitudinal beam flange relative to the starting marking point position. The process database pre-stores theoretical marking parameters corresponding to longitudinal beams of different specifications. These parameters are determined during the process design stage based on the design dimensions of the longitudinal beam and the marking process requirements, and can be dynamically updated based on the actual production process optimization results. The data processing unit synchronizes the retrieved theoretical pitch sequence and theoretical cumulative position sequence to the CNC punching and marking equipment control system and the feeding servo system, and each control system stores these parameters.

[0025] In some embodiments, for step S2, specifically, before the production process starts, the data processing unit assigns the initial dotting sequence number i to 1. This sequence number is the sequential identifier for dotting on the longitudinal beam flange, and it increments sequentially as the dotting process progresses. Simultaneously, the initial compensation amount is assigned to zero. The initial compensation amount is the compensation reference for the first dotting position. Since the first dotting position serves as the position reference for all subsequent dotting positions, there is no accumulation of previous dotting errors, so no error compensation is required. The data processing unit synchronizes the initial dotting sequence number and the initial compensation amount to each control system. Each control system sets its initial working state based on these parameters, ensuring the orderly start of the first dotting process.

[0026] In some embodiments, step S3 is designed to address the problems in the prior art where overall pressing easily leads to overall cross-sectional breathing deformation and the lack of time-sequence constraints causes uncontrolled impact deformation. By using zoned stiffness constraints and time-sequence control of pre-impact locking and post-impact release, the transient cross-sectional breathing deformation during impact is suppressed while avoiding residual stress lock-up. This creates a stable structural constraint environment for precise point marking. The principle is based on the impact energy propagation characteristics of a U-shaped cross-section. Zoned pressing achieves local dissipation of impact energy, time-sequence locking forms rigid constraints at critical impact stages, and releasing constraints after impact achieves orderly release of residual stress.

[0027] like Figure 2 As shown, Figure 2 This is a schematic diagram illustrating the process of calculating the final dot position and performing zone pressing and timing locking dot marking, as provided in the embodiments of this application.

[0028] In S201, the final marker position is calculated. The data processing unit retrieves the theoretical cumulative position of the i-th marker from the theoretical cumulative position sequence. and the current compensation amount According to the formula Calculate the final position of the i-th dot. When i=1, The final marked position is the theoretical cumulative position. After the calculation is completed, the data processing unit will assign the final marked position. The data is sent to the CNC punching and marking equipment control system and the feeding servo system. The feeding servo system controls the longitudinal beam to feed the material according to the parameters, so that the position of the longitudinal beam to be marked is moved to directly below the punch, thereby achieving precise positioning of the marking position.

[0029] In S202, the partition constraint pressure is set and the material handling is executed. Based on the cross-sectional dimensions of the longitudinal beam, the data processing unit retrieves the corresponding partition constraint pressure parameters from the specification database. The partition constraint pressure includes the flange constraint pressure. Rounded corners buffer pressure Web plate stabilizing pressure The partition constraint pressure parameters pre-stored in the specification database are obtained through trial production calibration. The calibration process involves selecting longitudinal beams with different cross-sectional dimensions and materials for trial production, testing the cross-sectional deformation suppression effect under different pressure configurations, and determining the optimal pressure parameters corresponding to each specification of longitudinal beam. (Flange constraint pressure) Rounded corners buffer pressure Web plate stabilizing pressure Satisfy numerical relationship The setting of this numerical relationship is to ensure that the energy generated by the point impact is dissipated locally on the wing surface first, and to avoid the energy being transferred to the fillet and web areas, which would cause overall cross-sectional breathing deformation.

[0030] The data processing unit sends the partition constraint pressure parameters to the pressure execution unit control system. After receiving the parameters, the pressure execution unit control system controls the airfoil pressure assembly, fillet pressure assembly, and web pressure assembly of the pressure unit to respectively adjust the airfoil constraint pressure. Rounded corners buffer pressure Web plate stabilizing pressure The output pressure, the contact length and contact sequence of each pressure component with the longitudinal beam are adapted to the pressure parameters. The wing pressure component has the longest contact length and contacts the longitudinal beam first, the rounded corner pressure component has the second longest contact length and contacts the beam with a delay, and the web pressure component has the shortest contact length and contacts the beam last, ensuring that the constraint effect of the zoned pressure is matched with the energy dissipation requirements.

[0031] In S203, the pre-punch locking timing control is implemented. The CNC punching equipment control system acquires the punch position signal output by the punch encoder in real time. This signal reflects the real-time motion status of the punch, including position information such as downward movement, return stroke, and stop. The CNC punching equipment control system determines the pre-punch locking timing parameters based on the punch position signal. This parameter represents the pre-locking time before the punch enters the downward phase. It is determined by the punch's stroke curve, which is a curve showing the punch displacement over time. This curve is determined by the hardware characteristics of the CNC punching equipment and can be fine-tuned through equipment calibration. For example, the faster the punch moves, the faster the pre-locking timing parameter... The configuration value can be appropriately increased to ensure that the pressing unit completes rigid locking before the punch impacts.

[0032] The control system of the CNC punching and dotting equipment detects that the punch is about to enter the downward phase and activates it in advance. The system continuously sends locking commands to the pressure execution unit control system. After receiving the commands, the pressure execution unit control system controls each pressure component to complete rigid locking within a preset time, so that a rigid constraint is formed between the pressure component and the longitudinal beam, limiting the deformation trend of the U-shaped section at the moment of point impact. After locking, the pressure of the pressure component is maintained at the set zone constraint pressure value to ensure the stability of the constraint.

[0033] In S204, the punching operation is performed in the locked state. After the pressure unit completes rigid locking, the CNC punching and marking equipment control system confirms the position of the longitudinal beam below the punch and the final punching position. Match, then control the plunger downwards, at the final target position. The longitudinal beam flange is subjected to a dotting impact operation. The impact force and depth of the punch are set according to the material properties and process requirements of the longitudinal beam to ensure the forming quality of the dots.

[0034] In S205, the post-punch release timing control. The CNC punching and dotting equipment control system determines the post-punch release timing parameters based on the punch position signal after detecting that the punch has completed its impact and entered the return phase, and after the punch return is complete. This parameter is the delayed release time after the punch returns to its original position, which is also determined by the punch stroke curve. The CNC punching equipment control system delays the release time after the punch returns to its original position. At any time, a release command is sent to the pressure execution unit control system. After receiving the command, the pressure execution unit control system controls each pressure component to complete the pressure release. The release rate is uniform to avoid vibration of the longitudinal beam caused by rapid release. After the release is completed, each pressure component maintains the basic support pressure to ensure the positional stability of the longitudinal beam in the subsequent feeding process.

[0035] In some embodiments, for step S4, this step is to solve the problem that there is no accurate method for determining the stability state of the cross section in the prior art, and that directly collecting error data is easy to cause the data to be contaminated by breathing deformation disturbance. By collecting and analyzing servo load data, the force balance state of the U-shaped cross section is accurately quantified. The principle is that the breathing deformation and vibration of the U-shaped cross section will be directly reflected on the servo load of the left and right wing pressure components. The smaller the difference in load between the left and right wings, the more balanced the cross section is and the better the stability state.

[0036] like Figure 3 As shown, Figure 3 This is a schematic diagram illustrating the process of collecting servo load data and determining the stability of the cross-section, as provided in an embodiment of this application.

[0037] In S301, servo load data is acquired. After the pressing unit completes the punching and releasing, the pressing execution unit control system acquires the servo load data of the left and right wing pressing components in real time, which is recorded as the left wing servo load. and right wing servo load The servo load data is acquired by the torque sensor of the servo motor of the pressure assembly. The torque signal is converted into servo load data using a torque-load conversion formula, which is calibrated by the hardware characteristics of the pressure execution unit and pre-stored in the pressure execution unit control system. The acquisition frequency of the servo load data is determined by the hardware acquisition module; for example, the acquisition frequency can be configured to be no less than 100Hz to ensure that the acquired data accurately reflects the load change status of the pressure assembly. The pressure execution unit control system will then process the acquired data... and It is synchronized to the data processing unit in real time.

[0038] In S302, load difference calculation. After receiving the servo load data from the left and right wings, the data processing unit calculates the load difference according to the formula. Calculate load difference ,in The load difference of the cross section after the i-th mark is completed reflects the force balance state of the U-shaped cross section. If the cross section has breathing deformation or vibration, the force on the left and right wings will be significantly different. The value will increase; if the cross-section is in a stable state, the forces on the left and right wings tend to be in equilibrium. The value will decrease.

[0039] In S303, the cross-sectional stability state is determined. The data processing unit will calculate the load difference. With preset threshold Comparison, preset threshold The critical load difference value at which the cross-section enters a steady state is obtained through trial production calibration. The calibration process involves selecting longitudinal beams of different specifications and materials for trial production, collecting load difference data of the cross-section from deformation and vibration to a steady state during the trial production process, and taking the maximum value of the data during the steady-state stage as the threshold. The initial value. Preset threshold. It has a dynamic update mechanism, and the update method is a weighted average method, that is, a new preset threshold. ,in This is the historical weighting coefficient, with a value ranging from 0.6 to 0.8, set by process engineers based on production stability. The threshold set for the new production batch. If The data processing unit then determines that the U-shaped cross-section has entered a stable state; if If the cross section has not entered a stable state, the pressure execution unit control system continues to collect servo load data, and the data processing unit repeats the load difference calculation and state judgment steps until the cross section enters a stable state.

[0040] In some embodiments, step S5 is to address the problem in the prior art that error sampling lacks stable state constraints and the sampled data includes transient errors caused by impact deformation, which cannot reflect the true airfoil deployment error. Error data is collected only under the stable state of the cross section to ensure that the pitch error only reflects the airfoil deployment error caused by roll forming. The principle is that the breathing deformation of the cross section under impact is a short-term disturbance that will decay in a short time. The deviation of the dot position under the stable state of the cross section is the inherent deployment error of the longitudinal beam, thus eliminating the interference of transient deformation.

[0041] like Figure 4 As shown, Figure 4 This is a schematic diagram illustrating the process of acquiring the actual marking position and calculating the pitch error according to an embodiment of this application. In S401, the actual marking position is acquired. After the data processing unit determines that the cross-section has entered a stable state, the feeding servo system acquires the actual marking position of the i-th marking point through a position encoder. The position encoder's acquisition accuracy matches the hardware accuracy of the feeding servo system. For example, the acquisition accuracy can be configured to 0.01mm to ensure the accuracy of the actual marking position data. The feeding servo system will acquire... The data is synchronized to the data processing unit, which then stores it.

[0042] In S402, pitch error is calculated. The data processing unit retrieves the actual position of the i-th marking point. and the actual position of the (i-1)th dot According to the formula Calculate the actual pitch of the i-th dot. When i=1, there is no actual position data for the (i-1)th marking point, so pitch error calculation is not performed. The subsequent pitch error calculation steps begin from i=2. Then, the data processing unit retrieves the theoretical pitch of the i-th marking point from the process database. According to the formula Calculate the pitch error of the i-th dot. ,in Let be the pitch error of the i-th dot. This indicates that the actual pitch is greater than the theoretical pitch, and there is a lead offset in the marking; if This indicates that the actual pitch is less than the theoretical pitch, and there is a lag offset in the dot pattern; if This indicates that the actual pitch matches the theoretical pitch, with no offset in the markings. The data processing unit will calculate the pitch error. The data is stored according to the wing surface category to form a reliable pitch error dataset.

[0043] In S403, error mutation detection and model self-updating are performed. After calculating the pitch error, the data processing unit executes the error mutation detection step, first retrieving the pitch error of the i-th data point. Pitch error of the (i-1)th dot According to the formula Calculate the continuous error gradient , It reflects the rate of change of pitch error between two adjacent dots. If this value exceeds the preset range, it indicates that the distribution pattern of the airfoil deployment error has changed abruptly, which may be caused by changes in process conditions such as roller pressing and changing, material batch changes, and mold wear.

[0044] The data processing unit will process the continuous error gradient. Compared with the preset mutation threshold Comparison, preset mutation threshold This is the critical value at which pitch error abruptly changes; it is obtained through trial production calibration. The higher the required process precision, the better. The smaller the value, the better. Then the data processing unit triggers the self-update process of the wing deployment error memory model. First, it performs weight decay on the model's historical training data. The weight decay formula is: ,in The historical weights after decay. The historical weights before decay. The weight decay coefficient is defined as the proportion of weight decay in historical training data. It is set by process engineers based on production experience, with a value ranging from 0.5 to 0.7. Weight decay reduces the impact of historical training data on the model, making the model more adaptable to the latest process conditions. After weight decay is completed, the data processing unit retrains the airfoil deployment error memory model based on the decayed weights and the latest reliable pitch error dataset, obtaining an updated model. If the pitch error does not change abruptly, the airfoil deployment error memory model remains unchanged.

[0045] In some embodiments, step S6 is to address the problem that error compensation in the prior art does not consider the cross-sectional stability state and cannot quantify the compensation range that the cross-section can withstand. By calculating the cross-sectional stability margin, the available compensation space of the cross-section is accurately quantified. The principle is that the cross-sectional stability margin is the difference between a preset threshold and the actual load. The larger the value, the better the cross-sectional stability state and the greater the error compensation range that can be withstood.

[0046] Specifically, the data processing unit retrieves the load difference of the i-th point. and preset threshold According to the formula Calculate the cross-sectional stability margin at the i-th marking point. ,in This represents the available compensation space for the cross-section, reflecting the error compensation range that the cross-section can withstand under its current condition. If... This indicates that the cross-section is in an unstable state, and error compensation is prohibited at this time to avoid inducing breathing deformation of the cross-section during the compensation operation; if This indicates that the cross-section is in a stable state, and can be determined according to... The numerical value determines the maximum magnitude of error compensation. The data processing unit will calculate the value... The data is stored and synchronized to various control systems.

[0047] In some embodiments, step S7 is to address the problem in the prior art that there is no accurate airfoil deployment error model, which makes it impossible to effectively fit and predict the nonlinear and locally abrupt deployment error caused by roll forming. The airfoil deployment error memory model constructed based on the Gaussian process regression model can accurately fit the nonlinear deployment error distribution of the airfoil along the length direction. It is stable for small sample data and can be updated with batches. The principle is that the Gaussian process regression model is a nonparametric regression model, which does not require a preset function form and can adaptively fit the nonlinear data distribution. The radial basis function kernel can effectively capture the local abrupt change characteristics of the data and adapt to the distribution characteristics of the airfoil deployment error.

[0048] The wing deployment error memory model employs a Gaussian process regression model. This model takes the point number as input and the predicted pitch error value as output, such as... Figure 5 As shown, Figure 5 This is a schematic diagram of the model structure provided in an embodiment of this application. The complete structure of the model includes an input layer, a kernel function layer, a regression layer, and a variance layer. These modules are connected sequentially. The output of the input layer serves as the input to the kernel function layer, and the output of the kernel function layer serves as the input to both the regression and variance layers. The regression and variance layers output results in parallel. The core function of the input layer is to standardize the input dot index i to eliminate the influence of dimensions. The standardization formula is: ,in This is the average of the dot numbers. The standard deviation of the dot index; the kernel layer uses radial basis functions as kernel functions, and the expression of the radial basis functions is: ,in For kernel function values, For signal variance, For length scale, and These are the hyperparameters of the model; the core function of the regression layer is to calculate the pitch error prediction value based on the posterior probability distribution of a Gaussian process, and output the pitch error prediction value of the i-th point. The core function of the variance layer is to calculate the uncertainty of the model's predictions and output the prediction uncertainty at the i-th point. , The smaller the value, the higher the reliability of the model's prediction results.

[0049] When the production process is executed for the first time and there is no historical reliable pitch error dataset, the data processing unit constructs an initial reliable pitch error dataset based on the pitch error of multiple points, including at least the i-th point, collected at the moment. The dataset is then divided into a training set and a validation set in a 7:3 ratio. The training set is used for learning the model's parameters, and the validation set is used for verifying the model's generalization ability.

[0050] The data processing unit first processes the hyperparameters of the kernel function. and Perform initialization, for example, The initial value can be configured to 1. The initial value can be configured to 1. Then, the standardized marker numbers from the training set are input into the model. The covariance matrix is ​​calculated through a kernel function layer, and the error prediction values ​​are fitted based on the posterior probability distribution formula of a Gaussian process. The hyperparameters are optimized using maximum likelihood estimation, by maximizing the log-likelihood function of the training set to find the optimal hyperparameters. The standardized marker numbers from the validation set are input into the optimized model, and the mean square error between the predicted values ​​and the actual pitch error on the validation set is calculated. If the mean square error meets the preset accuracy requirements, the model training is complete. If not, the initial hyperparameter values ​​are readjusted, and the fitting and optimization steps are repeated until the accuracy requirements are met. After training, the data processing unit saves the model structure and optimal hyperparameters to the model library, completing the initial construction of the wing deployment error memory model.

[0051] As the production process continues and the sample size of the reliable pitch error dataset increases, or when the model self-update process after error mutation detection is triggered, the data processing unit updates the airfoil deployment error memory model based on the newly added pitch error data or the latest reliable pitch error dataset. The model update uses incremental training, adding new data to the existing training set and re-executing hyperparameter optimization and model fitting steps. This eliminates the need for retraining the model, significantly improving the efficiency of model updates and ensuring that the model always accurately fits the latest airfoil deployment error distribution.

[0052] In some embodiments, step S8 addresses the problem in the prior art where the error compensation amplitude is fixed, and excessive compensation amplitude can easily trigger cross-sectional breathing deformation, forming an error amplification closed loop. By predicting the theoretical compensation amount through the wing deployment error memory model and limiting the compensation amplitude based on the cross-sectional stability margin, the coupling control of error compensation and cross-sectional stability is achieved. This effectively compensates for the accumulated deployment error while avoiding triggering cross-sectional breathing deformation. The principle is to introduce structural stability into the constraint condition of error compensation, so that the compensation amplitude matches the stable state of the cross-section, achieving synergy between macroscopic error compensation and microscopic deformation suppression.

[0053] The data processing unit inputs the index of the (i+1)th marker into the airfoil deployment error memory model, and the model outputs the predicted pitch error value of the (i+1)th marker. The data processing unit then accumulates the predicted pitch error values ​​from the second to the (i+1)th marker, according to the formula... Calculate the theoretical compensation amount for the (i+1)th marking point. This calculation method uses an integral mapping approach, which can effectively compensate for the cumulative unfolding error along the length direction and is adapted to the cumulative offset characteristics of the unfolding error of the wing surface after roll forming.

[0054] The data processing unit first determines the cross-sectional stability margin of the i-th marking point, thus determining the compensation limit and final compensation amount. ,like Then set the final compensation amount. Error compensation is prohibited; if Then, the following limiting operation is performed to determine the preset proportional coefficient k. This coefficient is defined as the ratio between the cross-sectional stability margin and the maximum tolerable compensation range. It is used to convert the cross-sectional stability margin into the maximum tolerable compensation range and is obtained through trial production calibration. The calibration process involves selecting longitudinal beams of different specifications to complete trial production, gradually increasing the compensation range under different cross-sectional stability margins, recording the critical compensation range at which the cross-section changes from stable to unstable, calculating the ratio of the critical compensation range to the cross-sectional stability margin, and taking the average of the ratios of multiple samples as the initial value of k. The value of k ranges from 0.8 to 1.0. For example, for high-strength steel thin-walled longitudinal beams, k can be configured as 0.8 to reserve a compensation margin; for ordinary steel thick-walled longitudinal beams, k can be configured as 1.0 to achieve maximum error compensation. The preset proportional coefficient k has a dynamic update mechanism, which can be manually adjusted by process engineers according to the cross-sectional stability and dotting accuracy during the production process, or automatically optimized by the data processing unit based on process data.

[0055] The data processing unit retrieves the cross-sectional stability margin of the i-th marker. And a preset proportional coefficient k, calculate k with The product of these, followed by the theoretical compensation amount. The absolute value of the product is compared with the product. If This indicates that the theoretical compensation amount exceeds the maximum compensation range that the cross-section can withstand under the current state. Therefore, the theoretical compensation amount is limited, resulting in the final compensation amount. ,in For a sign function, when hour, ,when hour, This limiting method ensures that the direction of the final compensation is consistent with the theoretical compensation, and the amplitude is the maximum amplitude that the cross-section can withstand; if This indicates that the theoretical compensation amount is within the range that the cross-section can withstand, and no amplitude limiting is required; the final compensation amount... The data processing unit will calculate the final compensation amount. The data is stored and synchronized to each control system to provide compensation parameters for calculating the final position of the next data point.

[0056] In some embodiments, for step S9, specifically, the data processing unit increments the dot number i by 1 to obtain a new dot number i+1, and then each control system performs a compensation based on the new dot number and the final compensation amount of the (i+1)th dot. Returning to the final dot position calculation step in the third step of this embodiment, the subsequent steps are executed sequentially, including zoned material pressing and timing-locked dot marking, cross-sectional stability judgment, pitch error calculation, cross-sectional stability margin calculation, model construction or update, compensation prediction and amplitude limiting, forming a closed-loop dot marking control process. The above steps are continuously executed in a loop until the data processing unit determines that all dot marking processes on the longitudinal beam flange are completed, at which point the loop terminates.

[0057] Therefore, this implementation method suppresses the breathing deformation of the impact transient section through partitioned stiffness constraints and time-sequence locking, obtains the true airfoil deployment error through reliable sampling under stable state constraints, constructs an accurate error memory model based on a Gaussian process regression model, and achieves error compensation for coupled constraints by combining section stability margin, thus forming a cross-scale coupled control system of section impact propagation regulation, reliable error sampling, and stability constraint mapping compensation. This process effectively solves the problems of non-uniform deployment error, impact transient section breathing deformation, and the error amplification closed loop formed by the two in the CNC marking process of the U-shaped longitudinal beam airfoil, significantly reduces the degree of section breathing deformation, ensures the stability of the error model, greatly improves the consistency of the subsequent marking and the accuracy of the double-wing symmetry matching, and achieves adaptiveness to changes in process states such as roll pressing and roll changing and material batch changes, thereby improving the stability of production batches.

[0058] It should be noted that although the operations of the method of this application are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. On the contrary, the steps depicted in the flowchart can be performed in a different order. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0059] Please see Figure 6 , Figure 6 This application provides a structural block diagram of a dotting production system for a CNC punching flange surface dotting production line for a longitudinal beam U-shaped structure. The system specifically includes: The parameter acquisition module 601 is used to acquire the theoretical marking parameters of the longitudinal beam to be processed, the theoretical marking parameters including the theoretical pitch sequence and the theoretical cumulative position sequence; The initialization module 602 is used to set the initial dot sequence number i=1 and the initial compensation amount to zero; The dotting execution module 603 is used to calculate the final dotting position of the i-th dotting point based on the theoretical cumulative position of the i-th dotting point in the theoretical cumulative position sequence and the current compensation amount, control the pressing unit to apply pressure to the longitudinal beam, lock the pressing unit before the punch makes the dotting impact, and control the punch to make the dotting at the final dotting position in the locked state, and release the pressing unit after the impact is completed. The cross-section state monitoring module 604 is used to collect servo load data of the left and right wing pressing components after the pressing unit is released, calculate the load difference based on the servo load data, and compare the load difference with a preset threshold to determine whether the cross section has entered a stable state. The error acquisition module 605 is used to acquire the actual position of the i-th marking point when the cross section is determined to be in a stable state, and to calculate the pitch error of the i-th marking point based on the actual marking point position and the theoretical marking point parameters. Margin calculation module 606 is used to calculate the cross-sectional stability margin of the i-th point based on the load difference and the preset threshold. The model building and updating module 607 is used to build or update the airfoil deployment error memory model based on the pitch error of multiple points, including at least the i-th point. The compensation amount prediction and limiting module 608 is used to predict the theoretical compensation amount of the (i+1)th mark based on the wing surface deployment error memory model, and to limit the theoretical compensation amount based on the cross-sectional stability margin of the (i)th mark, so as to obtain the final compensation amount of the (i+1)th mark. The loop control module 609 is used to set i=i+1 and repeatedly trigger the point-marking execution module, cross-section state monitoring module, error acquisition module, margin calculation module, model construction and update module, and compensation amount prediction and limiting module until all points are marked.

[0060] It should be noted that the working process of each module in the dotting production system of the longitudinal beam U-shaped CNC punching wing surface dotting production line described in this embodiment can refer to the working process of the dotting production process of the longitudinal beam U-shaped CNC punching wing surface dotting production line described in the above embodiment. The technical effect achieved is also the same as the dotting production process of the longitudinal beam U-shaped CNC punching wing surface dotting production line described in the above embodiment, and will not be repeated here.

[0061] The above description represents the preferred embodiments of the present invention. It should be noted that, for those skilled in the art, various improvements and modifications can be made without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A dotting production process for a CNC punching wing surface dotting production line for a longitudinal beam U-shaped structure, characterized in that, Includes the following steps: S1: Obtain the theoretical marking parameters of the longitudinal beam to be processed, including the theoretical pitch sequence and the theoretical cumulative position sequence; S2: Set the initial dot number i=1 and set the initial compensation amount to zero; S3: Based on the theoretical cumulative position of the i-th mark in the theoretical cumulative position sequence and the current compensation amount, calculate the final mark position of the i-th mark, control the pressing unit to apply pressure to the longitudinal beam, lock the pressing unit before the punch makes the mark impact, and control the punch to make the mark at the final mark position in the locked state, and release the pressing unit after the impact is completed. S4: After the pressing unit is released, collect the servo load data of the left and right wing pressing components, calculate the load difference based on the servo load data, and compare the load difference with a preset threshold to determine whether the cross section has entered a stable state. S5: When the cross section is determined to be in a stable state, the actual position of the i-th marking point is collected, and the pitch error of the i-th marking point is calculated based on the actual marking point position and the theoretical marking point parameters. S6: Calculate the cross-sectional stability margin of the i-th marking point based on the load difference and the preset threshold; S7: Construct or update the airfoil deployment error memory model based on the pitch error of multiple points, including at least the i-th point. S8: Predict the theoretical compensation amount of the (i+1)th mark based on the wing deployment error memory model, and limit the theoretical compensation amount based on the cross-sectional stability margin of the (i)th mark to obtain the final compensation amount of the (i+1)th mark. S9: Let i = i + 1, and repeat steps S3 to S9 until all dots are marked.

2. The dotting production process of the CNC punching wing surface dotting production line for a longitudinal beam U-shaped according to claim 1, characterized in that, The pressure applied to the longitudinal beam by the pressure control unit includes: setting zoned constraint pressure according to the cross-sectional dimensions of the longitudinal beam, wherein the zoned constraint pressure includes wing constraint pressure, rounded corner buffer pressure and web stabilizing pressure, and controlling the wing pressure assembly, rounded corner pressure assembly and web pressure assembly of the pressure control unit to output pressure according to the wing constraint pressure, rounded corner buffer pressure and web stabilizing pressure respectively; wherein the wing constraint pressure is greater than the rounded corner buffer pressure, and the rounded corner buffer pressure is greater than or equal to the web stabilizing pressure.

3. The dotting production process of a CNC punching wing surface dotting production line for a longitudinal beam U-shaped beam according to claim 2, characterized in that, Locking the pressing unit before the punch makes a dotting impact includes: determining the pre-punch locking timing parameters according to the punch position signal, and sending a locking command before the punch enters the downward phase at the time specified in the pre-punch locking timing parameters; releasing the pressing unit after the impact includes: determining the post-punch release timing parameters according to the punch position signal, and sending a release command after the punch returns to its original position at the time specified in the post-punch release timing parameters.

4. The dotting production process of a CNC punching wing surface dotting production line for a longitudinal beam U-shaped beam according to claim 1, characterized in that, The load difference is calculated according to the formula Calculation, where This refers to the servo load data of the left wing face pressure assembly. This is the servo load data for the right wing pressure assembly.

5. The dotting production process of a CNC punching wing surface dotting production line for a longitudinal beam U-shaped beam according to claim 1, characterized in that, The cross-sectional stability margin is calculated according to the formula. Calculation, where The preset threshold, The load difference is mentioned.

6. The dotting production process of a CNC punching wing surface dotting production line for a longitudinal beam U-shaped beam according to claim 1, characterized in that, The step of predicting the theoretical compensation amount for the (i+1)th marking point based on the wing deployment error memory model includes: outputting the pitch error prediction value for the (i+1)th marking point based on the wing deployment error memory model, and summing the pitch error prediction values ​​from the 2nd marking point to the (i+1)th marking point and taking the negative value to obtain the theoretical compensation amount.

7. The dotting production process of a CNC punching wing surface dotting production line for a longitudinal beam U-shaped beam according to claim 1, characterized in that, The step of limiting the theoretical compensation amount based on the cross-sectional stability margin of the i-th point includes: if the cross-sectional stability margin is less than or equal to zero, the final compensation amount is zero; otherwise, the absolute value of the theoretical compensation amount is compared with the value of the cross-sectional stability margin multiplied by the preset proportional coefficient. If the absolute value of the theoretical compensation amount is greater than the product, the final compensation amount is the product multiplied by the sign of the theoretical compensation amount; otherwise, the final compensation amount is the theoretical compensation amount.

8. The dotting production process of a CNC punching wing surface dotting production line for a longitudinal beam U-shaped beam according to claim 1, characterized in that, The airfoil deployment error memory model is a Gaussian process regression model. The airfoil deployment error memory model takes the dot sequence number as input and the pitch error prediction value as output. Its kernel function is a radial basis function, and the model hyperparameters are trained based on a reliable pitch error dataset.

9. The dotting production process of a CNC punching wing surface dotting production line for a longitudinal beam U-shaped beam according to claim 1, characterized in that, It also includes error mutation detection and model self-updating steps: calculating the gradient of the pitch error between two consecutive points; if the gradient is greater than a preset mutation threshold, then the historical training data of the wing deployment error memory model is weighted and decayed, and the model is retrained based on the decayed weights and the latest pitch error dataset.

10. A dotting production system for a CNC punching wing surface dotting production line for a longitudinal beam U-shaped structure, characterized in that, include: The parameter acquisition module is used to acquire the theoretical marking parameters of the longitudinal beam to be processed. The theoretical marking parameters include the theoretical pitch sequence and the theoretical cumulative position sequence. The initialization module is used to set the initial dot number i=1 and the initial compensation amount to zero; The dotting execution module is used to calculate the final dotting position of the i-th dotting point based on the theoretical cumulative position of the i-th dotting point in the theoretical cumulative position sequence and the current compensation amount, control the pressing unit to apply pressure to the longitudinal beam, lock the pressing unit before the punch makes the dotting impact, and control the punch to make the dotting at the final dotting position in the locked state, and release the pressing unit after the impact is completed. The cross-section state monitoring module is used to collect servo load data of the left and right wing pressing components after the pressing unit is released, calculate the load difference based on the servo load data, and compare the load difference with a preset threshold to determine whether the cross section has entered a stable state. The error acquisition module is used to acquire the actual position of the i-th marking point when the cross section enters a stable state, and to calculate the pitch error of the i-th marking point based on the actual marking point position and the theoretical marking point parameters. The margin calculation module is used to calculate the cross-sectional stability margin of the i-th point based on the load difference and the preset threshold. The model building and updating module is used to build or update the airfoil deployment error memory model based on the pitch error of multiple points, including at least the i-th point. The compensation amount prediction and limiting module is used to predict the theoretical compensation amount of the (i+1)th mark based on the wing surface deployment error memory model, and to limit the theoretical compensation amount based on the cross-sectional stability margin of the (i)th mark, so as to obtain the final compensation amount of the (i+1)th mark. The loop control module is used to set i=i+1 and repeatedly trigger the point-marking execution module, cross-section state monitoring module, error acquisition module, margin calculation module, model building and updating module, and compensation amount prediction and limiting module until all points are marked.

Images