A collaborative control and intelligent monitoring method for a multi-process automated production line

Abstract

The application provides a kind of multi-process automation production line's synergic control and intelligent monitoring method, comprising: in step, transfer processing material, collect surface image parameter and identify whether it conforms to set tolerance;Extract initial forming deviation value within set tolerance range, record the demolding time node of processing material leaving corresponding processing position;Track flow track and record arrival time node when reaching downstream assembly or welding step;Calculate flow residence duration according to demolding and arrival time node;Initial forming deviation value and flow residence duration are input into preset stress deformation decay mapping model, to predict the real-time micro-drift value generated by processing material in duration;Micro-drift value is converted into feedforward compensation instruction, to make downstream control node dynamically adjust mechanical parameters of assembly step or thermal energy parameters of welding step to counteract real-time micro-drift value. The application can neutralize material aging deformation deviation and eliminate assembly interference.

Description

Technical Field

[0001] This invention relates to the field of industrial automation manufacturing technology, and in particular to a collaborative control and intelligent monitoring method for multi-process automated production lines. Background Technology

[0002] In current multi-process automated production lines, materials need to be transferred sequentially between different physical machines to complete the continuous manufacturing process. The initial forming or cutting processes apply mechanical loads and thermodynamic influences to the material, resulting in complex residual stress distributions and temperature gradients within the material.

[0003] After the processed material completes its initial processing and leaves the physical constraints of the mold before entering the conveying device, the residual stress accumulated inside will be gradually released during the flow process over time. This stress relaxation evolution process will cause microscopic deformation of the processed material in the spatial dimension and dynamic displacement of its outer contour dimensions. The dwell time of different batches of materials between processes objectively varies, and the cumulative dimensional drift caused by aging will also be different.

[0004] Currently, downstream processing equipment generally uses pre-set static basic operating parameters to perform assembly or welding operations. The underlying control system based on fixed parameters lacks the means to synchronously acquire and adaptively feedforward compensation for the dynamic dimensional evolution of incoming materials. When processed materials with microscopic dimensional variations are transported to subsequent stages, the static processing parameters struggle to adaptively match the actual geometric and physical state of the material, leading to mechanical interference and assembly misalignment between subsequent components. Summary of the Invention

[0005] The purpose of this invention is to provide a collaborative control and intelligent monitoring method for multi-process automated production lines, in order to solve the technical problem that existing downstream processing equipment cannot adaptively match the micro-dimensional deviations caused by the release of internal residual stress during the flow and residence of processed materials when performing actions based on static basic operating parameters, which easily leads to assembly interference and fit deviations.

[0006] This invention provides a collaborative control and intelligent monitoring method for multi-process automated production lines, applicable to multi-process automated production lines including molding, cutting, assembly, welding, inspection, and packaging steps, and equipped with an image processing system and an intelligent monitoring system; the method includes the following steps:

[0007] The processed material is transferred between each of the steps, and the surface image parameters of the processed material are acquired by the image processing system. The intelligent monitoring system is used to identify whether the processed material meets the set tolerance range.

[0008] The method also includes a collaborative step:

[0009] In the molding step or the cutting step, the initial molding deviation value within the set tolerance range is extracted, and the demolding time node when the processed material leaves the corresponding processing position is recorded;

[0010] Track the flow trajectory of the processed material, and record the arrival time node when the processed material arrives at the corresponding processing position of the assembly step or the welding step;

[0011] The transit dwell time is calculated based on the demolding time node and the arrival time node;

[0012] The initial forming deviation value and the flow dwell time are input into a preset stress deformation attenuation mapping model to predict the real-time micro-drift value accumulated by the processed material during the flow dwell time.

[0013] The real-time micro-drift value is converted into a feedforward compensation command and sent to the downstream control node;

[0014] The downstream control node dynamically adjusts the mechanical execution parameters in the assembly step or the thermal output parameters in the welding step according to the feedforward compensation command, thereby counteracting the real-time micro-drift value during the processing operation.

[0015] Optionally, the molding step includes insert injection molding; in the molding step, the image processing system is used to monitor the pre-molding status of the insert.

[0016] By comparing the actual position coordinates of the insert with the reference position coordinates, when the position deviation is within the allowable limit, the local heating rod array set inside the mold is controlled to perform adaptive heating action to maintain the temperature of the local area within the set target range.

[0017] The cutting step includes a precision stamping operation; in the cutting step, a pre-set pre-springback compensation angle is obtained, and the actual springback amount is obtained by real-time measurement by a sensor;

[0018] With the help of a closed-loop feeding mechanism, error correction actions are performed based on the pre-springback compensation angle and the actual springback amount measured in real time, so as to maintain the macroscopic dimensions in a stable state.

[0019] Optionally, the precision stamping operation encompasses a multi-step composite feature manufacturing process; the multi-step composite feature manufacturing process includes:

[0020] The continuous manufacturing process is broken down into pre-bending, step-by-step rolling, and finishing and sizing processes.

[0021] In each sub-step of the step-by-step rolling action, an independently assigned compensation overpressure value is applied to the corresponding step corner.

[0022] Based on the number of workstations determined by simulation analysis, the variation in processing diameter between adjacent workstations is constrained within the set step gradient range to ensure that drawing and forming are completed before the material hardens.

[0023] Optionally, the detection step includes:

[0024] The image processing system is used to perform a sharp corner contact height check and appearance defect screening on the final composite component after it has undergone the assembly step and / or the welding step.

[0025] The verification results are compared with the preset standard template; when it is determined that there is an appearance defect or the height dimension exceeds the tolerance boundary, an abnormal trigger signal is sent to the central control terminal, and the rejection sorting branch is controlled to move the corresponding component out of the normal flow track.

[0026] Optionally, the packaging step includes:

[0027] When the detection step determines that the final synthesized component meets the factory standards, the final synthesized component is imported into the automatic packaging module.

[0028] The automatic packaging module is controlled to perform a heat-sealing and film-coating action on the final synthesized component.

[0029] A product manufacturing history data file is created, which includes the surface image parameters, the real-time micro-drift values, and the corresponding processing trajectory data. A data identification label carrying a link to the product manufacturing history data file is printed and affixed to the packaging surface.

[0030] Optionally, the process of constructing the preset stress deformation attenuation mapping model includes:

[0031] Collect sample data of springback deformation caused by residual heat and residual mechanical stress release in the processed materials of the historical processing batches after molding;

[0032] A multivariate mapping algorithm is established with initial deviation and turnover time as independent variables and actual deformation as dependent variable;

[0033] The multivariate mapping algorithm is used to form a mathematical relationship matrix that characterizes the gradual decay of internal stress in a material over time, and the mathematical relationship matrix is ​​solidified into the preset stress deformation decay mapping model.

[0034] Optionally, the step of dynamically adjusting the mechanical actuation parameters in the assembly step includes:

[0035] When the feedforward compensation command is directed to the assembly step, the pre-tightening compensation force value used to offset the real-time micro-drift value is parsed from the feedforward compensation command.

[0036] Update the pre-stored basic assembly downforce value in the control register to the pre-tightening compensation force value;

[0037] The servo drive mechanism applies mechanical pressing action to the processed material according to the updated pre-tightening compensation force value, causing the processed material to undergo reverse mechanical deformation, neutralizing the real-time micro-drift value, and effectively preventing assembly interference.

[0038] Optionally, the step of dynamically adjusting the heat output parameters in the welding step includes:

[0039] When the feedforward compensation command is directed to the welding step, the change in the contact surface gap width corresponding to the real-time micro-drift value is determined.

[0040] Based on the change in the width of the contact surface gap, calculate the heat capacity compensation required to maintain the stability of the molten pool.

[0041] Adjusting the basic heating power or pulse discharge duration of the welding equipment to match the actual output heat energy with the required heat capacity compensation, and changing the shrinkage tension generated during the solidification and cooling stage of the weld joint, thereby offsetting the real-time micro-drift value.

[0042] Optionally, the method further includes a closed-loop iterative optimization step for the feedforward compensation command:

[0043] Collect the final physical dimension error data output by the detection steps;

[0044] Analyze the correlation and fluctuation trend between the final inspection physical size error data and the real-time microscopic drift value;

[0045] By utilizing the aforementioned correlation fluctuation trend, a data optimization algorithm is used to reverse-correct the node weight coefficients within the preset stress deformation attenuation mapping model, thereby improving the prediction accuracy of subsequent processing batches.

[0046] Optionally, the coordination steps may further include an interception and blocking mechanism:

[0047] The absolute value of the real-time micro-drift value is compared with the maximum mechanical tolerance boundary that the downstream processing position can accommodate.

[0048] When the real-time micro-drift value exceeds the maximum mechanical tolerance boundary, it is determined that the processed material has undergone irreversible deformation beyond the limit due to excessive flow and residence time.

[0049] The pre-assigned flow code corresponding to the processed material is marked as an abnormal isolation state, and the lateral cylinder is controlled to blow the corresponding material away from the main conveyor belt before the processed material enters the processing position.

[0050] The present invention has achieved the following beneficial effects:

[0051] This invention synchronously collects the time nodes of material demolding and arrival at downstream machines, calculates the dwell time, and uses this time variable, along with the initial forming deviation value, to deduce the real-time microscopic drift value of the material caused by the accumulation of internal residual stress in a stress deformation attenuation mapping model. Based on this, it converts the data into feedforward compensation commands and sends them to the downstream execution machines. Upon receiving the commands, the underlying control nodes responsible for assembly or welding update their operating parameters accordingly, adjusting the mechanical preload or changing the heat output to execute the corresponding process actions based on compensation requirements. This feedforward collaboration strategy enables downstream production equipment to autonomously correct errors based on the actual material flow time. By applying reverse physical deformation at the end, it offsets the accumulated microscopic outward deformation from the previous flow, improving the adaptability of traditional processing machines that rely on static single-parameter operation. It reduces interference in physical fits and fluctuations in welding gaps caused by microscopic dimensional variations, ensuring the component connection accuracy and finished product assembly quality of continuous processing operations in multi-process automated production lines.

[0052] Other features and advantages of the invention will be set forth in the description which follows, 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 technical solutions particularly pointed out in the written description and the accompanying drawings.

[0053] 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

[0054] 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:

[0055] Figure 1 A schematic diagram of the system module composition structure of a multi-process automated production line provided in an embodiment of the present invention;

[0056] Figure 2 The main flowchart of the collaborative control and intelligent monitoring method for multi-process automated production lines provided in this embodiment of the invention;

[0057] Figure 3 A flowchart illustrating the construction process of a preset stress deformation attenuation mapping model provided in this embodiment of the invention;

[0058] Figure 4 A flowchart illustrating the dynamic adjustment of assembly and welding steps execution parameters provided in this embodiment of the invention;

[0059] Figure 5 A flowchart of the detection steps and automatic packaging steps provided in the embodiments of the present invention;

[0060] Figure 6 The flowchart illustrates the closed-loop iterative optimization steps and interception mechanism provided in this embodiment of the invention. Detailed Implementation

[0061] 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.

[0062] In the processing and transfer operations of multi-process automated production lines, the processed materials undergo spatial displacement sequentially between various physical machines. Due to the mechanical or thermodynamic effects of processing operations such as forming and cutting, residual stress distribution and temperature gradients are formed within the processed materials. When the processed materials leave the physical constraints of the mold and enter the conveying device, the residual stress within them is gradually released over time, causing microscopic deformation and dimensional shifts in the spatial dimension. If downstream processing equipment operates using preset static basic parameters, the dimensional variations caused by stress release will lead to misalignment of downstream components, mechanical interference, or changes in welding gaps. Therefore, this application provides a collaborative control and intelligent monitoring method for multi-process automated production lines. It calculates the microscopic dimensional drift using a transfer time parameter and a stress deformation attenuation mapping model, and performs dynamic feedforward parameter compensation operations in downstream control nodes.

[0063] This application discloses a collaborative control and intelligent monitoring method for multi-process automated production lines, such as... Figure 1 As shown, the equipment is used in a multi-process automated production line that includes molding, cutting, assembly, welding, inspection and packaging steps. The equipment is equipped with an image processing system and an intelligent monitoring system.

[0064] like Figure 2 As shown, the method includes the following steps:

[0065] Step S10: Transfer the processed material between each of the steps, and collect the surface image parameters of the processed material through the image processing system, and use the intelligent monitoring system to identify whether the processed material meets the set tolerance range.

[0066] Specifically, the multi-process automated production line includes a servo mainline conveyor mechanism, bypass diversion guide rails, and multi-axis transfer equipment in its physical hardware connections. Before the processed material leaves the corresponding processing positions of the forming and cutting steps, and enters the corresponding processing positions of the assembly and welding steps, image acquisition hardware is distributed to construct the image processing system. The image acquisition hardware includes an area array industrial camera and a corresponding programmable multi-band light source module. When the processed material passes through a photoelectric through-beam sensor, the sensor outputs a high-level trigger signal to the underlying programmable controller, which sends an exposure synchronization pulse to the image acquisition hardware. Upon receiving the synchronization pulse, the area array industrial camera opens its global shutter to perform image acquisition, acquiring the structured light reflection information and two-dimensional grayscale gradient information of the surface of the moving processed material, converting them into a two-dimensional digital image matrix, and establishing the two-dimensional digital image matrix as the surface image parameters. The image processing system receives the surface image parameters through an industrial communication network.

[0067] Furthermore, the image processing core of the intelligent monitoring system performs Gaussian smoothing filtering and edge gradient extraction on the surface image parameters to obtain a set of sub-pixel coordinate point clouds representing the spatial outer contour boundary and key structural hole features of the processed material. Based on the pre-stored internal and external calibration parameter matrices of the camera, a three-dimensional coordinate transformation logic is executed to convert the sub-pixel coordinate point cloud set from the image pixel coordinate system to a three-dimensional physical space metric coordinate system, calculating the actual geometric measurement values ​​corresponding to each measurement point of the currently processed material. Specifically, the Gaussian smoothing filtering operation uses a two-dimensional Gaussian convolution kernel matrix with a size of 5×5 pixels and a Gaussian distribution standard deviation... The preferred value is 1.2; the edge gradient extraction operation uses the Canny algorithm, with the high threshold parameter of the hysteresis dual threshold judgment logic set to 150 and the low threshold parameter set to 50. The camera's internal and external calibration parameter matrices are obtained offline based on Zhang Zhengyou's chessboard calibration method. The camera's internal calibration parameter matrix includes the equivalent focal length on the X-axis, the equivalent focal length on the Y-axis, and the coordinates of the optical principal point; the external calibration parameter matrix includes a 3×3 rotation matrix representing three-dimensional spatial rotation and a 3×1 translation vector representing spatial translation. The mapping relationship of the three-dimensional coordinate transformation logic is derived from the projection equation. Established. Among them, These are sub-pixel coordinates in the image pixel coordinate system. For depth scale factor, To calibrate the parameter matrix inside the camera, For the external calibration parameter matrix, This is to calculate the actual geometric measurement values ​​in the output three-dimensional physical space metric coordinate system. Among them, , These are the subpixel horizontal and vertical coordinates in the image pixel coordinate system, respectively; The transpose operator is used to represent a matrix or vector. To characterize rotation in three-dimensional space Rotation matrix components; To characterize spatial translation Translation vector components; , , These represent the X, Y, and Z axis physical coordinate components of the actual geometric measurement value in a three-dimensional physical space measurement coordinate system. Specifically, this refers to the camera's internal calibration parameter matrix. The data structure is defined as The upper triangular constant matrix. The elements in the first row and first column of its main diagonal, and the second row and second column, are the effective focal length parameters (in pixels) of the X and Y axes calculated through checkerboard corner feature extraction, respectively. The elements in the first row and third column, and the second row and third column, are the absolute offset constants of the pixel coordinates of the optical principal point. External calibration parameter matrix. Composition The floating-point numerical matrix, whose internal matrix parameters are obtained by using the Singular Value Decomposition (SVD) algorithm to obtain the least-squares solution for the corresponding three-dimensional point sets in the calibration board's world coordinate system and the camera sensor coordinate system. The intelligent monitoring system's data storage area pre-stores a set dimensional tolerance threshold, i.e., the set tolerance range. Specifically, the set tolerance range includes the maximum permissible positive deviation dimensional boundary and the maximum permissible negative deviation dimensional boundary. Specifically, the numerical boundaries of the set tolerance range are obtained by the intelligent monitoring system through parsing the three-dimensional computer-aided design (CAD) file of the processed material. The system extracts the nominal dimensional values ​​of the corresponding geometric measurement point coordinates from the file. Upper deviation constant With lower deviation constant The system thus establishes the maximum permissible positive deviation dimension boundary value as follows: The maximum permissible negative deviation dimension boundary value is Under microscopic drift monitoring conditions, and The range of values ​​is statically constrained in to Within the closed interval. The intelligent monitoring system compares the actual geometric measurement value with the numerical boundary of the set tolerance range; if the actual geometric measurement value is within the numerical boundary range, the intelligent monitoring system assigns a qualified attribute flag bit to the status register of the processed material and outputs a flow control signal to drive the main conveyor mechanism to continue conveying; if the actual geometric measurement value exceeds the set numerical boundary range, the intelligent monitoring system outputs an action trigger signal to control the extension of the diversion cylinder deployed on the side of the flow path, removing the processed material with non-compliant dimensions from the main flow path.

[0068] The method also includes a collaborative step:

[0069] Step S20: In the molding step or the cutting step, extract the initial molding deviation value within the set tolerance range, and record the demolding time node when the processed material leaves the corresponding processing position.

[0070] Understandably, for processed materials identified as conforming to the set tolerance range through previous steps, the intelligent monitoring system performs a spatial vector subtraction operation between the actual geometric measurement values ​​and the nominal design reference values. From the calculation result, it extracts the difference variable and directional feature vector located within the tolerance range, and defines this data column as the initial forming deviation value. The initial forming deviation value is used to quantify the microstructural morphology of the processed material at the moment it leaves the mold surface.

[0071] Both the forming main controller executing the forming step and the punching main controller executing the cutting step are connected to an industrial local area network equipped with a time synchronization protocol mechanism to achieve alignment of the system's global underlying hardware clock. When the mold performs an ejection action or the punch press slide returns to a set safe position coordinate, the position sensor in the relevant controller generates a state switching command. The intelligent monitoring system intercepts this state switching command and extracts the global synchronization clock data sequence within the corresponding hardware scan cycle. After formatting the clock data sequence, it establishes it as the demolding time node. The intelligent monitoring system generates a unique flow code for the current processed material and writes the flow code, the demolding time node, and the initial forming deviation value into the tracking database field of the system server.

[0072] Step S30: Track the flow trajectory of the processed material and record the arrival time node when the processed material arrives at the corresponding processing position of the assembly step or the welding step.

[0073] Specifically, a photoelectric incremental rotary encoder is installed on the main shaft of the main conveyor belt drive mechanism. The photoelectric incremental rotary encoder outputs an orthogonal pulse sequence corresponding to the conveyor speed. The intelligent monitoring system acquires this orthogonal pulse sequence through a high-speed counter input board and calculates the continuous linear displacement variable of the processed material on the conveyor belt by integrating it with a preset mechanical transmission ratio parameter. Simultaneously, radio frequency identification (RFID) antennas are arranged at nodes along the flow path to scan electronic tags installed on the processed material carrier and parse the corresponding flow code. By fusing the linear displacement variable and RFID data, the intelligent monitoring system updates the coordinate status of the processed material in the tracking database. When the processed material enters the pre-execution area of ​​the downstream assembly step or the positioning fixture area of ​​the welding step, the area limit sensor outputs a positioning feedback signal. When the downstream control node detects this positioning feedback signal, it executes a synchronous clock reading command and uploads the corresponding timestamp, defining the timestamp value as the arrival time node.

[0074] Step S40: Calculate the flow dwell time based on the demolding time node and the arrival time node.

[0075] Specifically, after receiving a data frame containing the arrival time node, the intelligent monitoring system retrieves the matching demolding time node from the tracking database according to the corresponding flow code. The intelligent monitoring system calculates the difference between the arrival time node and the demolding time node using an arithmetic operation unit, and after subtracting a pre-calibrated network communication delay constant, defines the resulting value as the flow dwell time. Specifically, the derivation steps for the pre-calibrated network communication delay constant are as follows: During the no-load initialization phase of the multi-process automated production line, the main control terminal of the intelligent monitoring system continuously sends 1000 probe data packets of a specific frame length to each physical workstation node via the IEEE 1588 Precision Time Protocol (PTP), and records the round-trip time difference from sending to receiving the response signal for each data packet; after removing the highest 5% and lowest 5% extreme values ​​from all round-trip time difference data, the arithmetic mean of the remaining data is calculated, and half of this arithmetic mean is fixed as the unidirectional network communication delay constant. In a gigabit industrial Ethernet hardware environment, the effective value range of this constant is calibrated as follows: to This transit dwell time characterizes the time spent on the physical transfer and waiting of processed materials between heterogeneous processes.

[0076] Step S50: Input the initial forming deviation value and the flow residence time into the preset stress deformation attenuation mapping model to predict the real-time micro-drift value accumulated by the processed material during the flow residence time.

[0077] It is understood that the data processing module of the intelligent monitoring system is pre-configured with the preset stress-deformation attenuation mapping model. The processing unit uses the initial forming deviation value, which includes three-dimensional coordinate direction parameters, as the reference space input variable of the model, and the flow residence time as the time extrapolation independent variable of the model. The preset stress-deformation attenuation mapping model extracts the physical parameter matrices such as the material's attenuation time constant and Poisson's ratio, substitutes the flow residence time into the preset damping attenuation integral kernel function for calculation, and adjusts the constant gain coefficient in combination with the data fed back by the workshop ambient temperature sensor, outputting a unitless feature matrix representing the deformation change rate. Specifically, the damping attenuation integral kernel function is defined based on the linear viscoelastic constitutive equation as: .in, The input is the flow dwell time; The elements on the main diagonal of the unitless eigenvalue matrix are scalars. The stress relaxation time constant is dynamically adjusted with temperature. For the natural constant The symbol for an exponential function with base 0; The absolute temperature is the input independent variable. The derivation of the equations follows Arrhenius's law: ,in The relaxation time constant for the material reference characteristic (its value range is constrained within) to ), The relaxation activation energy constant of the material (unit: ), The ideal gas constant (values) ), The absolute temperature measurement value fed back by the ambient temperature sensor in the workshop (unit: Furthermore, the left side of the equals sign in the formula... The stress relaxation time constant is dynamically adjusted with temperature. For the natural constant The exponential function operator is used as the base. Subsequently, a tensor dot product operation is performed between the unitless characteristic matrix and the spatial vector represented by the initial forming deviation value to obtain the incremental bias variable matrix. Specifically, in the above derivation equation, the ideal gas constant... The value is set to Characteristic relaxation time constant The values ​​are extracted by offline stress relaxation physical curve fitting of the material sample using a dynamic thermomechanical analyzer (DMA), and the effective value range is set to a limit within the system storage area. to Between; relaxation activation energy constant Established based on the time-temperature equivalence principle (TTS) of polymers, with units of [unit missing]. The internal data flow of the tensor dot product operation is as follows: [System computation core construction] Unitless eigendiagonal matrix The three independent elements on its main diagonal correspond to the deformation attenuation scalars in the orthogonal directions of the X, Y, and Z axes obtained independently from the integral kernel function; the control operator uses this diagonal matrix The components extracted from the initial molding deviation values Performing linear algebraic matrix multiplication on column vectors, the output of which is then established as a structure. The incremental bias variable matrix of the column vector. Finally, the incremental bias variable matrix and the initial forming deviation value are vector-summed in the same coordinate system to obtain the real-time micro-drift value. The real-time micro-drift value predicts the absolute coordinate change after the processed material structure undergoes stress relaxation shift under the action of flow time.

[0078] Step S60: Convert the real-time micro-drift value into a feedforward compensation command and send it to the downstream control node.

[0079] Specifically, the intelligent monitoring system includes a data protocol conversion driver module. This module calls the kinematic or thermodynamic parameter equations to execute a unit conversion algorithm, converting the real-time micro-drift value characterizing the spatial position error into the pulse compensation displacement of the servo controller, the torque control current limiting adjustment, or the power regulation duty cycle coefficient of the heat source controller. Specifically, the solution equation for the unit conversion algorithm is as follows: When the system command is converted into a pulse compensation displacement... At that time, the mapping equation is executed. ,in The pulse compensation displacement parameter is calculated by the system. The absolute distance component of the real-time micro-drift value in the direction of the actuator feed axis (unit: ), For servo ball screw lead parameters (unit: ), The single-turn resolution constant of the absolute encoder (unit: When converted to torque control current limiting adjustment amount At that time, execute equation ,in This is the torque control current limiting adjustment parameter calculated by the system. The corresponding component's axial equivalent structural stiffness constant (unit: N / mm). For the servo ball screw lead parameter (unit: mm / r, multiplied in the formula) (used for conversion to the standard unit meter) The torque constant (unit: N·m / A) for driving the servo motor. This is the mechanical thrust conversion efficiency constant; when converted to the power regulation duty cycle coefficient... At that time, execute equation ,in The power regulation duty cycle coefficient parameter is calculated by the system. As the baseline process heating duty cycle constant, This is the proportionality constant for the thermal deformation power response. Specifically, the exact method by which the system extracts the underlying parameters of the above unit transformation equation is: the axial equivalent structural stiffness constant of the component. (Unit is) The torque constant is obtained by extracting constants from the static compressive displacement result file of the three-dimensional finite element analysis (FEA) software offline simulation; (Unit is) The mechanical conversion efficiency is obtained by retrieving the factory-set hardware electromagnetic configuration parameters from the servo motor's read-only memory using the drive controller. The preset value is fixed at to Within a constant range. Subsequently, the generated underlying drive adjustment parameters are encapsulated into feedforward compensation instruction data messages according to a preset communication protocol, and then routed to the downstream control nodes of the designated execution station via the industrial network.

[0080] Step S70: The downstream control node dynamically adjusts the mechanical execution parameters in the assembly step or the thermal output parameters in the welding step according to the feedforward compensation command, thereby counteracting the real-time micro-drift value in the processing operation.

[0081] Understandably, after receiving and parsing the feedforward compensation instruction data message, the downstream control node extracts the compensation parameters from the message and overwrites and updates the preset baseline execution parameters in its internal operating register. For the specific processing operation of the assembly or welding step, the driving physical actuator applies a mechanical load or thermal output containing the feedforward correction based on the updated parameters. This offsets the dimensional deviations caused by component stress release during processing through mechanical forced deformation or thermophysical contraction mechanisms.

[0082] The physical implementation of the above-mentioned collaborative logic in various specific processes is described in detail below.

[0083] Step S80: The molding step includes insert injection molding; in the molding step, the image processing system is used to monitor the pre-mold closing state of the insert; the actual position coordinates of the insert are compared with the reference position coordinates, and when the position deviation value is within the allowable limit, the local heating rod array set inside the mold is controlled to perform adaptive heating action to maintain the local area temperature within the set target range.

[0084] Specifically, in the insert injection molding workstation, a multi-axis servo transfer robot grasps a metal insert and places it into the positioning hole of the moving mold. Before the mold performs a closing action, the controller triggers a light source and an industrial camera to acquire images. The image processing system parses the image matrix to obtain the three-dimensional point cloud coordinates of the insert within the mold, and extracts the actual position coordinates of the insert through inverse coordinate system projection. The logic operation unit performs spatial vector subtraction, compares the actual position coordinates of the insert with the reference position coordinates pre-stored in memory, and calculates the position deviation value including translation and rotation components. If the position deviation value is within a set allowable limit, the controller calculates the local variation distribution of fluid filling resistance based on the geometric direction of the position deviation value within the mold cavity space. For narrow regions where fluid resistance is predicted to increase, the controller increases the pulse width modulation duty cycle of the driving heating tube circuit in the corresponding region; for regions where fluid resistance decreases, the controller decreases the duty cycle of the corresponding heating tube. This controls the local heating rod array to perform adaptive heating, establishes a dynamic temperature gradient field on the cavity contact surface, changes the local apparent shear viscosity of the polymer melt, reduces the flow resistance increment, maintains the balance of melt filling speed, and reduces the residual stress caused by uneven flow field during the injection molding stage.

[0085] The set target range refers to the permissible temperature range of a local area of ​​the mold, pre-calibrated based on the heat distortion temperature of the insert material and the flow characteristics of the polymer melt. Its lower limit is the minimum mold temperature to ensure sufficient melt filling, and its upper limit is the maximum mold temperature to prevent excessive expansion of the insert or thermal degradation of the polymer. The system uses closed-loop control to ensure that the real-time temperature always converges within this range.

[0086] Step S90: The cutting step includes precision stamping operation; in the cutting step, a pre-set pre-springback compensation angle is obtained, and the actual springback amount is obtained by real-time measurement by a sensor; in conjunction with a closed-loop feeding mechanism, an error correction action is performed according to the pre-springback compensation angle and the actual springback amount obtained in real time to maintain the macroscopic dimensions in a stable state.

[0087] Specifically, the controller's non-volatile storage area pre-stores the basic bias angle parameters calculated based on the physical thickness and elastic modulus of the metal material, i.e., the pre-springback compensation angle. Specifically, the pre-springback compensation angle... The solution is performed based on the analytical equations of elastic-plastic pure bending springback: In the formula, To calculate the pre-springback compensation angle of the output; The nominal yield strength constant of the strip metal to be processed (unit: MPa). This is the constant elastic modulus of strip metal (unit: MPa). The physical thickness constant of the strip (unit: mm); This is the theoretical bending radius constant of the stamping die (unit: mm). The theoretical bending angle of the die is specified in the drawing. During the die opening and return phase of the progressive stamping cycle, a laser displacement scanning sensor located on the ejector side collects the point cloud coordinates of the strip surface after stamping. The three-dimensional coordinates are reconstructed by a signal processing algorithm, and the measured value of the actual bending angle is extracted. The algebraic difference between this actual bending angle and the angle specified in the drawing is calculated and defined as the actual springback. The closed-loop feeding mechanism consists of an AC servo motor, drive rollers, and an incremental encoder. The control unit calculates the difference between the actual springback and the pre-springback compensation angle to generate the displacement offset step parameter of the closed-loop feeding mechanism. Before starting the next stamping feeding action, the servo control board of the closed-loop feeding mechanism superimposes the displacement offset step parameter into the set reference feed pulse command, controls the servo motor to execute the corresponding additional rotational angular displacement, and drives the rollers to perform fine-tuning and correction actions on the strip feed reference position to maintain stable macroscopic output.

[0088] Step S100: The precision stamping operation covers a multi-step composite feature manufacturing process; the multi-step composite feature manufacturing process includes: breaking down the continuous manufacturing action into a pre-bending action, a step-by-step rolling action, and a finishing and sizing action; in each sub-step of the step-by-step rolling action, an independently allocated compensation overpressure value is applied to the corresponding step corner; based on the number of workstations determined by simulation analysis, the change in processing diameter between adjacent workstations is constrained within a set step gradient range to ensure that the drawing and forming is completed before the material hardens.

[0089] In the front station of the progressive die, the upper die punch applies a preliminary physical bending moment before the strip reaches its yield strength, performing the pre-bending action. The strip is then conveyed into a cluster of stations performing step-by-step rolling actions, where the wrap angle parameters of each die insert are arranged in a progressively increasing manner. During the step-by-step rolling action, a piezoelectric ceramic actuator or a servo-hydraulic wedge pusher is installed in the transition area of ​​the multi-step structure corner. When the punching system detects that the slider displacement has reached the set lower dead center coordinates, it outputs an actuation level signal to the corresponding micro-actuator. Based on this signal, the actuator extends by the corresponding displacement, applying a highly concentrated mechanical compressive stress load to the surface of the local step corner area of ​​the material, introducing compressive stress into the material lattice, i.e., loading the compensated overpressure value. The internal stress retention zone constructed by this compressive stress blocks the tensile stress released in adjacent areas when the material is unloaded from the die, suppressing the forming springback of the composite step feature.

[0090] The system sets the absolute reduction difference in the cavity diameter of adjacent workstations in a progressive die based on the material hardening exponent constant, ensuring that the change in machining diameter between adjacent workstations meets the set step gradient range. Specifically, the set step gradient range is mathematically boundaryed by combining the limit threshold of Hollomon's material strain hardening constitutive equation. Let the first... The machining diameter of the secondary workstation is The processing diameter of the adjacent preceding workstation is The diameter reduction ratio, which characterizes the change in machining diameter, is defined as follows: The strain hardening exponent constant of the unitless material obtained from the test is denoted as... (The value range for strip is set at) to (between), the system's decision constraints are: In the formula, A positive integer index representing the order of the workstation track layout; For the first The reduction ratio of the machining diameter at each pass station; For the first The machining diameter value of the substation; This refers to the machining diameter values ​​of adjacent preceding workstations. The system constrains the upper limit of the absolute reduction ratio of the machining diameter of all adjacent workstations within the solution set of this inequality; this defined solution set establishes the set step gradient range. Specifically, the material hardening exponent constant... The derivation steps are as follows: Perform a standard uniaxial tensile force test on metal strip samples from the same batch used in production, extract the true stress-strain data during the plastic deformation stage, perform linear regression fitting on the physical curve in a double logarithmic coordinate system, and extract the slope parameter of the fitted line, which is the constant. The system control register, tailored to the characteristics of cold-worked metals, sets the hardening exponent constant. The scope of the permitted assignment is defined in to Between. When the comparator circuit detects the diameter reduction ratio. When the calculated value exceeds the extreme threshold of the parameter on the right side of the constraint inequality, the algorithm immediately outputs an over-limit blocking signal, interrupting the die stepping design sequence to prevent physical tearing failure. This physical parameter constraint controls the cumulative rate of cold work dislocation multiplication hardening during the material's plastic deformation stage within a set threshold range, ensuring that the material completes the drawing and forming operation before losing its plastic ductility and fractures. After the aforementioned process, the strip enters the extrusion die station with a closed end structure to receive a static extrusion load, completing the finishing and sizing action to reduce non-roundness error.

[0091] like Figure 5 As shown, step S110: The detection step includes: linking the image processing system to perform a sharp corner contact height verification and appearance defect screening on the final synthesized component after the assembly step and the welding step; comparing the verification results with the preset standard template; when it is determined that there is an appearance defect or the height dimension exceeds the tolerance boundary, sending an abnormal trigger signal to the central control terminal, and controlling the rejection sorting branch to move the corresponding component out of the normal flow track.

[0092] Specifically, the final synthesized component, having undergone various processes, enters the optical inspection sealed chamber via a conveyor rail. Limit sensors trigger the structured light projector and area array camera to operate, and the image processing system's processing board captures a three-dimensional point cloud data dictionary of the component's surface. The point cloud processing algorithm calls the Random Sample Consensus (RANSAC) algorithm to fit and extract the three-dimensional spatial equation of the assembly reference surface. It then locates the top coordinates of the component's electrical contact sharp corner and calculates the projection vector distance of these coordinates in the reference surface normal direction, outputting the verification value of the sharp corner contact height. Specifically, the underlying extraction steps of the RANSAC algorithm are: randomly selecting three non-collinear three-dimensional coordinate points from the point cloud data to generate initial candidate plane equations. ,in, , , These are the constant coefficients of the normal vector of the candidate plane on the X, Y, and Z axes, respectively. Let be the intercept constant of the candidate plane equation. , , These are the independent variables of continuous spatial coordinates in a three-dimensional coordinate system; calculate the orthogonal geometric distance from the remaining point cloud to the candidate plane, and set the absolute value of the orthogonal geometric distance to a tolerance threshold for points within the distance (a system-defined constant). The data points are divided into an interior point set; 1000 iterative calculations are performed, and the parameter corresponding to the iteration with the largest number of interior points is extracted as the three-dimensional spatial equation of the assembly reference surface. The projection vector distance... The derivation and calculation formula is as follows ,in, The calculated distance value of the projection vector; , , , These are the normal constant coefficients and intercept constant terms of the X-axis, Y-axis, and Z-axis corresponding to the extracted three-dimensional spatial equations of the assembly reference surface; , , These are the independent coordinate numerical components of the absolute coordinates of the apex of the sharp corner in the X, Y, and Z axes. A synchronous control time sequence excites a multi-band shadowless light source array, and a planar array camera acquires two-dimensional image data. The system activates the gray-level co-occurrence matrix and morphological kernel, filters out background components, and extracts defect feature matrices representing scratches, indentations, or weld slag, performing appearance defect screening. Specifically, the appearance defect screening calls the gray-level co-occurrence matrix (GLCM) algorithm to extract local contrast feature values ​​and calls a 5×5 pixel structuring element to perform morphological opening operations to filter out high-frequency noise. The system's internally set defect maximum tolerance constant extraction benchmark is: the local contrast value extracted in the detection area is greater than three times the average contrast value of the global defect-free background area; simultaneously, according to the morphological connected component labeling algorithm, the total geometric area of ​​the connected component of a single bright pixel exceeds... When both of the above operational conditions are met simultaneously, the judgment logic is established and outputs the judgment result that there is an appearance defect.

[0093] The system's core calculation performs a logical comparison between the extracted high-resolution verification data and defect feature pixel matrix values ​​and the tolerance extreme value array and maximum defect tolerance constant (i.e., the preset standard template) set in the memory. If the comparison conditions do not meet the preset interval constraints, the system's alarm register state flips, sending the abnormal trigger signal to the main control terminal. The main control terminal locks the current coded position coordinates of the corresponding abnormal component. When the component moves with the conveyor belt to the preset rejection sorting branch entrance judgment point, it drives the bottom electromagnetic reversing valve to connect the pneumatic pipeline, controlling the lateral cylinder piston to quickly extend, pushing the component laterally away from the main conveyor belt and into the isolation collection box.

[0094] Step S120: The packaging step includes: when the detection step determines that the final synthesized component meets the factory standards, the final synthesized component is imported into the automatic packaging module; the automatic packaging module is controlled to perform a heat-sealing and film-coating action on the final synthesized component; a product manufacturing history data file containing the surface image parameters, the real-time microscopic drift values ​​and the corresponding processing trajectory data is established, and a data identification label carrying the link of the product manufacturing history data file is printed and affixed to the packaging surface.

[0095] Specifically, the final synthesized components that do not trigger the abnormal mechanism are sequentially incorporated into the insulating carrier tape cavity configured in the packaging module. The servo unwinding system continuously unwinds the heat-sealing cap film. Within the automated packaging station, a negative pressure suction cup component physically adheres the cap film to the edge of the carrier tape cavity. The heat-sealing pressure head, under closed-loop temperature control, performs a downward stroke driven by a cylinder, with its working end face contacting the film surface and maintaining a fixed heating time constant. The heat-sensitive resin coating at the bottom of the film undergoes a state transition upon heating, generating an adhesive physical effect. After pressure release and cooling, it cures, completing the heat-sealing coating process.

[0096] In the background digital process, the production line data integration server extracts the surface image parameters captured by the component at the demolding node, the real-time micro-drift values ​​calculated during the model feedforward prediction stage, and the timestamps and absolute coordinate data read at each transfer station from the databases of each control node as the processing trajectory data. The server uses a text formatting program to structure and merge the above datasets into a string sequence file, calls a hash logic encryption algorithm engine to generate a specific checksum, and constructs a file addressing link, i.e., a product manufacturing history data file. The graphics control module converts this link into a two-dimensional data matrix pixel bitmap and sends it to the memory of the industrial printer. The thermal transfer printer cures the carbon ribbon color base onto the antistatic label substrate based on the bitmap data to generate the data identification label. A four-degree-of-freedom labeling robot arm grasps the data identification label in a vacuum suction state and applies constant pressure to press it onto the corresponding outer interface of the laminated packaging.

[0097] like Figure 3 As shown, step S130: The construction process of the preset stress deformation attenuation mapping model includes: collecting sample data of springback deformation caused by the release of residual heat and residual mechanical stress after the processing of the material in the historical processing batches is completed; establishing a multivariate mapping algorithm with initial deviation and flow time as independent variables and actual deformation as dependent variable; using the multivariate mapping algorithm to form a mathematical relationship matrix to characterize the gradual attenuation law of the internal stress of the material over time, and solidifying the mathematical relationship matrix into the preset stress deformation attenuation mapping model.

[0098] Specifically, during the offline calibration phase of the control model's static parameters, a test batch of materials with identical material properties is introduced into the production line. At the initial zero-point moment after molding and demolding, and subsequently at a series of measurement nodes distributed along a logarithmic time axis, a coordinate measuring machine triggers probes to acquire the three-dimensional absolute coordinate values ​​of a predefined group of points on the material surface. Simultaneously, a temperature measurement array records the corresponding environmental temperature drop parameter curves. The processing system extracts the displacement offset of each discrete measurement time point relative to the initial demolding moment, and structurally stores the numerical set containing coordinate displacement, environmental temperature, and corresponding flow time, constructing a high-dimensional springback deformation sample data.

[0099] The algorithm calculation unit extracts the rebound deformation sample data. The distance deviation value captured at the initial time point is extracted and defined as the initial deviation independent variable of the multivariate equation system; the difference between each measurement time and the initial time is extracted and defined as the flow time independent variable; the measured coordinate displacement change at each time period is extracted and defined as the actual deformation dependent variable. A multivariate mapping algorithm architecture incorporating time decay terms is constructed within the computational framework.

[0100] The system's core computational logic initiates iterative optimization calculations, performing gradient descent operations within the sample data tensor to calculate the parameter coefficient set at which the sum of squared residuals between the predicted surface and the physically measured deformation points reaches a convergence threshold. After satisfying the convergence condition, the corresponding spatial attenuation weight variables and constant terms for each computation node are extracted, generating a floating-point mathematical relation matrix data package. The main control unit invokes a write instruction sequence to directly convert this mathematical relation matrix data package into low-level bytecode format and write it into the non-volatile storage sector of the execution board, completing the solidification and storage of the preset stress-deformation attenuation mapping model.

[0101] like Figure 4 As shown, step S140: the step of dynamically adjusting the mechanical execution parameters in the assembly step includes: when the feedforward compensation command points to the assembly step, parsing the pre-tightening compensation force value used to offset the real-time micro-drift value from the feedforward compensation command; updating the pre-stored basic assembly downward pressure value in the control register to the pre-tightening compensation force value; applying a mechanical pressing action to the processed material through the servo drive mechanism according to the updated pre-tightening compensation force value, causing the processed material to produce reverse mechanical deformation, neutralizing the real-time micro-drift value, and effectively preventing assembly interference.

[0102] Specifically, after receiving a feedforward compensation instruction data message containing compensation data, the downstream control node executing the assembly process extracts the real-time micro-drift value containing the spatial deformation dimension from the interpretation module. The processing unit reads the structural stiffness matrix constant table of the current processing component, performs a multiplication operation between the variable vector represented by the real-time micro-drift value and the stiffness matrix, and solves for the transient reverse thrust physical parameter required to neutralize the dimensional deformation by applying an external load and causing the material to undergo corresponding structural yielding. The calculated parameter is then established as the preload compensation force value. Specifically, the transient reverse thrust physical parameter is solved using matrix operations based on Hooke's law of mechanics: In the equation, The numerical three-dimensional vector matrix of preload compensation force (unit: ); The equivalent structural stiffness 3×3 diagonal matrix constants of the processed components extracted through offline finite element model static compressive analysis are given (unit: ); The extracted real-time microscopic drift numerical displacement input vector. Action extraction is performed. The orthogonal projection scalar component along the mechanical vertical pressing axis is established as the numerical reference for the preload compensation force calculated from the actual control current.

[0103] The servo controller performs a register instruction overwrite operation, updating the internal limiting torque current input setting constant from the basic assembly downward pressure value to the preload compensation force value that includes the required increment. The servo drive mechanism switches to torque control mode, receives the increased control current target parameter, increases the motor drive duty cycle configuration, and outputs the changed electromagnetic output torque. This electromagnetic torque is converted into a linear downward pressure load for the end-press assembly via the transmission structure, applying a corresponding mechanical pressing action to the assembly according to the updated value. This applied excessive mechanical load forces the processed material undergoing outward deformation to generate elastic compressive physical displacement at the contact surface. The resulting reverse mechanical deformation volume offsets the deformation expansion predicted by the real-time micro-drift value, eliminating physical assembly interference during the assembly insertion process.

[0104] Step S150: The step of dynamically adjusting the thermal output parameters in the welding step includes: when the feedforward compensation command points to the welding step, determining the change in the contact surface gap width corresponding to the real-time micro-drift value; calculating the thermal capacity compensation requirement required to maintain the stability of the molten pool based on the change in the contact surface gap width; adjusting the basic heating power or pulse discharge duration of the welding equipment to match the actual output thermal energy with the thermal capacity compensation requirement, changing the shrinkage tension generated during the solidification and cooling stage of the weld joint, thereby offsetting the real-time micro-drift value.

[0105] Specifically, the main chip of the welding machine's bottom-level controller decodes and obtains the real-time microscopic drift value. The processing chip, combined with the clamping and positioning coordinate mapping matrix of the fixture equipment, executes a three-dimensional geometric coordinate projection transformation program to calculate the spacing separation value generated by the material's deformation due to flow over time, projected onto the normal direction of the component welding joint reference surface. This spacing value is assigned and defined as the change in the contact surface gap width.

[0106] The controller's logic module loads arrays of thermodynamic constants, including the material's specific heat capacity, latent heat of fusion, and phase transition density. Through integral calculations based on the law of conservation of energy, it determines the additional Joules of heat energy required to fill the expanded physical gap space and maintain thermodynamic stability in the high-temperature molten pool. This calculated output is established as the required heat capacity compensation. Specifically, the Joule value is derived based on the integral formula of the law of conservation of thermodynamic energy: In the equation, To extrapolate the calculated heat capacity compensation requirement (unit: ); is the solid physical density constant of the metallic matrix; The input value is the amount of change in the width of the contact surface gap. This is the set theoretical physical length constant for the weld. The theoretical effective penetration constant for the weld is defined; where the product term is... Characterizes the physical volume of the expansion gap caused by dimensional separation; The isobaric specific heat capacity coefficient of the metal; This is the absolute temperature of the liquidus melting point of the material. The ambient reference temperature is fed back by the temperature sensor; The constant of latent heat of enthalpy required for metal phase transformation solidification (unit: ).

[0107] The control board adjusts the underlying drive based on the heat capacity compensation requirement parameters. For welding in continuous output mode, the power module output is adjusted to change the basic heating power scalar; for welding in discrete pulse output mode, the duty cycle limit of the pulse discharge duration during the gate opening cycle is adjusted. The parameter changes ensure that the actual radiant heat energy parameters projected onto the bonding interface match the heat capacity compensation requirement. The increased heat input causes the heated component to melt, forming a liquid high-temperature molten pool with an increased physical cross-sectional area and volume. During the subsequent convective cooling and transition of the liquid metal to a solid lattice, the change in physical density generates an increased centripetal condensation contraction tensile stress vector. This condensation contraction tension pulls the deformed and separated base metal on both sides of the joint to perform a physical approach displacement towards the center. The incremental phase transformation contraction displacement offsets the separation of the mating dimensions from a physical geometric perspective, thus offsetting the initial forming deviation.

[0108] like Figure 6 As shown, step S160: The method further includes a closed-loop iterative optimization step for the feedforward compensation command: collecting the final inspection physical size error data output by the detection step; analyzing the correlation fluctuation trend between the final inspection physical size error data and the real-time micro-drift value; using the correlation fluctuation trend, using a data optimization algorithm to reverse correct the node weight coefficients inside the preset stress deformation attenuation mapping model, thereby improving the prediction accuracy of subsequent processing batches.

[0109] Specifically, the system data acquisition process continuously collects the final inspection physical absolute offset data array of each output component issued by the coordinate measuring machine of the inspection station, i.e., the final inspection physical size error data. The data processing server, based on the flow coding record sequence, aligns the final inspection physical size error data with the real-time micro-drift values ​​output by the model during the feedforward stage within the same time reference coordinate system. The main control algorithm calls a multivariate statistical covariance analysis algorithm to compare these two sets of data. If the algorithm output graph reflects a continuous offset pattern with a monotonically non-closed distribution evolving over time, the system extracts the residual statistical characteristics characterizing the system error caused by this slowly changing external environmental variable, and identifies it as the correlated fluctuation trend.

[0110] The calculation module initiates the tuning engine subroutine. Using a gradient descent data optimization algorithm based on error backpropagation, and taking the aforementioned residual statistical characteristic matrix (i.e., the associated fluctuation trend) as the input dependent variable for the objective optimization calculation, it calculates the sensitive partial derivative parameter of the damping factor independent variable with respect to the total residual value in the constant system within the decay mapping model through back-derivative calculation. Specifically, the gradient descent data optimization algorithm internally constructs the objective cost function (mean squared error, MSE expression): .in, The numerical value of the constructed objective cost function (mean squared error); The positive integer index for the summation operation of batch samples; This refers to the total number of statistical batch samples extracted within a limited time sliding window. The first output of the feedforward model calculation A predicted micro-drift value, To extract the corresponding first value from the actual measurement The final physical size error data. The calculated sensitive partial derivative parameters are the target cost function with respect to the node weight coefficient vector. Gradient vector of solution The update execution equation for the system's reverse iterative correction is: ;in, For the objective cost function The gradient constant parameter is obtained by taking the partial derivative of the existing node weight constant coefficients. The coefficients of the existing node weight constants. To update the coefficient array, To constrain divergence, the learning rate step size proportional constant (its value is strictly limited to...) to (Within the range). The original register address block is then overwritten to achieve a low-level update. The control algorithm generates an update parameter difference with a compensation ratio based on the partial derivative parameter, executes low-level memory safe overwrite logic, and directly writes the updated value into the storage address area, overwriting the original node weight coefficients in the preset stress deformation attenuation mapping model. This online self-looping correction mechanism effectively reduces the prediction mismatch rate of the computational model introduced by changes in the production environment through low-level data refresh, improving the prediction accuracy of subsequent processing batches.

[0111] Step S170: The collaborative step further includes an interception and blocking mechanism: comparing the absolute value of the real-time micro-drift value with the maximum mechanical tolerance boundary that the downstream processing position can accommodate; when the real-time micro-drift value exceeds the maximum mechanical tolerance boundary, determining that the processed material has undergone irreversible deformation beyond the limit due to excessive flow residence time; marking the pre-assigned flow code corresponding to the processed material as an abnormal isolation state, and controlling the lateral cylinder to blow the corresponding material away from the main conveyor belt before the processed material enters the processing position.

[0112] Specifically, before the feedforward compensation execution parameter data frame is sent to the servo drive unit to trigger the mechanical action, the system safety check subroutine reads the physical extreme value dimension constant parameter of the downstream equipment fixture opening avoidance structure from the read-only memory and defines it as the maximum mechanical tolerance boundary. Specifically, the maximum mechanical tolerance boundary constraint constant is established. The algebraic extraction relation equation is: In the formula, The maximum mechanical tolerance boundary constraint constant is determined for calculation; This is the static constant for the maximum opening distance of the gripper when it performs an opening and avoidance action, extracted from the three-dimensional physical model of the downstream equipment; The nominal dimension parameters of the external maximum solid outline set for the corresponding material drawing; The safety gap redundancy constant (set value) is imposed to cover the errors of the dynamic repeatability positioning system. The underlying logic level extracts the spatial magnitude scale of the three-dimensional vector of the real-time micro-drift value, and determines whether the magnitude distance value is greater than the calculated value. A threshold constant is used; when the determination result is yes, the blocking condition is met. The comparison logic hardware acquires the input real-time micro-drift value vector and performs scalar-to-absolute conversion to extract the absolute spacing value. The hardware Boolean gate circuit executes a comparison instruction to compare the absolute spacing diameter of the real-time micro-drift value with the maximum mechanical tolerance boundary in terms of level magnitude.

[0113] When the comparison result indicates that the real-time micro-drift value exceeds the maximum mechanical tolerance boundary value allowed by the rigidity of the machine tool, the control system, based on solid failure theory, diagnoses that the processed material has lost the feasibility of elastic correction due to long-term retention and has undergone irreversible deformation beyond the limit. The network central scheduling processor immediately terminates the communication for sending feedforward compensation parameters to the processed material and updates the status field of the processed material identification code in the tracking database system form as an abnormal isolation status identifier prohibiting operation. At the same time, the system starts the physical position tracking subprocess. When the material entity carrying the scrap identification data arrives at the spatial coordinates of the safety interception section arranged in front of the downstream high-precision process position with the conveyor belt displacement, the digital output terminal of the area control station is turned on to output a drive voltage signal. This signal connects the control valve core of the high-flow electromagnetic reversing valve in the pneumatic pipeline, and high-pressure compressed air enters the power chamber of the side-arranged cylinder, controlling the piston push rod structure of the side cylinder to extend rapidly. The piston ejection displacement action pushes the corresponding material on the conveyor belt laterally away from the main movement trajectory and falls into the set waste part isolation collection tank, preventing the failed components that have undergone destructive deformation and distortion from entering the precision mold structure and causing physical collision and jamming.

[0114] 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 collaborative control and intelligent monitoring method for a multi-process automated production line, applied to a multi-process automated production line including molding, cutting, assembly, welding, inspection, and packaging steps, equipped with an image processing system and an intelligent monitoring system; the method includes the following steps: The processed material is transferred between each of the steps, and the surface image parameters of the processed material are acquired by the image processing system. The intelligent monitoring system is used to identify whether the processed material meets the set tolerance range. Its features are, The method also includes a collaborative step: In the molding step or the cutting step, the initial molding deviation value within the set tolerance range is extracted, and the demolding time node when the processed material leaves the corresponding processing position is recorded; Track the flow trajectory of the processed material, and record the arrival time node when the processed material arrives at the corresponding processing position of the assembly step or the welding step; The transit dwell time is calculated based on the demolding time node and the arrival time node; The initial forming deviation value and the flow dwell time are input into a preset stress deformation attenuation mapping model to predict the real-time micro-drift value accumulated by the processed material during the flow dwell time. The real-time micro-drift value is converted into a feedforward compensation command and sent to the downstream control node; The downstream control node dynamically adjusts the mechanical execution parameters in the assembly step or the thermal output parameters in the welding step according to the feedforward compensation command, thereby counteracting the real-time micro-drift value during the processing operation.

2. The collaborative control and intelligent monitoring method for multi-process automated production lines as described in claim 1, characterized in that, The molding step includes insert injection molding; in the molding step, the image processing system is used to monitor the pre-mold closing status of the insert. By comparing the actual position coordinates of the insert with the reference position coordinates, when the position deviation is within the allowable limit, the local heating rod array set inside the mold is controlled to perform adaptive heating action to maintain the temperature of the local area within the set target range. The cutting step includes a precision stamping operation; in the cutting step, a pre-set pre-springback compensation angle is obtained, and the actual springback amount is obtained by real-time measurement by a sensor; With the help of a closed-loop feeding mechanism, error correction actions are performed based on the pre-springback compensation angle and the actual springback amount measured in real time, so as to maintain the macroscopic dimensions in a stable state.

3. The collaborative control and intelligent monitoring method for multi-process automated production lines as described in claim 2, characterized in that, The precision stamping operation encompasses a multi-step composite feature manufacturing process; the multi-step composite feature manufacturing process includes: The continuous manufacturing process is broken down into pre-bending, step-by-step rolling, and finishing and sizing processes. In each sub-step of the step-by-step rolling action, an independently assigned compensation overpressure value is applied to the corresponding step corner. Based on the number of workstations determined by simulation analysis, the variation in processing diameter between adjacent workstations is constrained within the set step gradient range to ensure that drawing and forming are completed before the material hardens.

4. The collaborative control and intelligent monitoring method for multi-process automated production lines as described in claim 3, characterized in that, The detection steps include: The image processing system is used to perform a sharp corner contact height check and appearance defect screening on the final composite component after it has undergone the assembly step and / or the welding step. The verification results are compared with the preset standard template; when it is determined that there is an appearance defect or the height dimension exceeds the tolerance boundary, an abnormal trigger signal is sent to the central control terminal, and the rejection sorting branch is controlled to move the corresponding component out of the normal flow track.

5. The collaborative control and intelligent monitoring method for multi-process automated production lines as described in claim 4, characterized in that, The packaging step includes: When the detection step determines that the final synthesized component meets the factory standards, the final synthesized component is imported into the automatic packaging module. The automatic packaging module is controlled to perform a heat-sealing and film-coating action on the final synthesized component. A product manufacturing history data file is created, which includes the surface image parameters, the real-time micro-drift values, and the corresponding processing trajectory data. A data identification label carrying a link to the product manufacturing history data file is printed and affixed to the packaging surface.

6. The collaborative control and intelligent monitoring method for multi-process automated production lines as described in claim 1, characterized in that, The process of constructing the preset stress deformation attenuation mapping model includes: Collect sample data of springback deformation caused by residual heat and residual mechanical stress release in the processed materials of the historical processing batches after molding; A multivariate mapping algorithm is established with initial deviation and turnover time as independent variables and actual deformation as dependent variable; The multivariate mapping algorithm is used to form a mathematical relationship matrix that characterizes the gradual decay of internal stress in a material over time, and the mathematical relationship matrix is ​​solidified into the preset stress deformation decay mapping model.

7. The collaborative control and intelligent monitoring method for multi-process automated production lines as described in claim 6, characterized in that, The step of dynamically adjusting the mechanical actuation parameters in the assembly step includes: When the feedforward compensation command is directed to the assembly step, the pre-tightening compensation force value used to offset the real-time micro-drift value is parsed from the feedforward compensation command. Update the pre-stored basic assembly downforce value in the control register to the pre-tightening compensation force value; The servo drive mechanism applies mechanical pressing action to the processed material according to the updated pre-tightening compensation force value, causing the processed material to undergo reverse mechanical deformation, neutralizing the real-time micro-drift value, and effectively preventing assembly interference.

8. The collaborative control and intelligent monitoring method for multi-process automated production lines as described in claim 7, characterized in that, The step of dynamically adjusting the heat output parameters in the welding process includes: When the feedforward compensation command is directed to the welding step, the change in the contact surface gap width corresponding to the real-time micro-drift value is determined. Based on the change in the width of the contact surface gap, calculate the heat capacity compensation required to maintain the stability of the molten pool. Adjusting the basic heating power or pulse discharge duration of the welding equipment to match the actual output heat energy with the required heat capacity compensation, and changing the shrinkage tension generated during the solidification and cooling stage of the weld joint, thereby offsetting the real-time micro-drift value.

9. The collaborative control and intelligent monitoring method for multi-process automated production lines as described in claim 8, characterized in that, The method also includes a closed-loop iterative optimization step for the feedforward compensation command: Collect the final physical dimension error data output from the aforementioned detection steps; Analyze the correlation and fluctuation trend between the final inspection physical size error data and the real-time microscopic drift value; By utilizing the aforementioned correlation fluctuation trend, a data optimization algorithm is used to reverse-correct the node weight coefficients within the preset stress deformation attenuation mapping model, thereby improving the prediction accuracy of subsequent processing batches.

10. The collaborative control and intelligent monitoring method for multi-process automated production lines as described in claim 9, characterized in that, The coordinated steps also include an interception and blocking mechanism: The absolute value of the real-time micro-drift value is compared with the maximum mechanical tolerance boundary that the downstream processing position can accommodate. When the real-time micro-drift value exceeds the maximum mechanical tolerance boundary, it is determined that the processed material has undergone irreversible deformation beyond the limit due to excessive flow and residence time. The pre-assigned flow code corresponding to the processed material is marked as an abnormal isolation state, and the lateral cylinder is controlled to blow the corresponding material away from the main conveyor belt before the processed material enters the processing position.

Images