Uhmw-pe mold temperature and pressure decoupling adaptive control method based on digital twinning

Abstract

The present application relates to the technical field of die forming control, and particularly relates to a UHMW-PE die temperature and pressure decoupling adaptive control method based on digital twinning, which comprises the following steps: reconstructing the three-dimensional temperature field and crystallinity field inside the blank by constructing a physical reduced-order twin model; determining the closed core phase change area, and solving the volume shrinkage rate and core volume loss based on the PVT state equation; generating a negative pressure evolution index combined with the shear stress relaxation rate, and switching to a microscopic feeding mode when the critical condition is met; constructing a complex rheological impedance compensation function, generating a high-frequency micro-pulse displacement command combined with the core volume loss, and driving the pressure head to implement feeding; and finally collecting the tonnage reaction force response curve, and using the unscented Kalman filter algorithm to correct the twin model parameters in reverse. The present application realizes accurate perception and active feeding of the core shrinkage negative pressure, and effectively eliminates the shrinkage hole defects of the product.

Description

Technical Field

[0001] This invention relates to the field of molding control technology, and in particular to a UHMW-PE molding temperature and pressure decoupling adaptive control method based on digital twins. Background Technology

[0002] Ultra-high molecular weight polyethylene (UHMW-PE) is widely used in high-end industrial and medical implant fields due to its extremely excellent wear resistance, impact resistance, and self-lubricating properties. Its compression molding process typically employs high-pressure heating followed by prolonged pressure holding and cooling. Pressure signals are collected by pressure sensors installed on the mold wall to control the output tonnage of the hydraulic press, thereby maintaining stable macroscopic pressure during the compression molding process.

[0003] UHMW-PE melt has extremely high viscosity and extremely slow heat transfer. In the later stage of molding cooling and solidification, the outer layer of the blank will harden first to form a hard outer shell, which blocks the effective transmission of the external main holding pressure to the inside. At this time, the internal core continues to undergo crystallization phase transformation and shrinkage, which will generate severe microscopic negative pressure tensile stress in the closed area. Traditional wall pressure control cannot detect and eliminate this core shrinkage negative pressure, which can easily lead to shrinkage cavities and microscopic void defects inside the product. Summary of the Invention

[0004] To overcome the above shortcomings, this invention provides a UHMW-PE molding temperature and pressure decoupling adaptive control method based on digital twins, which aims to improve the problem that traditional wall pressure control cannot detect and eliminate the negative pressure of core phase change shrinkage, which makes the product prone to shrinkage cavities and micro voids.

[0005] This invention provides the following technical solution: a UHMW-PE molding temperature and pressure decoupling adaptive control method based on digital twins, comprising: S1. Construct a physical reduced-order twin model, collect real-time mold wall temperature, wall pressure and press head displacement data of the molding press and input them into the physical reduced-order twin model to reconstruct the three-dimensional temperature field and three-dimensional crystallinity field inside the billet; S2. Determine the closed core phase transition region based on the three-dimensional crystallinity field, and mark the grid nodes in the three-dimensional crystallinity field whose crystallinity is within the preset phase transition range as the phase transition front transition region; when the phase transition front transition region is completely surrounded in three-dimensional space by an outer solidification envelope surface with crystallinity higher than the preset hardening threshold, the surrounded phase transition front transition region and the internal uncrystallized region are determined as the closed core phase transition region; solve the volume shrinkage rate change rate and core volume defect of the closed core phase transition region based on the PVT state equation; S3. Solve for the shear stress relaxation rate of the closed core phase transition region based on the three-dimensional temperature field, solve for the dimensionless relaxation-contraction criterion based on the ratio of the volume shrinkage rate change rate to the shear stress relaxation rate, solve for the microscopic cavitation damage tensor of the closed core phase transition region based on the criterion, and map the principal component of the microscopic cavitation damage tensor to a negative pressure evolution index; when the negative pressure evolution index meets the critical switching condition, switch to the microscopic compensation control mode; S4. Derive the molecular chain shear stress relaxation time based on the three-dimensional temperature field and construct a complex rheological impedance compensation function. Calculate the target shrinkage displacement amplitude based on the core volume loss. Use the complex rheological impedance compensation function to perform phase and gain modulation on the target shrinkage displacement amplitude and superimpose a high-frequency oscillation carrier to generate a displacement control command containing a high-frequency micro-amplitude pulse waveform. S5. Send the high-frequency micro-pulse displacement command to the servo drive actuator of the molding machine to control the action of the pressure head to compensate for the shrinkage of the closed core phase change region; S6. Collect the tonnage reaction force response curve of the servo drive actuator, compare it with the tonnage response predicted by the physical reduced-order twin model using the unscented Kalman filter algorithm, and correct the parameters of the physical reduced-order twin model in reverse.

[0006] Furthermore, in S1, constructing the physically reduced twin model includes: High-order finite element simulation data of the molding process were collected, and the reduced-order spatial basis functions of the flow field and temperature field were extracted using the intrinsic orthogonal decomposition algorithm. A physical constraint surrogate model based on Gaussian process regression is constructed, and the surrogate model is used to establish online the nonlinear mapping relationship between the real-time mold wall temperature, wall pressure and indenter displacement data and the coefficients of the reduced-order space basis function; The coefficients of the reduced-order spatial basis function obtained by real-time calculation are linearly superimposed and reconstructed with the reduced-order spatial basis function to output the three-dimensional temperature field and three-dimensional crystallinity field inside the billet in real time.

[0007] Furthermore, in S2, solving for the rate of change of volume shrinkage and the volume defect of the closed core phase transition region based on the PVT equation of state includes: Based on the three-dimensional temperature field and wall pressure, the specific volume of each grid cell in the closed core phase transition region is calculated in real time using the dual-domain Tait PVT equation of state. The rate of change of volume shrinkage of each grid cell is determined by the differential of the specific volume with respect to time. The volume loss caused by phase transition shrinkage in each grid cell within the closed core phase transition region is accumulated by three-dimensional spatial integration to generate the core volume loss.

[0008] Furthermore, in S3, solving for the shear stress relaxation rate of the closed core phase transition region based on the three-dimensional temperature field includes: Based on the Arrhenius equation or the WLF equation, the temperature shift factor and relaxation time of the closed core phase transition region are calculated according to the three-dimensional temperature field. The shear relaxation modulus at the current temperature is derived using the generalized Maxwell rheological model, and the shear stress relaxation rate is solved by combining it with the shear strain rate of the phase transformation region of the closed core.

[0009] Furthermore, in S3, switching to the micro-compact control mode includes: Disconnect the static constant pressure holding control circuit that relies primarily on feedback from the pressure sensor on the molding machine wall. Switch to an adaptive servo displacement control mode that tracks the elimination of the core volume loss and adjusts the output displacement based on rheological impedance.

[0010] Furthermore, in S4, deriving the molecular chain shear stress relaxation time based on the three-dimensional temperature field and constructing the complex rheological impedance compensation function includes: Based on the Arrhenius equation, the time-temperature equivalence factor corresponding to the three-dimensional temperature field is calculated, and the molecular chain shear stress relaxation time of the closed core phase transition region is derived. By combining the molecular chain shear stress relaxation time with the dynamic shear viscosity of the melt in the closed core phase transition zone, a complex rheological impedance compensation function is constructed to quantify the flow resistance gain of the melt under dynamic disturbance.

[0011] Furthermore, in S6, the parameters of the reduced-order physical twin model are corrected by comparing the tonnage response predicted by the unscented Kalman filter algorithm with those predicted by the physical twin model, including: The thermal conductivity, thermal expansion coefficient, and crystallization kinetics parameters in the physical reduced-order twin model are used to construct an amplified state vector. The Sigma point set is generated using unscented transformation, input into the physical reduced-order twin model to predict the tonnage response, and the measurement residual between the tonnage response and the tonnage reaction force response curve is calculated. The amplified state vector is updated based on the measurement residuals and gain matrix to correct the parameters of the physical reduced-order twin model online.

[0012] Furthermore, the preset phase transition range is a crystallinity of 0.15 to 0.85, and the preset hardening threshold is a crystallinity of 0.85.

[0013] Furthermore, the critical switching condition is that the negative pressure evolution index reaches or exceeds 0.35.

[0014] The present invention has the following beneficial effects: 1. In this invention, by combining the three-dimensional crystallinity field to identify the phase transition region of the closed core and using the negative pressure evolution index to judge the critical state of cavitation instability in real time, the adaptive and precise switching of the control mode is realized, which solves the problem that traditional wall pressure control cannot sense and eliminate the shrinkage negative pressure of the core, which leads to the formation of shrinkage cavities and voids inside the product.

[0015] 2. In this invention, a high-frequency micro-pulse displacement command is generated by modulating a high-frequency sinusoidal carrier based on a complex rheological impedance compensation function. This realizes the reduction of flow resistance in the phase change region by utilizing the micro-vibration shear thinning effect, and solves the problems of the extremely high melt viscosity of ultra-high molecular weight polyethylene, which leads to difficulties in traditional macro-feeding, easy chain damage, and high equipment tonnage requirements.

[0016] 3. In this invention, by using a physical reduced-order twin model to reconstruct the internal microscopic field quantities, and by using an unscented Kalman filter algorithm to compare the tonnage residual to correct the model parameters, real-time visibility of the molding microscopic field quantities and online closed-loop correction of the model's physical property parameters are achieved, solving the problems of invisible internal state and inaccurate temperature and pressure control caused by the drastic drift of polymer physical property parameters with phase transition. Attached Figure Description

[0017] Figure 1 The flowchart shows the UHMW-PE molding temperature and pressure decoupling adaptive control method based on digital twin proposed in this invention. Figure 2 This is a flowchart illustrating the identification of the closed core phase transition region and the calculation of volume shrinkage in the UHMW-PE molding temperature and pressure decoupling adaptive control method based on digital twin proposed in this invention. Figure 3 This is a logic diagram for calculating the negative pressure evolution index and generating micro-compensation commands for the UHMW-PE molding temperature and pressure decoupling adaptive control method based on digital twin proposed in this invention. Figure 4 This is a logic diagram of the online parameter correction using an unscented Kalman filter for the UHMW-PE molding temperature and pressure decoupling adaptive control method based on digital twin proposed in this invention. Detailed Implementation

[0018] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] In embodiments of the present invention, the present invention provides a UHMW-PE molding temperature and pressure decoupling adaptive control method based on digital twins, such as... Figures 1-4 As shown, it includes the following steps: S1. Construct a physical reduced-order twin model, collect real-time mold wall temperature, wall pressure and indenter displacement data of the molding press and input them into the physical reduced-order twin model to reconstruct the three-dimensional temperature field and three-dimensional crystallinity field inside the billet. Furthermore, in S1, constructing the physically reduced twin model includes: High-order finite element simulation data of the molding process were collected, and the reduced-order spatial basis functions of the flow field and temperature field were extracted using the intrinsic orthogonal decomposition algorithm. A physical constraint surrogate model based on Gaussian process regression is constructed, and the surrogate model is used to establish the nonlinear mapping relationship between the real-time mold wall temperature, wall pressure and indenter displacement data and the coefficients of the reduced-order space basis function online; Based on the coefficients of the reduced-order spatial basis function obtained in real time, the reduced-order spatial basis function is linearly superimposed and reconstructed to output the three-dimensional temperature field and three-dimensional crystallinity field inside the billet in real time.

[0020] Specifically, in the implementation of constructing the physical reduced-order twin model, high-order finite element numerical simulation is first performed offline to simulate the unsteady heat conduction and viscoelastic flow field changes during the UHMW-PE compression molding process. The high-order simulation discretizes the spatial continuous region of the compression molding blank into a three-dimensional finite element mesh, and extracts a large amount of nodal temperature data and flow field velocity data at multiple time steps under multiple different combinations of process parameters to form a high-dimensional simulation data matrix.

[0021] An intrinsic orthogonal decomposition algorithm is performed on the high-dimensional simulation data matrix. First, the time mean vector is subtracted from the node data at each time step to construct a zero-mean pulsating field matrix. The covariance matrix of this matrix is ​​calculated, and its eigenvalues ​​and eigenvectors are solved. The top k eigenvectors with an energy percentage exceeding a preset threshold, such as 99.9%, are extracted as reduced-order spatial basis functions. The reduced-order spatial basis functions constitute a low-dimensional subspace capable of characterizing the main dynamic features of the complex thermal flow field inside the billet.

[0022] After obtaining the reduced-order spatial basis functions, a physical constraint surrogate model based on Gaussian process regression is constructed. The input feature vector of the surrogate model includes the real-time collected mold wall temperature, wall pressure, and indenter displacement of the mold press. During feature construction, the mold wall temperature change rate, wall pressure gradient, and displacement rate at three consecutive sampling times are combined into the input feature matrix. The output feature of the surrogate model is the coefficient vector of the reduced-order spatial basis functions at the corresponding time. The Gaussian process regression model adopts the standard squared exponential kernel function, and energy conservation and momentum conservation physical residuals are introduced as penalty terms during hyperparameter training to constrain the predicted output of the surrogate model, ensuring that it satisfies the basic laws of fluid thermodynamics while fitting the data mapping relationship. Before online operation, the hyperparameters of the surrogate model are optimized using a small number of representative samples obtained from offline simulation to determine the hyperparameter combination of the mean function and kernel function.

[0023] During the online control phase, the sampling system acquires real-time data on the mold wall temperature (measured by the mold wall sensor), the wall pressure (measured by the side sensor of the indenter), and the indenter displacement (measured by the grating ruler) at intervals of hundreds of milliseconds. These physical quantities are combined to construct an input feature vector, which is then directly input into a pre-trained Gaussian process regression physical constraint surrogate model. The surrogate model uses algebraic operations on the kernel matrix and the mean function to calculate and output the coefficient vector of the reduced-order spatial basis function corresponding to the current moment in real time.

[0024] After obtaining the coefficient vector of the reduced-order space basis functions, matrix multiplication and linear superposition operations are performed on it with the reduced-order space basis functions extracted and stored in the offline stage. For the reconstruction of the three-dimensional temperature field, the following reconstruction formula is used: ; in, This represents the reconstructed three-dimensional temperature field node vector. This represents the time-averaged vector of temperatures across all nodes in the offline dataset. This indicates the output of the proxy model at the current time. The coefficients of the reduced-order space basis functions Indicates the first A temperature-reduced spatial basis function This indicates the truncation order of the reduced-order space basis functions used for reconstruction. In a specific implementation case... The value is 8.

[0025] The reconstructed three-dimensional temperature field contains the real-time temperature values ​​of each grid node within the billet. This three-dimensional temperature field is input as intermediate data into the crystallization kinetics evolution calculation logic. Based on the Nakamura crystallization kinetic equation, the crystallinity evolution of each grid node under the current temperature distribution and cooling rate is calculated. By numerical integration over the time dimension, the real-time crystallinity of each node is calculated in real time, and then a complete three-dimensional crystallinity field inside the billet is assembled and output. The reconstructed three-dimensional temperature field and three-dimensional crystallinity field serve as data sources for subsequent phase transformation zone determination and volume shrinkage calculation, continuously participating in the real-time calculation of subsequent control cycles.

[0026] S2. Determine the closed core phase transition region based on the three-dimensional crystallinity field. Mark the grid nodes in the three-dimensional crystallinity field whose crystallinity is within the preset phase transition range as the phase transition front transition region. When the phase transition front transition region is completely surrounded in three-dimensional space by an outer solidification envelope surface with crystallinity higher than the preset hardening threshold, the surrounded phase transition front transition region and the internal uncrystallized region are determined as the closed core phase transition region. Solve for the volume shrinkage rate change rate and core volume defect of the closed core phase transition region based on the PVT equation of state. Furthermore, in S2, the solution for the volume shrinkage rate change and core volume defect of the closed core phase transition region based on the PVT equation of state includes: Based on the three-dimensional temperature field and wall pressure, the specific volume of each grid cell in the closed core phase transition region is calculated in real time using the dual-domain Tait PVT equation of state. The rate of change of volume shrinkage of each grid cell is determined by the differential of the specific volume with respect to time. The volume loss caused by phase transition contraction in each grid cell within the closed core phase transition region is accumulated by three-dimensional spatial integration to generate the core volume loss.

[0027] Specifically, after obtaining the reconstructed 3D crystallinity field, the computational logic enters the closed core phase transition region identification stage. The system traverses each node in the 3D finite element mesh and extracts the real-time crystallinity values ​​of the nodes. The lower limit of the preset phase transition interval is 0.15, and the upper limit is 0.85. When the crystallinity of a certain mesh node is between 0.15 and 0.85, the system marks the node and determines it as a phase transition front transition region node. At the same time, the preset hardening threshold is 0.85. The system uses a spatial connectivity analysis algorithm to search for closed 3D envelope surfaces formed by mesh nodes with crystallinity greater than 0.85. The spatial topology search logic scans layer by layer from the outer surface of the billet to the inside. If the set of nodes marked as phase transition front transition regions is detected to be completely wrapped by hardened layers with crystallinity greater than 0.85 in all three axes of 3D space, and there are no high-flow channels extending to the outer surface of the billet, then the phase transition front transition region nodes inside the closed envelope surface and the uncrystallized melt nodes with crystallinity less than 0.15 together constitute the closed core phase transition region. The system assembles and stores the grid cell index numbers belonging to the closed core phase transition region as the phase transition region grid set for the current control cycle.

[0028] For the identified closed core phase transition region, the system reads the three-dimensional temperature field data of the corresponding mesh element obtained from the aforementioned reconstruction, as well as the wall pressure data measured by the molding press at the current moment, and calculates the specific volume of each mesh element based on the dual-domain Tait PVT state equation. The dual-domain Tait PVT state equation is as follows: ; in, Indicates temperature With pressure The specific volume of the lower grid cell, in cubic meters per kilogram; The specific volume under zero pressure is represented by a temperature polynomial function of the polymer melt and solid regions. This is a universal dimensionless constant with a value of 0.0894; It is a function that expresses sensitivity to pressure and reflects the compressibility of polymers; This is an additional specific volume decay term for the crystallization phase transition. This represents the real-time temperature of the current grid cell. This refers to the wall pressure collected in real time.

[0029] After obtaining the volume ratio of each grid cell at the current sampling time to the previous sampling time, the differential of the volume ratio with respect to time is solved using the first-order forward difference method, and the volume shrinkage rate of each grid cell is calculated using normalization. ; in, This represents the rate of change of the volume shrinkage rate of the grid cell; This represents the specific volume of the grid cell at the current moment. Indicates the time sampling interval, within the current control cycle. The sampling interval is 0.01 seconds.

[0030] Subsequently, the system performs spatial integration and summation on all elements in the mesh set of the closed core phase transition region. For a single mesh element, the microscopic volume defect caused by the phase transition in the current time step is calculated based on the difference between its initial mass and its volume relative to the current and previous times. The volume defects of all mesh elements in the closed core phase transition region are then summed in three-dimensional space, using the following formula: ; in, This indicates the volume loss in the core, expressed in cubic meters. This represents the set of grid cells contained in the closed core phase transition region; Indicates the grid cell number in the set; and They represent the first Density and geometric volume of individual mesh cells in the initial reference state; and They represent the first The volume ratio of each grid cell between the previous and current time steps. The calculated core volume loss and volume shrinkage rate change are directly output to the subsequent negative pressure index evolution judgment logic and compensation instruction generation module.

[0031] S3. Solve for the shear stress relaxation rate of the closed core phase transition region based on the three-dimensional temperature field, solve for the dimensionless relaxation-contraction criterion based on the ratio of the volume shrinkage rate change rate to the shear stress relaxation rate, solve for the microscopic cavitation damage tensor of the closed core phase transition region based on the criterion, and map the principal component of the microscopic cavitation damage tensor to the negative pressure evolution index; when the negative pressure evolution index meets the critical switching condition, switch to the microscopic compensation control mode. Furthermore, in S3, the shear stress relaxation rate of the closed core phase transition region is solved based on the three-dimensional temperature field, including: Based on the Arrhenius equation or the WLF equation, the temperature shift factor and relaxation time of the closed core phase transition region are calculated according to the three-dimensional temperature field. The shear relaxation modulus at the current temperature is derived using the generalized Maxwell rheological model, and the shear stress relaxation rate is solved by combining the shear strain rate of the phase transformation region in the closed core.

[0032] Furthermore, in S3, switching to the micro-compact control mode includes: Disconnect the static constant pressure holding control circuit that relies primarily on feedback from the pressure sensor on the molding machine wall. Switch to an adaptive servo displacement control mode that tracks the elimination of core volume loss and adjusts the output displacement based on rheological impedance.

[0033] Specifically, after obtaining the reconstructed three-dimensional temperature field and the set of closed core phase transition regions, the data processing system performs rheological parameter evolution calculations for each grid node within the closed core phase transition region. First, the real-time average temperature and reference temperature of the closed core phase transition region are extracted. If the current temperature is higher than the polymer glass transition temperature and the region is in the melt flow region, the Williams-Randall-Ferry (WLF) equation is used to solve for the temperature shift factor. The calculation formula is as follows: ; in, Indicates the temperature offset factor; This represents the three-dimensional average temperature of the closed core phase transition region at the current moment, in Kelvin (K). This indicates the preset reference temperature, with a value of 463.15K; and These are empirical constants for materials. In the current embodiment, The value is 8.86. The value is set to 101.6 K. The relaxation time of the closed core phase transition region at the current temperature is obtained by multiplying the calculated temperature offset factor by the main relaxation time at the reference temperature. ; in, This represents the relaxation time of molecular chain shear stress at the current temperature; This represents the initial relaxation time at the reference temperature, with a value of 12.5 seconds.

[0034] The derived relaxation times are input into the generalized Maxwell rheological model to solve for the dynamic shear relaxation modulus at the current temperature. In this embodiment, the generalized Maxwell rheological model uses three parallel Maxwell elements with relaxation times of [missing information]. s、 s、 s, the corresponding relaxation moduli are respectively Pa、 Pa、 Pa. By combining the real-time monitored indentation velocity and spatial gradient to calculate the shear strain rate of the closed core phase transformation region, and differentiating the product of the shear relaxation modulus and the shear strain rate with respect to time, the shear stress relaxation rate of the closed core phase transformation region is calculated.

[0035] Subsequently, the system reads the previously obtained rate of change of volumetric shrinkage and the currently calculated shear stress relaxation rate, constructs the algebraic ratio between the two, and solves for the dimensionless relaxation-shrinkage coefficient. The calculation formula is as follows: ; in, This represents the dimensionless relaxation-contraction coefficient; This represents the rate of change of the average volume shrinkage rate of the closed core phase transition region obtained above; The relaxation time of molecular chain shear stress at the current temperature; This is the equivalent shear strain rate of the closed core phase transition region. This criterion characterizes the intensity of the non-equilibrium competition between the stress accumulation rate induced by phase transition volume shrinkage and the rate of molecular chain de-entanglement and relaxation.

[0036] In this embodiment, a simplified cavitation damage mapping model is used to define the first principal component of the microscopic cavitation damage tensor. Equal to the dimensionless relaxation-contraction coefficient ,Right now: ; Will The value is directly assigned as a negative pressure evolution index, that is: ; in, This indicates the negative pressure evolution index.

[0037] The data processing system performs real-time threshold comparison and judgment on the negative pressure evolution index. The system presets the cavitation instability critical value to be 0.35. When the calculated negative pressure evolution index reaches or exceeds 0.35, the control logic determines that the current closed core phase transition region has reached the critical state of microscopic tensile instability and triggers the critical switching signal.

[0038] Upon receiving the critical switching signal, the controller executes the smooth switching logic of the control mode. First, it disconnects the closed-loop PID control channel from the molding press wall pressure sensor in the main control loop to the hydraulic servo system, stopping the static constant pressure holding control aimed at maintaining the external wall pressure. Simultaneously, it enables the servo displacement adaptive control channel, switching the control mode to an adaptive servo displacement control mode that tracks the elimination of the previously calculated core volume loss and adjusts the output displacement based on rheological impedance. In this mode, the output variable of the control system changes from the tonnage setpoint to the high-frequency pulse displacement of the pressure head, providing the control loop basis for the generation and execution of subsequent micro-compensation commands.

[0039] S4. Derive the molecular chain shear stress relaxation time from the three-dimensional temperature field and construct a complex rheological impedance compensation function. Calculate the target shrinkage displacement amplitude based on the core volume defect. Use the complex rheological impedance compensation function to perform phase and gain modulation on the target shrinkage displacement amplitude and superimpose a high-frequency oscillation carrier to generate a displacement control command containing a high-frequency micro-amplitude pulse waveform. Furthermore, in S4, the relaxation time of molecular chain shear stress is derived from the three-dimensional temperature field, and a complex rheological impedance compensation function is constructed, including: Based on the Arrhenius equation, the time-temperature equivalence factor corresponding to the three-dimensional temperature field is calculated, and the relaxation time of molecular chain shear stress in the closed core phase transition region is derived. By combining the molecular chain shear stress relaxation time with the dynamic shear viscosity of the melt in the closed core phase transition zone, a complex rheological impedance compensation function is constructed to quantify the flow resistance gain of the melt under dynamic disturbance.

[0040] Specifically, after entering the adaptive servo displacement control mode, the system extracts real-time three-dimensional temperature field data of the closed core phase transition region and calculates the average nodal temperature of this region. If the current temperature is lower than the rheological transition temperature of the material in its molten state, the system uses the Arrhenius equation to calculate the mapping relationship between temperature change and molecular chain relaxation behavior, and calculates the time-temperature equivalence factor. ; in, Indicates the time-temperature equivalence factor; This represents the activation energy for the flow of the UHMW-PE molecular chain, with a value of 48.5 kJ / mol. This represents the molar gas constant, with a value of 8.314 J / (mol·K); This indicates the average temperature of the phase transition region in the closed core at the current moment; The reference temperature is 473.15 K. The system multiplies the obtained time-temperature equivalence factor with the characteristic relaxation time of the molecular chain at the reference temperature to calculate the shear stress relaxation time of the molecular chain in the closed core phase transition region at the current temperature.

[0041] Subsequently, the system reads the previously derived molecular chain shear stress relaxation time and the dynamic shear viscosity of the melt at the current temperature, and constructs a complex rheological impedance compensation function in the frequency domain. The complex rheological impedance compensation function is composed of a real damping component and an imaginary elastic recovery component, and the calculation formula is as follows: ; in, Represents the complex rheological impedance compensation function; This represents the dynamic shear viscosity of the melt in the closed core phase change zone at zero shear rate, and is taken as a value. Pa·s; The imaginary unit; Indicates the applied perturbation carrier angular frequency; This is the relaxation time of the molecular chain shear stress obtained above. This function quantifies the flow resistance phase lag and amplitude attenuation gain of the melt due to the viscoelastic response when subjected to high-frequency waveform perturbations.

[0042] During the control command generation phase, the system first extracts the previously calculated core volume deficit and the geometric cross-sectional area of ​​the molding die cavity. Then, using the algebraic ratio between the geometric volume and the cross-sectional area, it calculates the target compensation displacement amplitude. ; in, Indicates the magnitude of the target compensation displacement; This represents the core volume loss obtained above; This represents the effective cross-sectional area of ​​the mold cavity of the molding machine, with a value of 0.25 square meters.

[0043] After obtaining the target compensation displacement amplitude, the data processing module extracts the magnitude and complex angle of the complex rheological impedance compensation function at a preset carrier frequency, and performs phase lead correction and amplification gain modulation on the target compensation displacement amplitude. Subsequently, the system generates a high-frequency sinusoidal oscillation carrier with a frequency of 25 Hz, and superimposes the dynamically corrected amount after gain and phase modulation onto the target compensation displacement amplitude to reconstruct and generate a high-frequency micropulse displacement command. ; in, This indicates a high-frequency micropulse displacement command; The target compensation displacement amplitude calculated above; This represents the amplitude amplification factor of the complex rheological impedance compensation function in the frequency domain; This represents the angular frequency of a sinusoidal carrier wave, calculated as follows: The product of the carrier frequency and the 25Hz frequency; For the continuous time variables of the micro-compaction stage; The phase compensation angle is calculated using the arctangent function of the ratio of the imaginary to the real part of the complex rheological impedance compensation function.

[0044] The high-frequency micro-pulse displacement command generated by the calculation is directly output to the underlying drive controller of the servo control system. The drive controller adjusts the opening degree and action timing of the hydraulic servo valve in real time according to the command waveform.

[0045] S5. Send the high-frequency micro-pulse displacement command to the servo drive actuator of the molding machine to control the action of the pressure head to compensate for the phase change region of the closed core. Specifically, during the control command issuance and execution phase, the data processing system converts the high-frequency micro-pulse displacement command, which includes a high-frequency micro-amplitude pulse waveform, generated by the aforementioned calculations, into a digital signal. First, the system extracts the discrete displacement values ​​of the high-frequency micro-pulse displacement command at the current sampling moment and converts them into a data frame format conforming to the industrial fieldbus protocol. The digitally encoded displacement command is then periodically sent to the digital servo driver of the molding machine's servo drive actuator via the Ethernet control automation technology bus. The communication refresh frequency for the current control cycle is set to 1000 Hz.

[0046] After receiving the displacement command, the digital servo driver calculates the difference between the value and the absolute displacement value fed back in real time by the high-precision grating ruler installed on the side of the pressure head servo hydraulic cylinder, and calculates the displacement tracking deviation at the current moment: ; in, Indicates displacement tracking deviation; This indicates the high-frequency micropulse displacement command received at the current sampling moment; This indicates the actual displacement of the indenter measured in real time by the grating ruler.

[0047] The proportional-integral-derivative controller embedded in the servo driver extracts the displacement tracking deviation and, combined with the linear flow-voltage transmission characteristics of the electro-hydraulic servo valve, calculates and generates the servo valve control current signal: ; in, This indicates the drive current output to the electro-hydraulic servo valve coil; This is the proportional gain, with a value of 50 mA / m; The integral gain is 10 mA / (m·s); The differential gain is 0.5 mA·s / m. This is the time variable for integration.

[0048] The driving current is input to the high-frequency electro-hydraulic servo valve, which adjusts the flow rate and direction of hydraulic oil entering the rodless and rod chambers of the servo hydraulic cylinder by changing the axial displacement of the valve core. The hydraulic cylinder drives the pressure head to generate mechanical action consistent with the high-frequency micro-pulse displacement command. The pressure head superimposes a high-frequency micro-amplitude oscillation motion with an amplitude of 0.05 mm and a frequency of 25 Hz on the basis of the main propulsion displacement.

[0049] The high-frequency micro-amplitude pulses applied by the indenter transmit shear waves to the closed core phase change zone through the hardened outer shell of the mold. This high-frequency shear disturbance alters the rheological state of the entangled network of ultra-high molecular weight polyethylene (UHMWPE) polymer chains, significantly reducing the local shear viscosity of the melt. Simultaneously, the main thrust displacement of the indenter forces the semi-molten polymer in the transition zone at the phase change front continuously into the closed core phase change zone, filling the microscopic free volume loss caused by the shrinkage during the crystallization phase change. A grating ruler continuously feeds back the real-time absolute displacement data of the indenter to the servo driver, forming a high-frequency closed-loop position control flow. This ensures that the indenter displacement accurately tracks the micro-pulse command waveform until the closed core phase change zone completes full crystallization and solidification.

[0050] S6. Collect the tonnage reaction force response curve of the servo-driven actuator, compare it with the tonnage response predicted by the physical reduced-order twin model using the unscented Kalman filter algorithm, and correct the parameters of the physical reduced-order twin model in reverse.

[0051] Furthermore, in S6, the unscented Kalman filter algorithm is used to compare the tonnage response predicted by the physical reduced-order twin model, and the parameters of the physical reduced-order twin model are corrected in reverse, including: The thermal conductivity, thermal expansion coefficient, and crystallization kinetics parameters in the physical reduced-order twin model are constructed as an amplified state vector; The Sigma point set is generated using unscented transformation, input into the physical reduced-order twin model to predict the tonnage response, and the measurement residuals of the tonnage response and tonnage reaction force response curves are calculated. The parameters of the physical reduced-order twin model are corrected online by updating the amplified state vector based on the measurement residual and gain matrix.

[0052] Specifically, while the servo-driven actuator performs the compensation action, the system uses a pressure sensor mounted on the top of the servo hydraulic cylinder to collect real-time data on the tonnage reaction force of the pressure head at a sampling frequency of 100 Hz, and continuously assembles this data into a measured tonnage reaction force response curve over time. This measured curve serves as the data source for measurement updates and is transmitted in real-time to the online parameter correction module of the physical reduced-order twin model.

[0053] The online parameter correction module first constructs the augmented state vector for the unscented Kalman filter algorithm. The augmented state vector contains key physical parameters from the physical reduced-order twin model and internal system state parameters, as shown below: ; in, Indicates the first The amplified state vector at each sampling time; The value represents the thermal conductivity of ultra-high molecular weight polyethylene material, with an initial value of 0.42 W / (m·K). This represents the coefficient of thermal expansion, with an initial value of [value missing]. K ; This represents the crystallization kinetic rate constant, initially set to 0.025 s⁻¹. ; This represents the temperature of a representative node in the state vector. This indicates the representative degree of crystallinity at the nodes.

[0054] The data processing module performs an unscented transformation on the current time-incremented state vector and its covariance matrix, generating a symmetric set of Sigma points. This is done based on the dimension of the expanded state vector. (here) (Equal to 5), the generated Sigma point is calculated using the following formula: ; in, Indicates the generated first Sigma point vectors; This represents the posterior estimate of the expanded state vector at the previous time step; This represents the dimension of the expanded state vector, with a value of 5. This is a scalar scaling parameter with a value of 2; This represents the estimated covariance matrix of the amplified state vector at the previous time step; The first square root of a matrix represents the square root of the matrix. Column vector.

[0055] After generating the Sigma point set, the system substitutes the physical parameters extracted from each Sigma point into the physical reduced-order twin model, combines the current displacement and temperature inputs, performs forward state prediction and physical mechanics calculation, and derives the forward predicted tonnage response value corresponding to each Sigma point. The tonnage response values ​​predicted from all Sigma points are then weighted and summed according to preset weighting coefficients to calculate the average forward predicted tonnage response value of the model at the current moment.

[0056] Subsequently, the system calculates the difference between the real-time collected measured tonnage reaction force data and the predicted tonnage response mean to generate the measurement residual: ; in, Indicates the first The tonnage measurement residual at any given time; This indicates the tonnage reaction force value measured in real time by the pressure sensor; This represents the mean tonnage response calculated by forward prediction of the Sigma point set using a physically reduced-order twin model.

[0057] The system derives the Kalman gain matrix by calculating the cross-covariance matrix between the state vector and the measured predicted value, and dividing it by the autocovariance matrix of the measured predicted value. The Kalman gain matrix is ​​then compared with the measurement residuals. Multiplication is performed to update the a posteriori of the expanded state vector: ; in, This represents the posterior estimate of the updated amplified state vector; This represents the mean of the state vector predicted one step ahead. This represents the calculated Kalman gain matrix.

[0058] Through the above algebraic update process, the thermal conductivity in the amplified state vector is corrected in real time. Coefficient of thermal expansion and the crystallization kinetic rate constant The updated physical parameters immediately overwrite the corresponding parameter matrix in the physical reduced-order twin model, enabling the reduced-order twin model to output a more accurate three-dimensional temperature field and three-dimensional crystallinity field based on the corrected real physical property parameters in the next control cycle, thus achieving online closed-loop self-calibration of the twin model.

[0059] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A UHMW-PE molding temperature and pressure decoupling adaptive control method based on digital twin, characterized in that, include: S1. Construct a physical reduced-order twin model, collect real-time mold wall temperature, wall pressure and press head displacement data of the molding press and input them into the physical reduced-order twin model to reconstruct the three-dimensional temperature field and three-dimensional crystallinity field inside the billet; S2. Determine the closed core phase transition region based on the three-dimensional crystallinity field, and mark the grid nodes in the three-dimensional crystallinity field whose crystallinity is within the preset phase transition range as the phase transition front transition region; when the phase transition front transition region is completely surrounded in three-dimensional space by an outer solidification envelope surface with crystallinity higher than the preset hardening threshold, the surrounded phase transition front transition region and the internal uncrystallized region are determined as the closed core phase transition region; solve the volume shrinkage rate change rate and core volume defect of the closed core phase transition region based on the PVT state equation; S3. Solve for the shear stress relaxation rate of the closed core phase transition region based on the three-dimensional temperature field, solve for the dimensionless relaxation-contraction criterion based on the ratio of the volume shrinkage rate change rate to the shear stress relaxation rate, solve for the microscopic cavitation damage tensor of the closed core phase transition region based on the criterion, and map the principal component of the microscopic cavitation damage tensor to a negative pressure evolution index; when the negative pressure evolution index meets the critical switching condition, switch to the microscopic compensation control mode; S4. Derive the molecular chain shear stress relaxation time based on the three-dimensional temperature field and construct a complex rheological impedance compensation function. Calculate the target shrinkage displacement amplitude based on the core volume loss. Use the complex rheological impedance compensation function to perform phase and gain modulation on the target shrinkage displacement amplitude and superimpose a high-frequency oscillation carrier to generate a displacement control command containing a high-frequency micro-amplitude pulse waveform. S5. Send the high-frequency micro-pulse displacement command to the servo drive actuator of the molding machine to control the action of the pressure head to compensate for the shrinkage of the closed core phase change region; S6. Collect the tonnage reaction force response curve of the servo drive actuator, compare it with the tonnage response predicted by the physical reduced-order twin model using the unscented Kalman filter algorithm, and correct the parameters of the physical reduced-order twin model in reverse.

2. The UHMW-PE molding temperature and pressure decoupling adaptive control method based on digital twin as described in claim 1, characterized in that, In S1, constructing a physically reduced twin model includes: High-order finite element simulation data of the molding process were collected, and the reduced-order spatial basis functions of the flow field and temperature field were extracted using the intrinsic orthogonal decomposition algorithm. A physical constraint surrogate model based on Gaussian process regression is constructed, and the surrogate model is used to establish online the nonlinear mapping relationship between the real-time mold wall temperature, wall pressure and indenter displacement data and the coefficients of the reduced-order space basis function; The coefficients of the reduced-order spatial basis function obtained by real-time calculation are linearly superimposed and reconstructed with the reduced-order spatial basis function to output the three-dimensional temperature field and three-dimensional crystallinity field inside the billet in real time.

3. The UHMW-PE molding temperature and pressure decoupling adaptive control method based on digital twin as described in claim 1, characterized in that, In S2, the solution for the volume shrinkage rate change rate and core volume defect of the closed core phase transition region based on the PVT state equation includes: Based on the three-dimensional temperature field and wall pressure, the specific volume of each grid cell in the closed core phase transition region is calculated in real time using the dual-domain Tait PVT equation of state. The rate of change of volume shrinkage of each grid cell is determined by the differential of the specific volume with respect to time. The volume loss caused by phase transition shrinkage in each grid cell within the closed core phase transition region is accumulated by three-dimensional spatial integration to generate the core volume loss.

4. The UHMW-PE molding temperature and pressure decoupling adaptive control method based on digital twin as described in claim 1, characterized in that, In S3, the shear stress relaxation rate of the closed core phase transition region is calculated based on the three-dimensional temperature field, including: Based on the Arrhenius equation or the WLF equation, the temperature shift factor and relaxation time of the closed core phase transition region are calculated according to the three-dimensional temperature field. The shear relaxation modulus at the current temperature is derived using the generalized Maxwell rheological model, and the shear stress relaxation rate is solved by combining it with the shear strain rate of the phase transformation region of the closed core.

5. The UHMW-PE molding temperature and pressure decoupling adaptive control method based on digital twin as described in claim 1, characterized in that, In S3, switching to the micro-compact control mode includes: Disconnect the static constant pressure holding control circuit that relies primarily on feedback from the pressure sensor on the molding machine wall. Switch to an adaptive servo displacement control mode that tracks the elimination of the core volume loss and adjusts the output displacement based on rheological impedance.

6. The UHMW-PE molding temperature and pressure decoupling adaptive control method based on digital twin as described in claim 1, characterized in that, In S4, deriving the molecular chain shear stress relaxation time based on the three-dimensional temperature field and constructing the complex rheological impedance compensation function includes: Based on the Arrhenius equation, the time-temperature equivalence factor corresponding to the three-dimensional temperature field is calculated, and the molecular chain shear stress relaxation time of the closed core phase transition region is derived. By combining the molecular chain shear stress relaxation time with the dynamic shear viscosity of the melt in the closed core phase transition zone, a complex rheological impedance compensation function is constructed to quantify the flow resistance gain of the melt under dynamic disturbance.

7. The UHMW-PE molding temperature and pressure decoupling adaptive control method based on digital twin as described in claim 1, characterized in that, In S6, the parameters of the reduced-order physical twin model are corrected by comparing the tonnage response predicted by the unscented Kalman filter algorithm with those predicted by the physical twin model. The thermal conductivity, thermal expansion coefficient, and crystallization kinetics parameters in the physical reduced-order twin model are used to construct an amplified state vector. The Sigma point set is generated using unscented transformation, input into the physical reduced-order twin model to predict the tonnage response, and the measurement residual between the tonnage response and the tonnage reaction force response curve is calculated. The amplified state vector is updated based on the measurement residuals and gain matrix to correct the parameters of the physical reduced-order twin model online.

8. The UHMW-PE molding temperature and pressure decoupling adaptive control method based on digital twin as described in claim 1, characterized in that, The preset phase transition range is a crystallinity of 0.15 to 0.85, and the preset hardening threshold is a crystallinity of 0.

85.

9. The UHMW-PE molding temperature and pressure decoupling adaptive control method based on digital twin as described in claim 1, characterized in that, The critical switching condition is that the negative pressure evolution index reaches or exceeds 0.35.

Images