A method for preparing a photocurable two-component 3D printing elastic material

By real-time monitoring of the two-component flow rate and dynamic adjustment of exposure and layer thickness, and by using a physical information neural network to predict the curing shrinkage field, the problem of difficulty in real-time prediction and compensation of curing shrinkage behavior in 3D printing of photocurable two-component elastic materials is solved, thereby improving the interlayer bonding strength and mechanical properties of the molded parts.

CN122242143BActive Publication Date: 2026-11-10DONGGUAN JINGGE INTELLIGENT MANUFACTURING TECHNOLOGY CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202610355643.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2026-03-23
Publication Date
2026-11-10
Estimated Expiration
2046-03-23

AI Technical Summary

Technical Problem

In the 3D printing process of photocurable two-component elastic materials, the curing shrinkage behavior is difficult to predict and compensate in real time, resulting in insufficient interlayer bonding strength, warping and cracking risk of the molded parts.

Method used

A Coriolis mass flow meter is used to monitor the flow rate of the two components in real time. The curing shrinkage field is predicted by combining a model predictive control algorithm and a physical information neural network. The effects of oxygen inhibition are eliminated by a ternary thiol-olefin synergistic curing system and a hindered amine light stabilizer. The exposure amount and layer thickness are dynamically adjusted to achieve interlayer active double bond copolymerization and form a continuous transition region of chemical bonds.

Benefits of technology

It enables real-time prediction and compensation of shrinkage field in photocurable two-component elastic materials, improves interlayer bonding strength, reduces the risk of warping and cracking, and ensures the uniformity of mechanical properties of molded parts.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122242143B_ABST
    Figure CN122242143B_ABST
Patent Text Reader

Abstract

This invention provides a method for preparing photocurable two-component 3D printed elastic materials, belonging to the technical field of photocurable two-component 3D printed elastic material preparation. This invention achieves high-precision mixing ratio control of the two-component resin through a Coriolis mass flow meter and model predictive control algorithm. Simultaneously, it uses a Fick diffusion-reaction coupled curing depth iterative solution algorithm to accurately predict the curing profile. By dynamically adjusting the low-dose supplemental exposure and layer thickness, a cross-layer copolymer active interface layer is formed at the interlayer interface. Furthermore, the equivalent shrinkage stress output by an artificial intelligence model drives the interface quality assessment function to adjust process parameters in a closed loop. Finally, a post-curing treatment eliminates residual active double bonds. This invention solves the technical problems of difficulty in real-time prediction and compensation of curing shrinkage behavior during the 3D printing of photocurable two-component elastic materials, leading to insufficient interlayer interface bonding strength and a high risk of warping and cracking in the molded parts.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of photocurable two-component 3D printing elastic material preparation technology, specifically, it relates to a method for preparing photocurable two-component 3D printing elastic materials. Background Technology

[0002] Photopolymer 3D printing technology enables the rapid prototyping of complex elastic components by curing liquid resin layer by layer through exposure, and has wide applications in medical devices, flexible electronics, and soft robotics. In current photopolymer printing of two-component elastic resin systems, two peristaltic pumps or syringe pumps are typically used to supply materials separately, which are then combined in a static mixer. The resin is cured layer by layer with fixed exposure parameters, and the interlayer interfaces rely on the natural wetting and adhesion of the resin. The overall process lacks the ability to online sense and dynamically compensate for the curing shrinkage field.

[0003] However, the volume change caused by polymerization shrinkage during the curing process of two-component elastic resin systems exhibits significant spatial non-uniformity. The shrinkage rate field varies drastically under different layer thicknesses, exposure levels, and formulation parameters. Traditional finite element offline simulation methods take minutes to hours to calculate, making it impossible to complete the real-time calculation of the single-layer shrinkage field within the printing cycle. This results in the printing path not being able to dynamically compensate based on the layer-by-layer shrinkage behavior. At the same time, the oxygen inhibition effect leads to a low conversion rate of active double bonds at the interlayer interface. Traditional fixed exposure strategies cannot accurately control the proportion of residual active double bonds at the interface, resulting in insufficient interlayer covalent crosslinking density.

[0004] In current photopolymer elastomer printing, the lack of a closed-loop control scheme that combines a real-time prediction model of the curing shrinkage field with a synergistic regulation mechanism of interlayer interfacial active double bonds makes the molded parts highly susceptible to interlayer delamination, warping, and cracking defects under the cumulative shrinkage stress layer by layer, making it difficult to meet the interlayer fatigue life requirements under large strain cyclic loading. In other words, existing technologies suffer from the technical problem of difficulty in real-time prediction and compensation of curing shrinkage behavior during the 3D printing of photopolymer two-component elastomers, resulting in insufficient interlayer interfacial bonding strength and a high risk of warping and cracking in the molded parts. Summary of the Invention

[0005] In view of this, the present invention provides a method for preparing photocurable two-component 3D printing elastic materials, which can solve the technical problem that in the process of photocurable two-component 3D printing elastic materials, the curing shrinkage behavior is difficult to predict and compensate in real time, resulting in insufficient interlayer bonding strength, warping and cracking risk of the molded parts.

[0006] This invention is achieved as follows: This invention provides a method for preparing a photocurable two-component 3D printing elastic material, comprising the following steps:

[0007] Component A and Component B of the two-component elastic resin system are placed in two gear pump-driven feeding units respectively. The instantaneous mass flow rates of the two components are collected in real time by Coriolis mass flow meters. The dual-pump flow coupling dynamic model is driven by a model predictive control algorithm to control the mixing mass ratio error of component A and component B within ±0.5%.

[0008] The printing cavity is filled with nitrogen until the oxygen concentration inside the cavity is below 500 ppm. At the same time, a ternary thiol-olefin synergistic curing system is introduced into the mixed formulation of the two-component elastic resin system, and a hindered amine light stabilizer is added to eliminate the effect of oxygen inhibition on the surface curing integrity.

[0009] Using the current layer thickness, current layer exposure, and mixing formula parameters as input, the pre-trained curing shrinkage field prediction model is called to output the layer-by-layer volume shrinkage rate field. The Fick diffusion-reaction coupled curing depth iterative solution algorithm is used to calculate the curing profile of each layer, and the printing path is compensated online based on the layer-by-layer volume shrinkage rate field.

[0010] After each layer is cured, a low dose of supplemental exposure is applied to the current cured layer to retain the target proportion of active double bonds at the interlayer interface. The current layer thickness is set to less than 20 μm so that the new layer resin copolymerizes with the active double bonds to form an active interface layer.

[0011] Extract the layer-by-layer equivalent shrinkage stress value output from the curing shrinkage field prediction model, calculate the interface quality assessment function value, and dynamically adjust the low-dose supplementary exposure and the current layer thickness according to the interval to which the interface quality assessment function value belongs, so that the interlayer bonding strength is maintained within the target range.

[0012] After printing, the molded parts undergo overall post-curing treatment to eliminate residual active double bonds and obtain photocured two-component elastic material molded parts with uniform mechanical properties.

[0013] The Coriolis mass flow meter works as follows: when fluid flows through the vibrating tube, the Coriolis force causes the vibrating tube to generate a phase difference proportional to the instantaneous mass flow rate. The controller calculates the phase difference in real time to obtain the instantaneous mass flow rate. The sampling period reaches the sub-hundred-millisecond level, thereby providing a high-frequency instantaneous mass flow rate feedback signal for the model predictive control algorithm.

[0014] Among them, the model predictive control algorithm uses the dual-pump flow coupling dynamics model as the prediction model. It solves the optimal control sequence in the finite time domain in each control cycle and incorporates the ratio error of the instantaneous mass flow rates of component A and component B into the objective function. The dual-pump flow coupling dynamics model obtains the model parameters by performing a system identification experiment on the measured instantaneous mass flow rate data under different back pressure, speed and viscosity conditions of the two-component elastic resin system, and fitting the data using the least squares method.

[0015] In the two-component elastic resin system, component A consists of 40%–60% polyether-type polyurethane acrylate oligomer, 10%–20% aliphatic polyurethane acrylate oligomer, 5%–15% isobornyl acrylate, 5%–10% polyethylene glycol diacrylate, 1%–3% 2-hydroxy-2-methyl-1-phenyl-1-propanone, and 0.5%–1.5% hindered amine light stabilizer; component B consists of 20%–40% pentaerythritol tetra(3-mercaptopropionate), 10%–20% triallyl isocyanurate, 30%–50% polyethylene glycol diacrylate, and 1%–3% 2-hydroxy-2-methyl-1-phenyl-1-propanone; the mixing mass ratio of component A to component B is 1:1.

[0016] The working principle of the ternary thiol-olefin synergistic curing system is as follows: after the photoinitiator generates free radicals, pentaerythritol tetra(3-mercaptopropionate) transfers hydrogen atoms to triallyl isocyanurate through a sulfur-hydrogen transfer reaction to generate carbon free radicals. The carbon free radicals then copolymerize with the acrylate double bonds in the polyether-type polyurethane acrylate oligomer, bypassing the competitive reaction between dissolved oxygen and chain-growing free radicals. The hindered amine light stabilizer sebacic acid bis(2,2,6,6-tetramethyl-4-piperidinyl) ester further consumes residual dissolved oxygen by capturing oxygen free radicals.

[0017] The solidification shrinkage field prediction model adopts a physical information neural network architecture. The output layer outputs the layer-by-layer volume shrinkage rate field and the layer-by-layer equivalent shrinkage stress value. The intermediate layer of the network adopts a sinusoidal activation function architecture. The loss function includes both data fitting terms and mechanical equilibrium equation residual terms. The boundary condition terms are internally configured with an adaptive weight balancing mechanism to dynamically adjust the weights of each physical residual term.

[0018] The steps for establishing the training dataset for the solidification shrinkage field prediction model are as follows: For the current layer thickness range of 10–50 μm and the current layer exposure range of 10–200 μm… A three-dimensional curing shrinkage simulation model was established using finite element simulation software to cover the entire process parameter space, including the mixed formulation parameters. Through offline finite element simulation calculations, layer-by-layer volume shrinkage rate field and layer-by-layer equivalent shrinkage stress value samples covering the entire process parameter space were obtained to form a training dataset.

[0019] The working principle of the Fick diffusion-reaction coupled curing depth iterative solution algorithm is as follows: A Beer-Lambert attenuation model for ultraviolet light propagating along the depth direction is established, along with a set of free radical diffusion-reaction kinetic equations containing oxygen-containing polymerization inhibition competition terms. These two sets of equations are coupled to form a set of partial differential equations. The spatial grid is discretized using the finite difference method in the depth direction, and the adaptive Runge-Kutta method is used for integration in the time direction. The convergence criterion is the conversion rate field between two adjacent steps. norm difference less than .

[0020] The initial value of the low-dose supplemental exposure of the active interface layer is determined as follows: a group of elastic samples with different low-dose supplemental exposures are prepared, and the interlayer peel strength and surface acrylate double bond retention rate of each group of samples are tested. The lowest low-dose supplemental exposure corresponding to the interlayer peel strength meeting the design lower limit and the acrylate double bond retention rate not less than 15% is taken as the initial value.

[0021] Among them, the interface quality evaluation function The statement is as follows: ;in This serves as a standard reference value for the interface quality evaluation function. This is the standard reference value for the equivalent shrinkage stress layer by layer. To supplement the standard reference value for low-dose exposure, The current standard reference value for layer thickness is determined by preparing a set of elastic specimens with different parameter combinations and regressing the parameter combinations corresponding to the interlayer peel strength meeting the design lower limit.

[0022] Among them, based on the interface quality evaluation function value The rule for dynamically adjusting the low-dose supplemental exposure and the current layer thickness within the specified interval is as follows: when At that time, the low-dose supplemental exposure will be increased by 10%; when At the same time, maintain a low dose of supplemental exposure and keep the current layer thickness unchanged; when When, reduce the current layer thickness by 20%; when At this time, the current layer thickness is reduced by 20% and the low-dose supplemental exposure is increased by 20%, and the curing shrinkage field prediction model is triggered to recalculate the online printing path compensation amount of the current layer.

[0023] The threshold of oxygen concentration below 500 ppm in the cavity was determined as follows: elastic film samples were prepared with the same mixing formula parameters and current layer exposure under different oxygen concentration gradients in the cavity. The thickness of the adhesive layer on the surface of each group of samples and the surface elongation at break were tested. The highest oxygen concentration in the cavity corresponding to the surface elongation at break not being lower than 95% of the bulk elongation at break was taken as the control upper limit.

[0024] The current layer thickness is set to be below 20 μm based on the following method: elastic specimens with current layer thicknesses of 10 μm, 20 μm, 30 μm and 50 μm are prepared, and the interlaminar fatigue delamination life of each group of specimens under large strain cyclic loading is tested. The maximum current layer thickness corresponding to the interlaminar fatigue delamination life is not less than 80% of the predicted value of homogeneous material is used as the upper limit of the setting.

[0025] The exposure time and light intensity of the overall post-curing treatment were determined as follows: Fourier transform infrared spectroscopy was performed on the molded sample with different overall post-curing exposure times, and the shortest exposure time corresponding to when the characteristic peak area of ​​the acrylate double bond decreased to less than 5% of the initial value was taken as the standard value of the overall post-curing exposure time.

[0026] The overall post-curing process involves placing the printed part in an ultraviolet light box and exposing it uniformly with ultraviolet light. This allows residual active double bonds that were not fully converted during the printing process due to the limitations of the layer-by-layer process to further participate in the cross-linking reaction, thereby improving the uniformity of the overall cross-linking density of the part.

[0027] Among them, the curing shrinkage field prediction model embeds the basic equations of elasticity into the loss function, so that the network prediction results can always conform to the laws of mechanical equilibrium while meeting the data fitting accuracy. Thus, it can still maintain physically consistent prediction output in the process parameter range not covered by the training data, and enable online printing path compensation to have the ability to generalize to new mixed formula parameters and new current layer thickness.

[0028] This invention constructs a curing shrinkage field prediction model with a physical information neural network at its core. Taking current layer thickness, exposure amount, and mixing formula parameters as input, it outputs the layer-by-layer volumetric shrinkage rate field and equivalent shrinkage stress value within millisecond-level inference time. The prediction results directly drive online compensation of the printing path, fundamentally eliminating the limitation of traditional offline simulation methods that cannot respond in real time to changes in layer-by-layer process parameters. This invention also introduces a precise curing profile prediction mechanism based on the Fick diffusion-reaction coupled curing depth iterative solution algorithm. This mechanism solves the light propagation attenuation law and free radical reaction kinetics constraints simultaneously, allowing the curing profile information of each layer to directly participate in the online correction of the shrinkage field prediction, forming a physical closed loop for curing behavior prediction. Furthermore, this invention dynamically adjusts the low-dose supplementary exposure amount and layer thickness through an interface quality assessment function, maintaining the retention rate of active double bonds at the interlayer interface within the target range. This ensures that the new layer resin and residual active double bonds undergo cross-layer copolymerization, transforming the interlayer interface into a region of continuous chemical bond transition. In summary, the present invention solves the technical problems mentioned in the background art, such as the difficulty in predicting and compensating for the curing shrinkage behavior during the 3D printing of photocurable two-component elastic materials, which leads to insufficient interlayer bonding strength and a high risk of warping and cracking of the molded parts. Attached Figure Description

[0029] Figure 1 This is a flowchart of the method of the present invention.

[0030] Figure 2 The training loss convergence curve of the prediction model for the curing shrinkage field of a two-component elastic resin system.

[0031] Figure 3 This is a comparison chart of the predicted values ​​of the volumetric shrinkage rate field layer by layer in the entire process parameter space and the reference values ​​of the solidification shrinkage field prediction model. Detailed Implementation

[0032] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below.

[0033] like Figure 1 The diagram shown is a flowchart of a method for preparing a photocurable two-component 3D printing elastic material according to the present invention. This method includes the following steps:

[0034] S01. Component A and Component B of the two-component elastic resin system are placed in two gear pump-driven feeding units respectively. The instantaneous mass flow rates of the two components are collected in real time by a Coriolis mass flow meter. The dual-pump flow coupling dynamic model is driven by a model predictive control algorithm to control the mixing mass ratio error of component A and component B within ±0.5%.

[0035] S02. Fill the printing cavity with nitrogen until the oxygen concentration inside the cavity is below 500ppm. At the same time, introduce a ternary thiol-olefin synergistic curing system into the mixed formulation of the two-component elastic resin system and add a hindered amine light stabilizer to eliminate the influence of oxygen inhibition on the surface curing integrity.

[0036] S03. Using the current layer thickness, current layer exposure, and mixing formula parameters as input, the pre-trained curing shrinkage field prediction model is called to output the layer-by-layer volume shrinkage rate field. The Fick diffusion-reaction coupled curing depth iterative solution algorithm is used to calculate the curing profile of each layer. The printing path is compensated online based on the layer-by-layer volume shrinkage rate field.

[0037] S04. After each layer is cured, a low dose of supplemental exposure is applied to the current cured layer to retain the target proportion of active double bonds at the interlayer interface. The current layer thickness is set to less than 20 μm so that the new layer resin copolymerizes with the active double bonds to form an active interface layer.

[0038] S05. Extract the layer-by-layer equivalent shrinkage stress value output by the curing shrinkage field prediction model, calculate the interface quality assessment function value, and dynamically adjust the low-dose supplementary exposure amount and the current layer thickness according to the interval to which the interface quality assessment function value belongs, so that the interlayer bonding strength is maintained within the target range.

[0039] S06. After printing, the molded part is subjected to overall post-curing treatment to eliminate residual active double bonds and obtain a photocurable two-component elastic material molded part with uniform mechanical properties.

[0040] The specific formulation of the two-component elastic resin system is as follows: Component A consists of 40%–60% polyether-type polyurethane acrylate oligomer, 10%–20% aliphatic polyurethane acrylate oligomer, 5%–15% reactive diluent monomer isobornyl acrylate, 5%–10% reactive diluent monomer polyethylene glycol diacrylate, 1%–3% photoinitiator 2-hydroxy-2-methyl-1-phenyl-1-propanone, and 0.5%–1.5% hindered amine light stabilizer, with the sum of the mass ratios of all components being 100%; Component B consists of 20%–40% polyfunctional thiol monomer pentaerythritol tetra(3-mercaptopropionate) and olefin monomer triallyl isocyanurate. The system comprises 10%–20% ester, 30%–50% reactive diluent monomer polyethylene glycol diacrylate, and 1%–3% photoinitiator 2-hydroxy-2-methyl-1-phenyl-1-propanone, with the sum of the mass ratios of all components being 100%. The mixing mass ratio of component A to component B is 1:1. The ternary thiol-olefin synergistic curing system is composed of the polyfunctional thiol monomer pentaerythritol tetra(3-mercaptopropionate), the olefin monomer triallyl isocyanurate in component B, and the acrylate double bonds in the polyether-type polyurethane acrylate oligomer in component A. The hindered amine light stabilizer is sebacic acid bis(2,2,6,6-tetramethyl-4-piperidinyl) ester.

[0041] The Coriolis mass flow meter works as follows: when fluid flows through the vibrating tube, the Coriolis force causes the vibrating tube to generate a phase difference proportional to the instantaneous mass flow rate. The controller calculates the phase difference in real time to obtain the instantaneous mass flow rate, and the sampling period reaches the sub-hundred-millisecond level, thereby providing a high-frequency instantaneous mass flow rate feedback signal for the model predictive control algorithm.

[0042] The working principle of the model predictive control algorithm is as follows: using the dual-pump flow coupling dynamics model as the prediction model, the optimal control sequence in the finite time domain is solved in each control cycle. The ratio error of the instantaneous mass flow rates of component A and component B is incorporated into the objective function. By predicting the flow deviation in the next few steps and applying compensation commands in advance, the algorithm overcomes the hysteresis of traditional proportional-integral-derivative control in response to nonlinear viscosity coupling. The model predictive control algorithm incorporates oxygen inhibition factors into the flow coupling prediction model, so that the dual-pump coordinated response takes into account both viscosity nonlinearity and back pressure fluctuations, thereby maintaining a high-precision and stable mixing mass ratio throughout the entire printing cycle. The dual-pump flow coupling dynamics model is obtained by systematically identifying the measured instantaneous mass flow rate data under different back pressures, rotation speeds, and viscosity conditions of the two-component elastic resin system through experiments. The model parameters are obtained by least squares fitting. The experimental temperature covers the actual printing temperature range, and the experimental back pressure covers the actual printing back pressure range.

[0043] The working principle of the ternary thiol-olefin synergistic curing system is as follows: After the photoinitiator generates free radicals, the multifunctional thiol monomer pentaerythritol tetra(3-mercaptopropionate) transfers hydrogen atoms to the olefin monomer triallyl isocyanurate through a sulfur-hydrogen transfer reaction, generating carbon free radicals. These carbon free radicals then copolymerize with the acrylate double bonds in the polyether-type polyurethane acrylate oligomer, thereby bypassing the competitive reaction between dissolved oxygen and chain-growing free radicals, allowing the surface curing layer to achieve full conversion under low-activity free radical concentration conditions. The hindered amine light stabilizer sebacic acid bis(2,2,6,6-tetramethyl-4-piperidinyl) ester further consumes residual dissolved oxygen by capturing oxygen free radicals, working synergistically with the ternary thiol-olefin synergistic curing system.

[0044] The specific structure of the curing shrinkage field prediction model is as follows: It employs a physical information neural network architecture. The input layer receives the current layer thickness, current layer exposure, and mixing formula parameters. The output layer outputs the layer-by-layer volumetric shrinkage rate field and the layer-by-layer equivalent shrinkage stress value. The intermediate layer of the network uses a sinusoidal activation function architecture to fit the high-frequency spatially varying stress field. The loss function simultaneously includes data fitting terms and mechanical equilibrium equation residual terms. The boundary condition term incorporates an internal adaptive weight balancing mechanism to dynamically adjust the weights of each physical residual term, preventing a single constraint term from dominating the gradient direction and causing other constraints to fail. The specific steps for establishing the training dataset for the solidified shrinkage field prediction model include: targeting a current layer thickness range of 10–50 μm and a current layer exposure range of 10–200 μm. A three-dimensional curing shrinkage simulation model was established using finite element simulation software, encompassing the entire process parameter space of the mixed formulation parameters. Through offline finite element simulation calculations, layer-by-layer volumetric shrinkage rate fields and layer-by-layer equivalent shrinkage stress value samples covering the entire process parameter space were obtained, forming a training dataset. The specific steps for training the curing shrinkage field prediction model include: using finite element simulation samples as supervisory data, and using the residual terms of the mechanical equilibrium equations... Using boundary condition residuals as physical constraints, the loss function is jointly optimized. An adaptive weight balancing mechanism is adopted to dynamically adjust the weight ratio of the data fitting term and the physical residual term during training. After training convergence, the calculation time of the single-layer layer-by-layer volume shrinkage rate field during the inference stage is compressed to the millisecond level, realizing online printing path compensation.

[0045] The curing shrinkage field prediction model embeds the basic equations of elasticity into the loss function, ensuring that the network prediction results meet the data fitting accuracy requirements while always conforming to the laws of mechanical equilibrium. This allows the model to maintain physically consistent prediction outputs even in process parameter ranges not covered by the training data, avoiding non-physical interpretations in the extrapolation range of purely data-driven models. This enables online printing path compensation to generalize to new mixed formula parameters and new current layer thicknesses, ultimately effectively suppressing the risk of warpage and cracking of molded parts across the entire process parameter space.

[0046] The working principle of the Fick diffusion-reaction coupled curing depth iterative solution algorithm is as follows: A Beer-Lambert attenuation model is established for ultraviolet light propagating along the depth direction in a two-component elastic resin system, describing the exponential decay of light intensity with depth. Simultaneously, a set of free radical diffusion-reaction kinetic equations is established, containing an oxygen inhibition competition term, describing the consumption process of chain-growing free radicals by dissolved oxygen. These two sets of equations are coupled to form a set of partial differential equations. The spatial grid is discretized using the finite difference method in the depth direction, and the adaptive Runge-Kutta method is used for integration in the time direction. Within each time step, the absorbance of the two-component elastic resin system is first updated based on the current monomer conversion rate, then the light intensity distribution is updated, and then the conversion rate increment is calculated, achieving a bidirectional coupled iterative solution of light propagation and chemical reaction. The convergence criterion is the conversion rate field between two adjacent steps. norm difference less than This allows for accurate prediction of the cured profile under any current layer exposure parameter.

[0047] The Fick diffusion-reaction coupled curing depth iterative solution algorithm couples the light propagation physical model with the free radical reaction kinetic equation, so that the curing profile prediction result satisfies both the light attenuation law and the chemical reaction kinetic constraint. Therefore, even when the dissolved oxygen concentration, monomer type or the current layer exposure changes, it can give prediction results consistent with the actual curing behavior, providing a high-confidence theoretical basis for online printing path compensation and current layer exposure optimization. It also forms a closed loop with the input of the curing shrinkage field prediction model, so that the curing profile information of each layer directly participates in the online correction of the layer-by-layer volume shrinkage rate field.

[0048] The active interface layer refers to the layer on which, after each layer has cured, a low-dose supplementary exposure is applied to the current cured layer, allowing a certain proportion of unreacted acrylate double bonds to remain on the surface. When the next layer of fresh resin is applied, these unreacted acrylate double bonds copolymerize with the monomers in the new resin layer during the exposure process, thereby forming a covalent cross-linked network across adjacent layers at the interlayer interface. This transforms the interlayer interface, which originally had a low conversion rate of acrylate double bonds, into a region of continuous chemical bond transition. The initial value of the low-dose supplementary exposure is determined as follows: a group of elastic samples with different low-dose supplementary exposures are prepared, and the interlayer peel strength and surface acrylate double bond retention rate of each group of samples are tested. The lowest low-dose supplementary exposure corresponding to the interlayer peel strength meeting the design lower limit and the acrylate double bond retention rate not less than 15% is used as the initial value, and this value is determined after analyzing multiple experimental data.

[0049] The interface quality assessment function is constructed by using the layer-by-layer equivalent shrinkage stress values ​​output by the curing shrinkage field prediction model. Current low-dose supplemental exposure With the current layer thickness Interface quality evaluation function with variables as independent variables The statement is as follows:

[0050] ;

[0051] in This serves as a standard reference value for the interface quality evaluation function. This is the standard reference value for the equivalent shrinkage stress layer by layer. To supplement the standard reference value for low-dose exposure, This is the current standard reference value for layer thickness; , , The following method was used to determine the interlayer peel strength of each set of elastic samples with different combinations of current layer thickness and low-dose supplemental exposure. The parameter combination corresponding to the interlayer peel strength meeting the design lower limit was used as the standard reference value, and this was determined after regression analysis of multiple experimental data. The method was based on the interface quality assessment function value. The rules for dynamically adjusting the low-dose supplemental exposure and the current layer thickness within the specified range are as follows: When At that time, the low-dose supplemental exposure is increased by 10%, while the current layer thickness remains unchanged; when At the same time, maintain a low dose of supplemental exposure and keep the current layer thickness unchanged; when When, reduce the current layer thickness by 20%; when At this time, the current layer thickness is reduced by 20% and the low-dose supplemental exposure is increased by 20%, and the curing shrinkage field prediction model is triggered to recalculate the online printing path compensation amount of the current layer.

[0052] The threshold of oxygen concentration below 500 ppm in the cavity is determined by the following method: elastic film samples are prepared under different oxygen concentration gradients in the cavity with the same mixing formula parameters and current layer exposure. The thickness of the adhesive layer on the surface of each group of samples and the surface elongation at break are tested. The highest oxygen concentration in the cavity corresponding to the surface elongation at break not being lower than 95% of the bulk elongation at break is used as the control upper limit. This is determined after analyzing multiple sets of experimental data.

[0053] The setting of the current layer thickness to below 20 μm is based on the following method: elastic specimens with current layer thicknesses of 10 μm, 20 μm, 30 μm, and 50 μm are prepared, and the interlaminar fatigue delamination life of each group of specimens under large strain cyclic loading is tested. The maximum current layer thickness corresponding to the interlaminar fatigue delamination life not being less than 80% of the predicted value of homogeneous materials is used as the upper limit of the setting, and the result is determined after regression analysis of experimental data.

[0054] The overall post-curing treatment refers to placing the printed part in an ultraviolet light box and uniformly exposing the entire part to ultraviolet light. This allows residual active double bonds that were not fully converted during the printing process due to the limitations of the layer-by-layer process to further participate in the cross-linking reaction, thereby improving the uniformity of the overall cross-linking density of the part. The exposure time and light intensity of the overall post-curing treatment are determined as follows: Fourier transform infrared spectroscopy tests are performed on the part samples with different overall post-curing exposure times. The shortest exposure time corresponding to the decrease of the characteristic peak area of ​​the acrylate double bonds to less than 5% of the initial value is taken as the standard value of the overall post-curing exposure time. This value is determined after multiple experimental verifications.

[0055] The specific implementation of step S01 is as follows: Component A and Component B are respectively filled into their respective sealed storage tanks and connected to a gear pump-driven feeding unit via pipelines. The speed of the gear pump is precisely controlled by a servo motor. A Coriolis mass flow meter is installed on each of the discharge pipelines of component A and component B. Utilizing the physical principle that the Coriolis force causes a phase difference proportional to the instantaneous mass flow rate when fluid flows through the vibrating tube, the controller calculates the phase difference in real time with a sampling period on the order of sub-milliseconds to obtain a high-frequency instantaneous mass flow rate feedback signal. This feedback signal is fed into a model predictive control algorithm. The algorithm uses a dual-pump flow coupling dynamics model as the prediction model and, within each control cycle, continuously solves for the optimal pump speed sequence based on finite-time optimal control theory. The ratio error of the instantaneous mass flow rates of components A and B is incorporated into the objective function. By predicting the flow deviation in the next few steps and applying compensation commands in advance, the algorithm overcomes the hysteresis of traditional proportional-integral-derivative control in response to nonlinear viscosity coupling. The parameters of the dual-pump flow coupling dynamic model were obtained in advance through system identification experiments. The experiments covered the actual printing temperature range and back pressure range. After collecting measured instantaneous mass flow data under various operating conditions, the model parameters were fitted using the least squares method to ensure that the model can accurately describe the dual disturbances of viscosity nonlinearity and back pressure fluctuation, thereby stabilizing the mixing mass ratio error within ±0.5% throughout the entire printing cycle.

[0056] The specific implementation of step S02 is as follows: Before printing begins, high-purity nitrogen is continuously introduced into the sealed printing cavity. The oxygen concentration inside the cavity is monitored in real time by an oxygen sensor. The printing program can only be started when the oxygen concentration drops below 500 ppm. The upper limit of 500 ppm for controlling the oxygen concentration inside the cavity is determined as follows: elastic film samples are prepared under different oxygen concentration gradient environments with the same mixing formula parameters. The surface elongation at break of each group of samples is tested. The highest oxygen concentration corresponding to the surface elongation at break not being lower than 95% of the bulk elongation at break is taken as the upper limit of control. After multiple sets of experiments, 500 ppm was determined. At the mixing formula level, the multifunctional thiol monomer pentaerythritol tetra(3-mercaptopropionate) and the olefin monomer triallyl isocyanurate are introduced into component B. Together with the acrylate double bonds of the polyether-type polyurethane acrylate oligomer in component A, they form a ternary thiol-olefin synergistic curing system. After the photoinitiator generates free radicals, pentaerythritol tetra(3-mercaptopropionate) generates carbon free radicals through a sulfhydryl-hydrogen transfer reaction. These carbon free radicals then copolymerize with the double bonds of the acrylate, thus bypassing the competitive reaction between dissolved oxygen and chain-growing free radicals. Simultaneously, bis(2,2,6,6-tetramethyl-4-piperidinyl) sebacic acid is added to component A as a hindered amine light stabilizer. This stabilizer further consumes residual dissolved oxygen by capturing oxygen free radicals, synergistically eliminating the impact of oxygen inhibition on the integrity of surface curing in conjunction with the ternary thiol-olefin co-curing system.

[0057] The specific implementation of step S03 is as follows: Before each layer exposure, using the current layer thickness, current layer exposure amount, and mixing formula parameters as input, a pre-trained curing shrinkage field prediction model is invoked for inference. This model employs a physical information neural network architecture, with the intermediate layers using a sinusoidal activation function to fit the high-frequency spatially varying stress field. The output layer provides the layer-by-layer volumetric shrinkage rate field and the layer-by-layer equivalent shrinkage stress value, compressing the inference time to the millisecond level. The training dataset for the curing shrinkage field prediction model is used in finite element simulation software for layer thicknesses of 10–50 μm and exposure amounts of 10–200 μm. The entire process parameter space, including the mixed formulation parameters, is obtained through offline simulation. During training, finite element simulation samples are used as supervised data, and the residual terms of the mechanical equilibrium equations are also included. The boundary condition residuals are embedded as physical constraints in the loss function. An adaptive weight balancing mechanism dynamically adjusts the weight ratio between the data fitting term and the physical residual term to ensure that the network can still provide physically consistent predictions in the extrapolation interval. Simultaneously, the Fick diffusion-reaction coupled curing depth iterative solution algorithm is used to calculate the curing profile of each layer. This algorithm couples the Beer-Lambert light attenuation model with the free radical diffusion-reaction kinetic equations of the oxygen-containing polymerization inhibition competition term to form a system of partial differential equations. It uses the finite difference method for discretization in the depth direction and the adaptive Runge-Kutta method for integration in the time direction. At each step, the absorbance is updated based on the current monomer conversion rate, and then the light intensity distribution is updated. norm difference less than As a convergence criterion, the obtained solidified contour information directly participates in the online correction of the layer-by-layer volume shrinkage rate field, and the corrected shrinkage rate field drives the real-time calculation of the online compensation amount of the printing path.

[0058] The specific implementation of step S04 is as follows: After each layer has been normally exposed and cured, an additional low-dose supplementary exposure is applied to the current cured layer to make the conversion rate of acrylate double bonds on the surface of the current cured layer lower than the level of complete curing, thereby retaining a certain proportion of unreacted active double bonds. The initial value of the low-dose supplementary exposure is determined in advance by the following method: a group of elastic samples with different low-dose supplementary exposures are prepared, and the interlayer peel strength and surface acrylate double bond retention rate of each group of samples are tested. The lowest low-dose supplementary exposure corresponding to the interlayer peel strength meeting the design lower limit and the acrylate double bond retention rate not less than 15% is taken as the initial value, and determined after multiple experimental data analysis. The current layer thickness is set to be below 20 μm. The basis for determining this upper limit is: interlayer fatigue delamination life test under large strain cyclic loading is performed on elastic samples with layer thicknesses of 10 μm, 20 μm, 30 μm, and 50 μm. The maximum layer thickness corresponding to the interlayer fatigue delamination life not less than 80% of the predicted value of homogeneous materials is taken as the set upper limit, and determined to be 20 μm by experimental data regression analysis. After the new layer of resin is applied, during the next layer exposure process, the monomers in the new layer of resin undergo a copolymerization reaction with the active double bonds retained on the surface of the current cured layer, forming a covalent cross-linked network that spans two adjacent layers at the interlayer interface, transforming the interlayer interface into a region of continuous chemical bond transition.

[0059] The specific implementation of step S05 is as follows: Extract the layer-by-layer equivalent shrinkage stress value of the current layer from the real-time output of the curing shrinkage field prediction model in step S03. Combined with the current low-dose supplemental exposure With the current layer thickness Based on the interface quality evaluation function Calculate the current layer interface quality evaluation function value The standard reference value is... , , A set of elastic specimens with varying combinations of layer thickness and low-dose supplemental exposure were prepared, and the interlayer peel strength was tested. The parameter combination that met the design lower limit was determined through regression analysis of multiple experimental data. For the specified interval, the following dynamic adjustment rules apply: When At that time, the low-dose supplemental exposure was increased by 10%, while the layer thickness remained unchanged; when When, both remain unchanged; when When, reduce the current layer thickness by 20%; when At this time, the current layer thickness is reduced by 20%, the low-dose supplemental exposure is increased by 20%, and the curing shrinkage field prediction model is triggered to recalculate the online printing path compensation for the current layer. This closed-loop adjustment mechanism quantifies the combined effect of equivalent shrinkage stress, exposure, and layer thickness into a single scalar. This ensures that the interlayer bond strength remains within the target range.

[0060] The specific implementation of step S06 is as follows: After the printed part is removed from the printing platform, it is placed in an ultraviolet light box equipped with a multi-directional uniform ultraviolet light source for overall post-curing treatment. This allows residual active double bonds that were not fully converted during the printing process due to layer-by-layer limitations to further participate in the cross-linking reaction, improving the uniformity of the overall cross-linking density of the part. The exposure time and light intensity of the overall post-curing treatment are determined as follows: Fourier transform infrared spectroscopy tests are performed on part samples with different overall post-curing exposure times. The shortest exposure time corresponding to the decrease of the characteristic peak area of ​​the acrylate double bonds to less than 5% of the initial value is taken as the standard value of the overall post-curing exposure time. After multiple experimental verifications, the reference value is approximately 15-30 min, and the reference value of the light intensity is approximately 20-50. The specific values ​​are adjusted according to the mixing formula parameters. The overall post-curing treatment makes the cross-linking density of each layer of the molded part more uniform, eliminating the uneven local double bond conversion rate caused by differences in layer thickness and exposure in the layer-by-layer printing process, and finally obtaining a photocurable two-component elastic material molded part with uniform mechanical properties.

[0061] It should be noted that the first key technical idea of ​​this invention is to construct a curing shrinkage field prediction model using a physical information neural network. While traditional finite element methods can accurately calculate the shrinkage field, single-layer calculations take minutes, failing to meet the real-time requirements of online printing. The physical information neural network embeds constraints from the elasticity equilibrium equations into the loss function, ensuring that the network predictions conform to mechanical equilibrium laws while maintaining data fitting accuracy. The inference phase computation time is compressed to the millisecond level, and it still provides physically consistent outputs even in process parameter ranges not covered by the training data, fundamentally resolving the contradiction between real-time performance and generalization. The second key technical idea is to accurately predict the curing profile using a Fick diffusion-reaction coupled curing depth iterative solution algorithm. This algorithm bidirectionally couples light propagation attenuation with free radical reaction kinetics, ensuring that the curing profile prediction simultaneously satisfies both photophysical and chemical kinetic constraints. The resulting curing profile information directly participates in the online correction of the shrinkage field prediction, forming a closed loop. The third key technical approach is to use an interface quality assessment function to drive the dynamic synergistic regulation of low-dose supplemental exposure and layer thickness. This quantifies the combined effect of equivalent shrinkage stress, exposure, and layer thickness into a single scalar, ensuring that the retention rate of interlayer active double bonds and the copolymer crosslinking density remain within the target range. When these three technical approaches work synergistically, curing profile prediction provides a high-confidence input for shrinkage field prediction. The equivalent shrinkage stress output from the shrinkage field prediction drives the interface quality assessment function, and the regulation results of the interface quality assessment function are fed back to the exposure and layer thickness settings of the next layer, thereby affecting the curing profile and shrinkage field prediction of the next layer. This forms a layer-by-layer rolling closed-loop control, ensuring that shrinkage compensation and interface strengthening are simultaneously guaranteed at both the physical and chemical levels. Compared to the effect of any single mechanism acting alone, this approach has an inseparable synergistic gain.

[0062] It should be noted that this invention also solves the following technical problem: In photopolymerization printing of a two-component elastic resin system, the viscosity of component A and component B exhibits significant nonlinear coupling characteristics with temperature and back pressure. Traditional proportional-integral-derivative (PI-DE) control, unable to predict flow deviations in the next few steps, exhibits a delayed response to back pressure fluctuations, leading to a step deviation in the instantaneous ratio of the two flow paths. This, in turn, causes local inaccuracies in the mixing formulation parameters, affecting the consistency of curing shrinkage behavior. This invention drives a dual-pump flow coupling dynamic model through a model predictive control algorithm, uses finite-time optimal control theory to solve for the optimal pump speed sequence, and applies compensation commands in advance. This ensures that the mixing mass ratio error remains stably controlled within ±0.5% even under the dual disturbances of nonlinear viscosity coupling and back pressure fluctuations, thus eliminating the interference of instantaneous mixing ratio inaccuracies on the accuracy of subsequent curing shrinkage prediction and path compensation from a mechanistic perspective. Furthermore, this invention also solves the problem of incomplete surface curing caused by oxygen inhibition. The competitive reaction between dissolved oxygen and chain-growing free radicals makes it difficult for surface-cured layers to achieve full conversion in open environments. Traditional nitrogen protection solutions, limited by sealing constraints, often cannot control oxygen concentrations to sufficiently low levels, while simply reducing oxygen concentration increases equipment complexity. This invention, through the synergistic effect of a ternary thiol-olefin synergistic curing system and a hindered amine light stabilizer, bypasses the competitive pathway of dissolved oxygen at the chemical mechanism level and actively consumes residual dissolved oxygen. This allows the surface-cured layer to achieve full conversion even under low-reactive-radical concentration conditions, providing dual protection with the nitrogen protection solution and eliminating the stringent dependence on extremely low oxygen concentrations.

[0063] Specifically, the principle of this invention is as follows: This invention can solve the above-mentioned technical problems, and its logical root lies in incorporating the real-time prediction of the curing shrinkage field and the active control of the interlayer interface into the same closed-loop control framework. The physical information neural network, by embedding elastic mechanical equilibrium equation constraints in the loss function, ensures that the network always satisfies the laws of mechanical equilibrium while fitting finite element simulation supervision data. This allows it to output a physically consistent shrinkage rate field even in process parameter ranges not covered by training data, avoiding non-physical solutions generated by purely data-driven models in the extrapolation range. The millisecond-level computation speed in the inference stage enables the prediction result to complete the path compensation calculation before each layer exposure, eliminating the fundamental contradiction between offline simulation methods and online printing cycle time. The Fick diffusion-reaction coupled curing depth iterative solution algorithm bidirectionally couples the Beer-Lambert model of light intensity attenuation with the free radical reaction kinetic equation of the oxygen-containing polymerization inhibition competition term, ensuring that the curing profile prediction simultaneously satisfies both the physical constraints of light propagation and the kinetic constraints of chemical reaction, providing a high-confidence layer-by-layer curing profile input for shrinkage field prediction. The interface quality assessment function quantifies the combined effect of equivalent shrinkage stress, low-dose supplemental exposure, and layer thickness into a single scalar. Based on its corresponding range, process parameters are dynamically adjusted to ensure that the retention rate of active double bonds and the copolymer crosslinking density remain within the target range, guaranteeing the continuous stability of interlayer bond strength. The synergistic effect of these three mechanisms ensures that curing shrinkage compensation and interlayer interface strengthening are simultaneously guaranteed at both the physical and chemical levels, thereby eliminating the root causes of warpage, cracking, and interlayer delamination from a mechanistic perspective.

[0064] The following provides a specific embodiment 1 of the present invention, and the specific implementation of each step in this embodiment 1 is described in detail below.

[0065] The specific implementation method of step S01 is as follows: a two-component elastic resin system Components and The components are placed in two gear pump-driven feeding units. The Coriolis mass flow meter uses the phase difference generated when the fluid flows through the vibrating tube, which is proportional to the instantaneous mass flow rate, to calculate the flow rate in real time. Instantaneous mass flow rate of components and Instantaneous mass flow rate of components The sampling period reaches the sub-hundred-millisecond level. The dual-pump flow coupling dynamics model was established through system identification experiments, and the measured data under different back pressures, rotational speeds, and viscosity conditions were fitted using the least squares method. The model is as follows:

[0066] ;

[0067] In the formula, Component identification for Components in Instantaneous mass flow rate at time t, in units for Component instantaneous mass flow rate standard reference value, unit for Component gear pumps in Rotational speed at any given time, in units for Standard reference value for component rotation speed, unit for Components in Back pressure at any moment, unit for Component back pressure standard reference value, unit for Components at temperature Viscosity at the specified level, in units Current printing temperature, in °C for Standard reference value for component viscosity, unit , , The dimensionless model parameters obtained by least squares fitting This is the index of the term in the rotational speed polynomial, with a value of [value]. This is the index of the term in the backpressure polynomial, with a value of [value]. , The order is a polynomial, with empirical values ​​of 3 and 2 respectively. for The component flow rate model has a dimensionless error term. The model parameters were obtained through system identification experiments, including step 1: collecting data under conditions covering both the actual printing temperature range and the actual printing back pressure range. , Two components at different rotation speeds Different back pressures and different viscosities Measured instantaneous mass flow rate data under the given conditions; Step 2: Fit the above measured data using the least squares method and solve... , , The model parameters are obtained. The model predictive control algorithm solves for the optimal control sequence in the finite time domain in each control cycle, with the objective function as follows:

[0068] ;

[0069] In the formula, For the first Step mixing mass ratio setting value, dimensionless, default is 1 For the first The predicted value of the mixing mass ratio in the first step, i.e. Dimensionless This is a standard reference value for the mixing mass ratio, dimensionless, and the default value is 1. for Component gear pump in the first Step speed control increment, unit for Component speed control increment standard reference value, unit This is a dimensionless control weighting coefficient, with an empirical value of 0.1 to 1.0. To predict the number of time-domain steps To control the number of time-domain steps This is the time-domain step index, with a value of or The objective function value is dimensionless. Mixed mass ratio error. Controlled within ±0.5%, that is , dimensionless.

[0070] The specific implementation of step S02 is as follows: Nitrogen gas is introduced into the printing cavity until the oxygen concentration in the cavity is lower than 500 ppm. At the same time, a ternary thiol-olefin synergistic curing system is introduced into the mixed formula and a hindered amine light stabilizer is added. By bypassing the competition between dissolved oxygen and chain growth free radicals through the sulfur-hydrogen transfer reaction, the influence of oxygen inhibition on the surface curing integrity is synergistically eliminated.

[0071] The specific implementation method of step S03 is as follows: based on the current layer thickness (unit Current layer exposure (unit ) and mixing formulation parameter vector As input, the pre-trained solidification shrinkage field prediction model is invoked to output the layer-by-layer volumetric shrinkage rate field. and the equivalent shrinkage stress value layer by layer The Fick diffusion-reaction coupled curing depth iterative solution algorithm is then used to calculate the curing profile of each layer. In the Fick diffusion-reaction coupled curing depth iterative solution algorithm, the attenuation of ultraviolet light intensity along the depth direction follows the Beer-Lambert model, as follows:

[0072] ;

[0073] In the formula, For depth place, time Light intensity per hour, unit The intensity of incident light, in units For depth coordinates, units This is a standard reference value for depth, in units of... This is the standard reference value for absorbance coefficient, in units of... As the monomer conversion rate Linearly varying absorbance coefficient, in units Its expression is as follows:

[0074] ;

[0075] In the formula, The absorbance coefficient when the monomer is completely unconverted, in units of... Determined by experiment Absorption coefficient at complete monomer conversion, in units Determined by experiment For depth place, time The monomer conversion rate at time t, dimensionless and ranging from 0 to 1, is obtained by solving the following kinetic equation. The monomer conversion rate evolution equation for the oxygen-containing inhibition competition term is as follows:

[0076] ;

[0077] In the formula, the left side of the equation This is the partial derivative of the dimensionless conversion rate with respect to dimensionless time. All terms on the right-hand side have been normalized to standard reference values, and the dimensions are unified to dimensionless. This is a time standard reference value, in units of... The chain growth rate constant is expressed in units of... Initial monomer concentration, unit This is the standard reference value for the chain growth rate constant, in units of... The oxygen inhibition rate constant is expressed in units of... For depth place, time Dissolved oxygen concentration at time, in units Initial dissolved oxygen concentration, in units This is the standard reference value for the oxygen inhibition rate constant, in units of... This is a standard reference value for light intensity, in units of... . The experimental determination included: Step 1: Preparing a set of thin film samples with different photoinitiator concentrations and light intensities, and measuring the conversion rate-time curves for each set; Step 2: Fitting the conversion rate-time curves using the least squares method to obtain... . The experimental determination included step 1: preparing a thin film sample under conditions of known initial dissolved oxygen concentration; and step 2: measuring the surface conversion rate-time curve and comparing it with the conversion rate curve under an oxygen-free environment, using the difference to obtain the result. The evolution of dissolved oxygen concentration follows the Fick diffusion equation, as follows:

[0078] ;

[0079] In the formula, The diffusion coefficient of dissolved oxygen in the resin is expressed in units of... The experimental determination includes the following steps: Step 1: preparing a resin film sample with a known initial dissolved oxygen concentration; Step 2: measuring the change curve of dissolved oxygen concentration over time at different depths under an inert atmosphere; Step 3: fitting the above concentration-time-depth data using Fick's second law to obtain... This is the standard reference value for the diffusion coefficient, in units of... The iterative convergence criterion is the conversion rate field of two adjacent steps. Norm difference satisfies:

[0080] ;

[0081] In the formula, For the first The transformation rate field vector obtained from the next iteration is dimensionless. Index of iteration steps For vectors Norm. Based on the layer-by-layer volumetric shrinkage rate field. Perform online compensation for the printing path, compensation amount satisfy:

[0082] ;

[0083] In the formula, To print in-plane coordinates Path compensation amount at location, unit This is a standard reference value for compensation, in units of... The method is obtained through experimental calibration, including step 1: preparing a set of elastic specimens with known volume shrinkage rates and measuring the actual molding size deviation; step 2: determining the dimensional deviation based on the linear regression results of shrinkage rate and dimensional deviation. This is a dimensionless standard reference value for volumetric shrinkage rate, obtained from the corresponding value under the baseline working condition in the finite element simulation training dataset. , To print coordinates in the plane, units .

[0084] The specific implementation of step S04 is as follows: After each layer is cured, a low dose of supplemental exposure is applied to the current cured layer. (unit This ensures that the interlayer interface retains the target proportion of active double bonds, with the current layer thickness set to 20. The new layer of resin copolymerizes with the active double bonds to form an active interface layer. The initial values ​​were determined through the following experiments, including step 1: preparing a set of Different elastic specimens; Step 2: Test the interlaminar peel strength and surface acrylate double bond retention rate of each group of specimens; Step 3: Determine the minimum value corresponding to the interlaminar peel strength meeting the design lower limit and the acrylate double bond retention rate not less than 15%. As initial values.

[0085] The specific implementation of step S05 is as follows: Interface quality evaluation function The formula is expressed as follows:

[0086] ;

[0087] In the formula, The value of the interface quality evaluation function for the current layer is dimensionless. The standard reference value for the interface quality evaluation function is dimensionless and derived from... , , Substituting into the formula, we can calculate the result. This refers to the layer-by-layer equivalent shrinkage stress value output by the solidification shrinkage field prediction model, in units of... This is the standard reference value for the equivalent shrinkage stress layer by layer, in units of To supplement the current low dose of exposure, unit To supplement the standard reference value for low-dose exposure, in units For the current layer thickness, unit This is the current standard reference value for layer thickness, in units of... In the above formula, all ratio terms are dimensionless, and both sides of the equation are dimensionless quantities. , , The determination was made through experimental regression analysis, including step 1: preparing a set of elastic samples with different combinations of current layer thickness and low-dose supplemental exposure; step 2: testing the interlayer peel strength of each set of samples, using the parameter combination corresponding to the interlayer peel strength meeting the design lower limit as the standard reference value, and determining it after multiple experimental data regression analyses. Based on... The dynamic adjustment rule for the value's interval is: when At that time, Increase by 10%, Keep unchanged; when hour, and All remain unchanged; when At that time, Reduce by 20%; when At that time, Reduce by 20% and Increase by 20% and trigger the curing shrinkage field prediction model to recalculate the online printing path compensation amount for the current layer.

[0088] The specific implementation method of step S06 is as follows: After printing, the molded part is placed in an ultraviolet light box for overall post-curing treatment, and the entire molded part is uniformly exposed to further cross-link the residual active double bonds, thereby improving the uniformity of the overall cross-linking density of the molded part. The exposure time of the overall post-curing treatment is determined by Fourier transform infrared spectroscopy experiments, including step 1: Fourier transform infrared spectroscopy tests on molded part samples with different post-curing exposure times; step 2: The shortest exposure time corresponding to the decrease of the characteristic peak area of ​​acrylate double bonds to less than 5% of the initial value is used as the standard exposure time, which is determined after multiple experimental verifications, thereby obtaining a photocurable two-component elastic material molded part with uniform mechanical properties.

[0089] To better understand and implement this invention, the following is a specific application scenario of the invention, Example 2: To verify the technical solution of this invention, technicians built a photocurable two-component 3D printing test platform. A two-component elastic resin system with polyether-type polyurethane acrylate oligomer as the main component was selected. Component A and component B were mixed in a 1:1 mass ratio. Component A consisted of 50% polyether-type polyurethane acrylate oligomer, 15% aliphatic polyurethane acrylate oligomer, 10% isobornyl acrylate, 8% polyethylene glycol diacrylate, 2% 2-hydroxy-2-methyl-1-phenyl-1-propanone, and 1% sebacic acid bis(2,2,6,6-tetramethyl-4-piperidinyl) ester. Component B consisted of 30% pentaerythritol tetra(3-mercaptopropionate), 15% triallyl isocyanurate, 53% polyethylene glycol diacrylate, and 2% 2-hydroxy-2-methyl-1-phenyl-1-propanone. The sum of the mass ratios of all components was 100%. The printing chamber is filled with high-purity nitrogen gas, and the oxygen concentration inside the chamber is monitored by an oxygen sensor and stably maintained below 480ppm, meeting the control upper limit requirement of 500ppm.

[0090] Technicians conducted a system identification experiment on the dual-pump flow coupling dynamics model. The experimental temperature range was 25–45℃, and the back pressure range was 0.05–0.3 MPa. After collecting measured instantaneous mass flow data under various operating conditions, the model parameters were fitted using the least squares method. The sampling period of the Coriolis mass flow meter was set to 80 ms. The prediction time domain of the model predictive control algorithm was 10 steps, and the control time domain was 3 steps. The real-time monitoring data of the mixing mass ratio during the entire printing cycle is shown in Table 1.

[0091] Table 1. Statistics of Mixed Quality Ratio Errors Throughout the Printing Cycle

[0092]

[0093] As shown in Table 1, the mixing mass ratio error did not exceed the control limit of ±0.5% throughout the entire printing cycle. The model predictive control algorithm has a good suppression effect on the dual disturbances of nonlinear viscosity coupling and back pressure fluctuation.

[0094] The training dataset for the curing shrinkage field prediction model was obtained using finite element simulation software at layer thicknesses of 10–50 μm and exposure rates of 10–200 μm. Offline simulation was performed within the entire process parameter space, generating a total of 2400 working condition samples. The physical information neural network uses the residual terms of the mechanical equilibrium equations... The boundary condition residuals are used as physical constraints for joint training. An adaptive weight balancing mechanism dynamically adjusts the weight ratio between the data fitting term and the physical residual term during training. Figure 2 As shown, the inference time after training convergence is 3.2ms / layer, which meets the real-time requirements for online printing. Figure 3 As shown, the prediction results of the solidification shrinkage field prediction model for the layer-by-layer volume shrinkage rate field in the entire process parameter space are in good agreement with the reference values ​​of the finite element simulation. Moreover, it can still maintain physical consistency in the extrapolation interval outside the training data boundary, which reflects the role of physical constraint embedding in improving the generalization ability.

[0095] The Fick diffusion-reaction coupled curing depth iterative solution algorithm is used with a dissolved oxygen concentration of 480 ppm and a current layer exposure of 60. The algorithm predicts the solidification profile using a layer thickness of 15 μm as input. It employs 100 finite difference nodes in the depth direction and uses an adaptive Runge-Kutta integration method for time steps, with the conversion rate field between adjacent steps being used as the input. norm difference less than As a convergence criterion, the curing depth output after the iterative solution converged was 18.3 μm, which deviated from the measured curing depth of 18.6 μm by less than 2%, verifying the accuracy of the bidirectional coupling iterative solution of the light propagation physical model and the chemical reaction kinetics constraint.

[0096] Technicians conducted calibration experiments on the initial values ​​of low-dose supplemental exposure, preparing low-dose supplemental exposure values ​​of 5, 10, 15, 20, and 25. The elastic specimens were tested, and the interlaminar peel strength and surface acrylate double bond retention rate of each group of specimens were measured. The results are shown in Table 2.

[0097] Table 2 Results of Low-Dose Supplemental Exposure Calibration Experiment

[0098]

[0099] As shown in Table 2, the low-dose supplemental exposure is 15. At that time, the interlayer peel strength met the design lower limit of 2.2. Furthermore, the retention rate of acrylate double bonds is not less than 15%, therefore 15% As the initial value for low-dose supplemental exposure, the current layer thickness is set to 15 μm, meeting the requirement of less than 20 μm, ensuring sufficient translayer copolymerization between the new layer resin and the active double bonds.

[0100] The standard reference value for the interface quality assessment function was determined through regression analysis. , , , Set to 1.00. During the printing process, the curing shrinkage field prediction model outputs the equivalent shrinkage stress values ​​of each layer in real time, and the interface quality assessment function is based on each layer. The low-dose supplemental exposure and layer thickness were dynamically adjusted within the specified range. During a complete 1200-layer printing cycle, the low-dose supplemental exposure was increased by 10% 87 times, the layer thickness was reduced by 20% 23 times, and the layer thickness was reduced by 20% while the low-dose supplemental exposure was increased by 20% and path compensation was recalculated 6 times. The interlayer adhesion strength remained within the target range, and no interlayer peeling or cracking occurred.

[0101] After printing, the molded parts are placed in an ultraviolet light chamber for overall post-curing treatment, with the light intensity set to 35. The exposure time was set to 20 minutes. Fourier transform infrared spectroscopy confirmed that the characteristic peak area of ​​the double bond of acrylate decreased to 3.8% of the initial value, which met the convergence criterion of less than 5%. The crosslinking density of each layer of the molded part tended to be consistent, and the mechanical properties were uniform.

[0102] The advancements of this invention compared to traditional methods are reflected in the following aspects: Traditional offline finite element simulation methods cannot match the printing cycle due to computation time, and can only rely on empirical parameters to fix the printing path. However, the physical information neural network of this invention compresses the computation time of the inference stage to the millisecond level by embedding the mechanical equilibrium equation constraint into the loss function, and can still maintain physical consistency in the extrapolation interval, fundamentally opening up the technical path between real-time prediction of curing shrinkage field and online path compensation. Traditional fixed exposure strategies cannot dynamically adjust the retention rate of interlayer active double bonds based on the layer-by-layer shrinkage stress state. However, this invention quantifies the comprehensive effect of equivalent shrinkage stress, exposure amount and layer thickness into a single scalar through the interface quality evaluation function and implements closed-loop adjustment, so that the interlayer covalent crosslinking density is continuously maintained within the target range, and the interlayer interface is transformed from a weak adhesion region dependent on physical wetting into a chemical bond continuous transition region, eliminating the root cause of interlayer delamination at the chemical mechanism level.

[0103] It should be noted that the variables involved in this invention are explained in detail in Tables 3 and 4.

[0104] Table 3. Variable Explanation Table (Part 1)

[0105]

[0106] Table 4. Variable Explanation Table (Part Two)

[0107]

[0108] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for preparing a photocurable two-component 3D printing elastic material, characterized in that, Includes the following steps: Component A and Component B of the two-component elastic resin system are placed in two gear pump-driven feeding units respectively. The instantaneous mass flow rates of the two components are collected in real time by Coriolis mass flow meters. The dual-pump flow coupling dynamic model is driven by a model predictive control algorithm to control the mixing mass ratio error of component A and component B within ±0.5%. The printing cavity is filled with nitrogen until the oxygen concentration inside the cavity is below 500 ppm. At the same time, a ternary thiol-olefin synergistic curing system is introduced into the mixed formulation of the two-component elastic resin system, and a hindered amine light stabilizer is added to eliminate the effect of oxygen inhibition on the surface curing integrity. Before each layer exposure, the current layer thickness, current layer exposure amount, and mixing formula parameters are used as inputs to call a pre-trained curing shrinkage field prediction model for inference. This curing shrinkage field prediction model employs a physical information neural network architecture, with the intermediate layers using sinusoidal activation functions to fit the high-frequency spatially varying stress field. The output layer provides the layer-by-layer volumetric shrinkage rate field and the layer-by-layer equivalent shrinkage stress value, compressing the inference time to the millisecond level. The training dataset for the curing shrinkage field prediction model is used in finite element simulation software for layer thicknesses of 10–50 μm and exposure amounts of 10–200 μm. The entire process parameter space of the mixed formulation parameters is obtained through offline simulation. During training, finite element simulation samples are used as supervision data, and the residual terms of the mechanical equilibrium equations are included. The boundary condition residuals are embedded as physical constraints in the loss function. An adaptive weight balancing mechanism dynamically adjusts the weight ratio between the data fitting term and the physical residual term to ensure the network provides physically consistent predictions within the extrapolation interval. The Fick diffusion-reaction coupled curing depth iterative solution algorithm is used to calculate the curing profile of each layer. This algorithm couples the Beer-Lambert light attenuation model with the free radical diffusion-reaction kinetic equations of the oxygen-containing polymerization inhibition competition term to form a system of partial differential equations. The depth direction is discretized using the finite difference method, and the time direction is integrated using the adaptive Runge-Kutta method. At each step, the absorbance is updated based on the current monomer conversion rate, and then the light intensity distribution is updated. norm difference less than As a convergence criterion, the obtained solidified contour information directly participates in the online correction of the layer-by-layer volume shrinkage rate field, and the corrected shrinkage rate field drives the real-time calculation of the online compensation amount of the printing path. After each layer is cured, a low dose of supplemental exposure is applied to the current cured layer to retain the target proportion of active double bonds at the interlayer interface. The current layer thickness is set to less than 20 μm so that the new layer resin copolymerizes with the active double bonds to form an active interface layer. Extract the layer-by-layer equivalent shrinkage stress value output from the curing shrinkage field prediction model, calculate the interface quality assessment function value, and dynamically adjust the low-dose supplementary exposure and the current layer thickness according to the interval to which the interface quality assessment function value belongs, so that the interlayer bonding strength is maintained within the target range. After printing, the molded parts undergo overall post-curing treatment to eliminate residual active double bonds and obtain photocured two-component elastic material molded parts with uniform mechanical properties.

2. The method for preparing photocurable two-component 3D printing elastic material according to claim 1, characterized in that, The working principle of the Coriolis mass flow meter is as follows: when the fluid flows through the vibrating tube, the Coriolis force causes the vibrating tube to generate a phase difference proportional to the instantaneous mass flow rate. The controller calculates the phase difference in real time to obtain the instantaneous mass flow rate. The sampling period reaches the sub-hundred-millisecond level, thereby providing a high-frequency instantaneous mass flow rate feedback signal for the model predictive control algorithm.

3. The method for preparing photocurable two-component 3D printing elastic material according to claim 2, characterized in that, The model predictive control algorithm uses the dual-pump flow coupling dynamics model as the prediction model. It solves the optimal control sequence in the finite time domain in each control cycle. The ratio error of the instantaneous mass flow rates of component A and component B is included in the objective function. The dual-pump flow coupling dynamics model is systematically identified by experimentally measuring the instantaneous mass flow rate data under different back pressures, rotational speeds, and viscosity conditions of the two-component elastic resin system. The model parameters are obtained by fitting the data using the least squares method.

4. The method for preparing a photocurable two-component 3D printing elastic material according to claim 3, characterized in that, In the mixed formulation of the two-component elastic resin system, component A consists of 40%–60% polyether-type polyurethane acrylate oligomer, 10%–20% aliphatic polyurethane acrylate oligomer, 5%–15% isobornyl acrylate, 5%–10% polyethylene glycol diacrylate, 1%–3% 2-hydroxy-2-methyl-1-phenyl-1-propanone, and 0.5%–1.5% hindered amine light stabilizer; component B consists of 20%–40% pentaerythritol tetrakis, 10%–20% triallyl isocyanurate, 30%–50% polyethylene glycol diacrylate, and 1%–3% 2-hydroxy-2-methyl-1-phenyl-1-propanone; the mixing mass ratio of component A to component B is 1:

1.

5. The method for preparing a photocurable two-component 3D printing elastic material according to claim 4, characterized in that, The working principle of the ternary thiol-olefin synergistic curing system is as follows: After the photoinitiator generates free radicals, pentaerythritol tetra transfers hydrogen atoms to triallyl isocyanurate through a sulfur-hydrogen transfer reaction to generate carbon free radicals. The carbon free radicals then copolymerize with the acrylate double bonds in the polyether-type polyurethane acrylate oligomer, bypassing the competitive reaction between dissolved oxygen and chain-growing free radicals. The hindered amine light stabilizer sebacic acid diester further consumes residual dissolved oxygen by capturing oxygen free radicals.

6. The method for preparing photocurable two-component 3D printing elastic material according to claim 5, characterized in that, The working principle of the Fick diffusion-reaction coupled curing depth iterative solution algorithm is as follows: A Beer-Lambert attenuation model for ultraviolet light propagating along the depth direction is established, along with a set of free radical diffusion-reaction kinetic equations containing oxygen-containing polymerization inhibition competition terms. These two sets of equations are coupled to form a set of partial differential equations. The spatial grid is discretized using the finite difference method in the depth direction, and the adaptive Runge-Kutta method is used for integration in the time direction. The convergence criterion is the conversion rate field between two adjacent steps. norm difference less than .

7. The method for preparing a photocurable two-component 3D printing elastic material according to claim 6, characterized in that, The initial value of the low-dose supplemental exposure of the active interface layer is determined by the following method: a group of elastic samples with different low-dose supplemental exposures are prepared, and the interlayer peel strength and surface acrylate double bond retention rate of each group of samples are tested. The lowest low-dose supplemental exposure corresponding to the interlayer peel strength meeting the design lower limit and the acrylate double bond retention rate not less than 15% is taken as the initial value.

8. The method for preparing photocurable two-component 3D printing elastic material according to claim 7, characterized in that, The threshold of oxygen concentration below 500 ppm in the cavity was determined as follows: elastic film samples were prepared with the same mixing formulation parameters and current layer exposure under different oxygen concentration gradients in the cavity. The thickness of the adhesive layer on the surface of each group of samples and the surface elongation at break were tested. The highest oxygen concentration in the cavity corresponding to the surface elongation at break not being lower than 95% of the bulk elongation at break was taken as the control upper limit.

Citation Information

Patent Citations

  • Additive manufacturing temperature field prediction method fusing physical characteristics and physical constraints

    CN118821504A

  • Full-section 3D printing combined component

    CN210288863U