3D printing auxiliary arc-shaped mirror wall forming method of flexible optical lens array

Abstract

The invention discloses a 3D printing auxiliary arc-shaped mirror wall forming method for a flexible optical lens array. The method comprises the steps that an ideal optical model of the flexible optical lens array is established, and physical parameters of a forming material are obtained; performing multi-physical field coupling simulation on the whole forming process to predict and quantify an initial deviation field generated by physical deformation; according to the deviation field, a pre-distortion mold three-dimensional model capable of actively compensating deformation is generated through iterative calculation of a reverse topological optimization algorithm; printing a high-precision physical mold according to the model by using additive manufacturing equipment; injecting a liquid flexible material into the physical mold for curing and molding, and demolding to obtain a flexible optical lens array; according to the invention, active and high-precision control of forming deviation is realized, the product precision and research and development efficiency are improved, the manufacturing cost is reduced, and a set of standardized and intelligent solution is provided for precision manufacturing of complex flexible optical elements.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of computer-aided technology, and in particular to a 3D printing assisted arc mirror wall forming method of a flexible optical lens array. BACKGROUND

[0002] With the rapid development of cutting-edge technologies such as wearable devices, augmented / virtual reality (AR / VR), biomedical imaging, and robot vision, flexible optical lens arrays with advantages such as lightweight, conformal, and large-area preparation have become a hot topic in research and application. Currently, the preparation of such optical elements mainly relies on soft lithography or mold replication technology. The core process is to first manufacture a rigid negative mold with an ideal optical surface through precision machining or lithography, and then inject liquid optical polymer (such as polydimethylsiloxane PDMS) into the mold, and finally obtain a flexible product through solidification, demolding, and other processes. This method relies on the accurate replication of the mold to the geometry of the final product in principle.

[0003] However, in actual precision manufacturing, the above-mentioned traditional process path has an insurmountable technical bottleneck, i.e., low forming precision fidelity. The fundamental reason is that during the transition from liquid to solid, the liquid optical polymer is accompanied by complex multi-physical field coupling effects. First, the polymerization reaction of the material will cause significant solidification volume shrinkage. Second, the temperature change during solidification and the thermal stress and thermal shrinkage after solidification will cause the product to deform. Finally, at the moment of demolding, the accumulated residual stress in the element will be released, causing inevitable elastic rebound. The combined effect of these physical deformations causes the final formed flexible lens array, especially the arc mirror wall surface that determines its optical performance, to deviate from the theoretical design shape of the mold by several microns to several tens of microns, severely degrading the imaging quality and optical performance of the element. The traditional process solution to this problem mainly relies on the experience of process personnel to repeatedly try and error and manually modify the mold, which is not only a costly, time-consuming, and non-standardized process, but also lacks predictability and universality when dealing with different materials and different structural elements, and has failed to meet the requirements of modern precision optical manufacturing for high precision, high efficiency, and high reliability. SUMMARY

[0004] This section is intended to summarize some aspects of the embodiments of the present application and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the abstract and title of the specification to avoid obscuring the purpose of this section, abstract, and title, and such simplifications or omissions cannot be used to limit the scope of the present application.

[0005] In view of the above existing problems, the present application is proposed. Therefore, the present application provides a 3D printing auxiliary arc mirror wall forming method of a flexible optical lens array to solve the problems proposed in the background art.

[0006] To solve the above technical problems, the present application provides the following technical solutions: a 3D printing auxiliary arc mirror wall forming method of a flexible optical lens array, comprising: establishing an ideal optical model of the flexible optical lens array, and obtaining physical parameters of a liquid flexible material used for forming; based on the ideal optical model, predicting and quantifying an initial deviation field generated by the liquid flexible material due to physical deformation during the forming process through multi-physical field coupling simulation of the entire forming process of the liquid flexible material; according to the initial deviation field, generating a pre-distortion mold three-dimensional model capable of compensating for the physical deformation through iterative calculation by a reverse topology optimization algorithm; using an additive manufacturing device to print a physical mold according to the pre-distortion mold three-dimensional model; injecting the liquid flexible material into the physical mold for solidification forming, and then demolding to obtain a flexible optical lens array.

[0007] As a preferred scheme of the 3D printing auxiliary arc mirror wall forming method of the flexible optical lens array, the physical parameters include: rheological property parameters, thermal physical property parameters, solidification kinetics parameters, solidification shrinkage property parameters, and mechanical property parameters after solidification of the liquid flexible material.

[0008] As a preferred scheme of the 3D printing auxiliary arc mirror wall forming method of the flexible optical lens array, the multi-physical field coupling simulation includes: fluid dynamics simulation of the injection and filling process of the liquid flexible material; heat transfer and chemical reaction kinetics coupling simulation of the solidification process of the material; and structural mechanics simulation of stress and strain and elastic rebound of the material during solidification and demolding.

[0009] As a preferred scheme of the 3D printing auxiliary arc mirror wall forming method of the flexible optical lens array, the process of predicting and quantifying the initial deviation field includes: generating an initial mold model complementary to the geometric shape of the ideal optical model; using the initial mold model to perform the multi-physical field coupling simulation to obtain a simulated deformed product model; calculating normal displacement of points on the surface of the simulated deformed product model relative to corresponding points of the ideal optical model, thereby forming the initial deviation field.

[0010] As a preferred solution of the 3D printing assisted arc mirror wall forming method of the flexible optical lens array of the present application, wherein: the objective function of the inverse topology optimization algorithm is to minimize the geometric shape error between the surface of the final formed product and the surface of the ideal optical model.

[0011] As a preferred solution of the 3D printing assisted arc mirror wall forming method of the flexible optical lens array of the present application, wherein: the gradient of the objective function with respect to the geometry of the mold surface is calculated using the adjoint state method, and the pre-distortion mold three-dimensional model is iteratively updated according to the gradient until the objective function converges.

[0012] As a preferred solution of the 3D printing assisted arc mirror wall forming method of the flexible optical lens array of the present application, wherein: after generating the pre-distortion mold three-dimensional model, before printing the physical mold, further comprising: Collecting real appearance data of the flexible optical lens array obtained in actual production, and comparing it with the appearance data predicted by multi-physical field coupling simulation to obtain a residual field; Based on a plurality of the residual fields accumulated in history and corresponding process parameters and mold geometric features, training an artificial intelligence model for predicting residual error; Using the residual error predicted by the artificial intelligence model to correct the generated pre-distortion mold three-dimensional model.

[0013] As a preferred solution of the 3D printing assisted arc mirror wall forming method of the flexible optical lens array of the present application, wherein: the artificial intelligence model is a graph convolution network model, and the graph convolution network model is used to directly process lens array surface data represented in a non-Euclidean grid structure.

[0014] As a preferred solution of the 3D printing assisted arc mirror wall forming method of the flexible optical lens array of the present application, wherein: the additive manufacturing equipment is a light-curing 3D printer with a resolution better than 50 microns.

[0015] As a preferred solution of the 3D printing assisted arc mirror wall forming method of the flexible optical lens array of the present application, wherein: after obtaining the flexible optical lens array, further comprising: Performing three-dimensional appearance measurement on the obtained flexible optical lens array; Using the measured three-dimensional appearance data as new samples to update the artificial intelligence model for predicting residual error, so as to realize adaptive optimization of the forming method.

[0016] Compared with the prior art, the beneficial effects of the scheme of the present application are: 1. By multi-physical field coupling simulation, the physical deformation in the forming process is accurately predicted, and a "reverse" pre-distortion compensation mold is generated by using a reverse topology optimization algorithm. Through the active control method of "prediction first, then compensation", the geometric deviation problem caused by solidification shrinkage, thermal shrinkage and elastic rebound is fundamentally solved. The forming precision of the final product can be improved to the micron or even sub-micron level, while the research and development cycle is shortened and the mold testing cost is reduced; 2. The pre-distortion mold generated by the optimization algorithm usually has complex and non-uniform micro-topological features on the surface, which is difficult or impossible to economically and efficiently process by traditional subtractive manufacturing (such as CNC). The present application uses high-resolution light-cured 3D printing technology to quickly and accurately convert the digital model carrying all the compensation information into a physical mold, realizing the conversion from digital twin to physical reality; 3. By collecting the topographic data of real products, continuously training and updating the model, random factors and systematic errors that the simulation model cannot fully capture can be learned and compensated for, so that the entire forming method has intelligent characteristics of self-learning and self-improvement, and can continuously improve precision and stability with the accumulation of production data. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained without creative labor. Among them: Figure 1 The overall flowchart of the 3D printing auxiliary arc mirror wall forming method of the flexible optical lens array of an embodiment of the present application. DETAILED DESCRIPTION

[0018] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should be within the scope of protection of the present application.

[0019] In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present application. However, it will be apparent to one skilled in the art that the present application can be practiced without the specific details set forth in this description. In other instances, well-known methods, procedures, components, and circuits have not been described in detail so as not to unnecessarily obscure aspects of the present application.

[0020] It should also be noted that, as used in the specification and in the claims, the article "a", "an", or "the" is intended to mean that there is one or more of the features or elements. Unless otherwise indicated, the use of the terms "coupled" and / or "connected", and any variations thereof, are intended to mean either an indirect or direct connection in an electrical circuit and / or in signal transmission between components. In addition, terms such as first and second, top and bottom, and / or other positional descriptions are used herein for purposes of illustration, but not limitation. Terms such as "first", "second", "third", etc. are used herein not to imply an importance or a temporal meaning, but simply to distinguish one element from another.

[0021] The present application is described in detail below with reference to the attached drawings.

[0022] In the description of the present application, it should be noted that the terms "upper", "lower", "inner" and "outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application. In addition, the terms "first", "second" or "third" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance.

[0023] Unless otherwise expressly specified and limited, the terms "mounting", "connection", "connection" in the present application should be understood broadly, for example: it can be fixed connection, detachable connection or integral connection; it can also be mechanical connection, electrical connection or direct connection, it can also be indirectly connected through intermediate medium, or it can be the internal communication of two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0024] Embodiment 1 Reference Figure 1 For the first embodiment of the present application, the embodiment provides a 3D printing auxiliary arc mirror wall forming method of a flexible optical lens array, comprising: S1, an ideal optical model of a flexible optical lens array is established, and the physical parameters of a liquid flexible material used for forming are obtained; It should be noted that this step is divided into establishing an ideal optical model and obtaining the physical parameters of the liquid flexible material. Further, the three-dimensional geometric model of the flexible optical lens array with ideal optical performance, which is designed and optimized by professional optical design software (such as ZEMAX, Code V, etc.), is imported into computer-aided design (CAD) or finite element analysis (FEA) software; Specifically, the three-dimensional geometric model is a digital expression, which contains the precise curvature, position, size and array arrangement of the arc-shaped mirror wall of each lens unit. It is emphasized that, in order to ensure the quality of subsequent finite element meshing and the convergence of calculation, the model must be a "watertight" manifold geometry, i.e. without any open boundary or geometric defects; the three-dimensional geometric model is denoted as , which is the basis and final goal of all subsequent calculations and comparisons; Further, in order to truly reproduce the physical behavior of materials in the digital twin environment, comprehensive physical performance characterization experiments need to be conducted on the selected liquid flexible material (for example, polydimethylsiloxane), and the obtained parameter data is digitized to form a material database; Specifically, these obtained parameters are physical parameters, which include: Rheological property parameters: used to describe the behavior of the material during flow and filling; the viscosity ( ) of the material under different temperatures (T) and shear rates ( ) is measured by a rotational rheometer, and the experimental data can be fitted by the Cross-WLF model to accurately describe the non-Newtonian fluid characteristics and temperature dependence of the polymer melt, in this embodiment, the mathematical formula can be expressed as: wherein, is the zero-shear viscosity, and its dependence on temperature is described by the Williams-Landel-Ferry (WLF) equation; is the shear stress sensitivity parameter; q is the power-law index; by fitting operation, the rheological property parameter set of the material: and the related constants of the WLF equation} can be obtained; Thermal physical property parameters: including the density, specific heat capacity, thermal conductivity and thermal expansion coefficient of the material, which describe the response of the material during heat transfer and temperature change, and are the basis for thermal-structural coupling simulation; Curing Kinetics Parameters: used to describe the chemical reaction rate of the material from liquid to solid; by differential scanning calorimetry (DSC) at different heating rates, the relationship between the degree of curing and the change of time and temperature can be obtained, and these parameters can be fitted by the Kamal-Sourour autocatalytic reaction model to accurately describe the reaction kinetics characteristics of the curing process, in this embodiment, the mathematical formula of the Kamal-Sourour autocatalytic reaction model can be expressed as: wherein, and is the reaction rate constant following the Arrhenius equation, m and n are the reaction order, by fitting the DSC curve, a series of curing kinetics parameters such as activation energy, pre-exponential factor and reaction order can be obtained, at the same time, the total reaction heat can also be obtained for calculating the internal heat source in the curing process; Curing Shrinkage Characteristics Parameters: one of the main physical factors leading to molding deviation; by high-precision pycnometer method or pressure-volume-temperature (PVT) analyzer, the volume shrinkage rate of the material during the entire curing process (i.e. the degree of curing from 0 to 1) can be measured, which can be usually established as a function related to the degree of curing; Mechanical Properties Parameters after Curing: used to describe the elastic deformation behavior of the completely cured material under the action of force (such as demolding force), by dynamic mechanical analyzer (DMA) or tensile testing machine, the Young's modulus and Poisson's ratio of the material can be measured, which are crucial for accurately simulating the elastic rebound after demolding; S2, based on the ideal optical model, by multi-physical field coupling simulation of the entire molding process of the liquid flexible material, an initial deviation field generated by the physical deformation of the liquid flexible material during the molding process is predicted and quantified; It should be noted that this step aims to reproduce and quantify the physical deformation of the flexible optical lens array in the entire "injection-curing-demolding" process chain through high-fidelity numerical simulation; Further, according to the three-dimensional geometric model established in S1 step , an initial mold model completely complementary to the geometric shape thereof is generated by Boolean operation, denoted as ; it should be explained that this initial mold model represents the mold designed based on the ideal product shape without any compensation in the traditional process; Further, the initial mold model and the cavity defined thereby (i.e. the future product area) are discretized by high-quality finite element mesh, wherein the discretization process adopts adaptive mesh technology, and the mesh is encrypted at the key optical functional surfaces such as the arc-shaped mirror wall of the lens array, so as to ensure high-precision calculation results; Further, based on the divided finite element model and the physical parameters obtained in S1, a series of tightly coupled physical field simulations are performed according to the time sequence of the actual process. The physical simulation process is divided into three stages: The first stage: fluid dynamics simulation (filling process), which mainly simulates the process of filling the mold cavity with liquid flexible material under a certain injection pressure and speed. In this stage, the flow field is mainly solved based on the Navier-Stokes equation, and the Volume of Fluid (VOF) method is used to track the melt front. It should be noted that the purpose of this simulation is to obtain the initial distribution of temperature field, pressure field and material molecular orientation in the cavity at the end of filling, and these distributions are used as initial conditions for the subsequent solidification simulation. The second stage: coupled simulation of heat transfer and chemical reaction kinetics (solidification process), which mainly simulates the process of chemical crosslinking reaction of the material in the mold cavity under a certain solidification temperature curve. In this stage, the transient heat conduction equation and the solidification kinetics equation (such as the Kamal-Sourour model described in S1) need to be solved simultaneously. In this embodiment, the control equation of the heat conduction equation is represented as: wherein, is the internal heat source generated by the chemical reaction exothermic, which is proportional to the solidification reaction rate, i.e. By solving this coupled equation set, the temperature field and the solidification degree field of the material at any time and any position during the solidification process can be obtained; k is the thermal conductivity, is the density of the material, is the specific heat capacity, represents the gradient; The third stage: structural mechanics simulation (solidification shrinkage, thermal stress and elastic rebound), which mainly inputs the temperature field and solidification degree field obtained in the second stage as loads. Specifically, first, the total strain caused by solidification shrinkage and thermal shrinkage is calculated according to the solidification shrinkage characteristics and the thermal expansion coefficient of the material. Then, based on the mechanical properties of the solidified material (Young's modulus and Poisson's ratio), the nonlinear finite element method is used to solve the residual stress field under the constraint of the mold. Finally, the demolding process is simulated, i.e. removing the displacement constraint of the mold on the product, allowing the product to deform freely under the action of residual stress until a new mechanical equilibrium state is reached. The geometric model corresponding to this final equilibrium state is the simulated deformed product model, denoted as ; Further, the simulated deformed product model After that, it needs to be compared with the three-dimensional geometric model precisely to quantify the forming deviation; Specifically, on the optical functional surface of the product , a series of discrete points are taken , the corresponding points on the three-dimensional geometric model surface are found (the process is usually based on the nearest point or normal projection principle); then, the normal displacement between each point pair is calculated , that is, the displacement of each discrete point in the normal direction of the three-dimensional geometric model at , and all these displacement amounts constitute a scalar field, that is, the initial deviation field we want to predict and quantify ; It needs to be explained that the deviation field visually shows the degree of "bulging" or "concave" at each point on the product surface in the form of a three-dimensional topological map, which is a direct basis for the reverse compensation design of the mold; S3, according to the initial deviation field, an iterative calculation is performed through a reverse topological optimization algorithm to generate a pre-distortion mold three-dimensional model that can compensate for physical deformation; It needs to be explained that this step aims to reverse the forming deviation predicted in step S2 into an accurate modification of the mold geometry through an optimization algorithm, so as to actively "offset" the physical deformation that will occur, and this process is achieved through an iterative reverse topological optimization cycle; Further, the mold modification problem is formalized as a mathematical topological optimization problem, and a topological optimization problem usually consists of the following three parts: Design variables: the geometry of the mold cavity surface; among them, in the discretized finite element model, the design variable is expressed as the normal displacement of the mold surface grid node; Objective function: the definition of this objective function is to minimize the geometric shape error between the final molded product surface and the three-dimensional geometric model surface; specifically, it is to minimize the L2 norm square sum of the normal deviation of all surface nodes between the simulated molded product model and the three-dimensional geometric model , and its objective function can be expressed as: wherein, is the total number of nodes on the product optical surface, is the normal displacement of the th node under simulation prediction; Constraints: The entire multiphysics coupled simulation process itself constitutes a constraint on this optimization problem. That is, for any given mold shape, the product deformation is uniquely determined by the simulation process in step S2. It should be noted that in order to minimize the objective function using gradient-based optimization algorithms (such as gradient descent, conjugate gradient, etc.) It is necessary to calculate the objective function. Gradient or sensitivity relative to each mold surface design variable (i.e., nodal normal displacement) ( For the mold surface (Normal displacement of each node); If the traditional finite difference method is used to calculate the gradient, each gradient calculation requires perturbation of each design variable and rerunning a complete multiphysics simulation, which is computationally very expensive and not feasible in practice. Furthermore, to address the aforementioned issues, this invention preferably employs the Adjoint Method to efficiently calculate gradients. It should be explained that this Adjoint Method, by solving an "adjoint problem" corresponding to the original physical problem (the state problem), can simultaneously obtain the gradient of the objective function relative to all design variables within the computational cost of an additional simulation (adjoint simulation). This makes the cost of gradient calculation independent of the number of design variables, greatly improving optimization efficiency. Moreover, the entire calculation process can conceptually be understood as solving a large set of coupled equations containing state equations and adjoint equations, thereby obtaining sensitivity information. Furthermore, in obtaining the calculated gradient Then, the geometry of the mold can be iteratively updated, and at the 1st... In this iteration, the node coordinates on the mold surface Update using the following formula to generate new mold geometry. : in, It is the first The step size (learning rate) of each iteration can be determined using methods such as line search. It should be noted that the physical meaning of the above formula update process is as follows: in the simulation prediction, for the "bulging" (positive deviation) area, the mold surface is "deepened" inward; for the "depression" (negative deviation) area, the mold surface is "raised" outward. Furthermore, the initial mold geometry This refers to the initial mold model used in step S2. Then, the loop is executed: (i) Forward simulation: using the current mold geometry Run the multiphysics simulation in step S2 to obtain the geometry of the product after deformation. and objective function value ; (ii) Reverse simulation (adjoint solution): Solve the adjoint equation and calculate the gradient of the objective function with respect to the current mold geometry. ; (iii) Update the mold: Calculate the new mold geometry based on the update formula above. ; (iv) Convergence judgment: Check the objective function value Is it less than the preset convergence threshold? (For example, the average surface error corresponding to the submicron level), or whether the number of iterations has reached the upper limit. If the above conditions are met, the loop ends; otherwise, return to the above loop (i) to continue the next iteration. It should be noted that when the iteration converges, the final obtained 3D model of the mold This is the pre-distorted mold 3D model we are looking for. The surface topology of this model has undergone precise, non-uniform "reverse" distortion. Its design purpose is to enable the product to "self-correct" to the ideal geometry after undergoing a real physical forming process. In addition, in order to further improve the accuracy of the three-dimensional model of the pre-distorted mold, especially to compensate for random factors or unmodeled physical effects that the simulation model itself cannot fully capture, a correction scheme based on digital twin feedback is also provided in this embodiment. Furthermore, after generating the pre-distorted mold 3D model, but before 3D printing, the following correction steps are performed: From past production batches, high-precision 3D scanners (such as white light interferometers) are used to collect the actual product's real shape data. This real shape data is then compared with the simulation prediction results made using the same process parameters and molds to obtain a residual field. This residual field represents the difference between the "real world" and the "digital twin". As production data accumulates in each batch, a series of residual fields and their corresponding process parameters and mold geometric features are eventually obtained. By utilizing this data, an artificial intelligence model can be trained. In this embodiment, the artificial intelligence model adopts a model based on Graph Convolutional Network (GCN). It should be explained that GCN is chosen because the surface data of the lens array is essentially non-Euclidean mesh structure data, and GCN can learn directly on this structure data, thereby effectively capturing the complex nonlinear relationship between geometric features and residual distribution. Further, for a brand new, first-time production of flexible optical lens array products, since there is no historical production data, the artificial intelligence model (GCN model) for predicting residual errors is untrained in the initial stage, in which case, the present solution executes the following procedure: The step of using the GCN model for correction can be skipped; that is, the "pre-distortion mold three-dimensional model" generated by the inverse topology optimization algorithm is directly used as the final model for printing, at which time, the compensation of the molding precision mainly relies on the systematic deviation compensation based on physical simulation; The flexible optical lens array obtained in the first and subsequent initial production batches is measured for three-dimensional topography to obtain real topography data, and the real data is compared with the corresponding simulation prediction data to calculate the first batch of "residual field" data, which, together with the corresponding process parameters and mold geometric characteristics, constitutes the initial data set for training the GCN model; When a sufficient number of sample data is accumulated (i.e., after several production batches), the GCN model can be initially trained, and thereafter, when a new batch or a new product of the same type is produced, the initially trained model is used to predict residual errors and make secondary corrections to the pre-distortion mold; As production continues, new data samples are continuously added to the training set, and the GCN model is continuously updated and optimized, so that its prediction ability becomes increasingly accurate and can be generally applicable to the development of any new product; For a newly generated pre-distortion mold model , the trained GCN model is used to input the geometric characteristics of the pre-distortion mold model and the target process parameters to predict the most likely residual field that will occur in this molding, and then the predicted residual field (denoted as , which represents an additional normal displacement field) is used as an additional compensation amount to adjust the surface grid nodes of the pre-distortion mold model to obtain the final mold model ; wherein the correction process can be mathematically understood as: Through this process, the final mold geometry includes both the systematic deviation compensation based on physical simulation and the random and model error compensation based on data-driven; It should be noted that the final pre-distortion mold three-dimensional model not only compensates for predictable systematic physical deformation, but also feeds forward compensates for randomness and model errors through machine learning, thereby maximizing the molding precision of the final product; In addition, in order to construct an intelligent manufacturing system that can continuously learn and improve itself, the present embodiment also includes a feedback and optimization closed loop step: The real 3D topography point cloud data of the curved mirror wall surface of the newly prepared flexible optical lens array is obtained by using a high-precision non-contact three-dimensional measurement device such as a white light interferometer or a laser confocal microscope to perform a comprehensive scan on the curved mirror wall surface of the newly prepared flexible optical lens array; The measured real topography data is compared with the three-dimensional geometric model to calculate the actual residual error of this time of forming, and then this set of new complete data containing process parameters, mold geometry, simulation predicted deviation, and actual measured deviation is added to the training data set of the artificial intelligence model (GCN model) in the S3 step as a new high-quality sample; The artificial intelligence model is retrained or fine-tuned periodically or after a certain number of new samples are accumulated, and in this way, the model can continuously learn and capture deeper physical laws or random disturbance factors that the simulation fails to completely cover, thereby making its prediction ability of residual error stronger and stronger, and when the S3 step is performed next time to correct the mold, the retrained or fine-tuned model will be able to provide more accurate compensation suggestions; S4, using an additive manufacturing device, printing a physical mold according to the pre-distortion mold three-dimensional model; It should be noted that this step aims to convert the digital pre-distortion mold three-dimensional model carrying all the compensation information in the S3 step into a high-precision physical entity; Further, the pre-distortion mold three-dimensional model finally generated in the S3 step (if corrected by the artificial intelligence model, it is the final mold model ; otherwise ) is exported to a file format commonly used in the field of additive manufacturing, such as STL (Standard Tessellation Language) format, which describes the complex three-dimensional surface topology of the mold; Further, in this embodiment, in order to ensure that the micron-level compensation features in the pre-distortion model can be accurately reproduced, an additive manufacturing device with a resolution better than 50 microns is preferably used; specifically, a light-cured 3D printer is preferably used, and its mainstream technologies include stereolithography (SLA), digital light processing (DLP), or continuous liquid interface production (CLIP), etc. Further, the printing process using the printer is as follows: First, import the STL file of the mold into a dedicated slicing software and set key printing parameters, such as selecting a very small layer thickness (such as 10-25 microns) to minimize the "step effect" on the curved surface; according to the characteristics of the selected resin material, optimize the single-layer exposure time and energy to ensure sufficient curing and no size deviation caused by over-curing; Then, the printing direction of the mold is set in the software; in order to ensure that the optical functional surface has the best surface quality, it should be directed away from the build platform, and the support structure should not be generated directly on the surface, and all necessary support structures should be designed in the non-functional area of the mold (such as the back, side wall or positioning hole) to facilitate subsequent removal without damaging the key cavity surface; Finally, start the light-curing 3D printer, and expose and cure the photosensitive resin layer by layer according to the slicing path until the entire physical mold is printed. It should be noted that the photosensitive resin material selected should have high hardness, high thermal stability (to withstand the curing temperature of flexible materials) and low adhesion or chemical inertia with flexible materials, for example, high-temperature resistant resin or ceramic filled composite resin can be selected; Further, after printing is completed, in order to obtain a final usable physical mold, a post-processing procedure is also required: The printed mold is removed from the build platform and placed in an ultrasonic cleaning machine, and isopropanol (IPA) and other solvents are used to thoroughly clean the surface of the unhardened liquid resin; All support structures are removed manually or using tools; The cleaned mold is placed in a specific wavelength ultraviolet (UV) curing box for secondary curing for tens of minutes to several hours, aiming to make the chemical cross-linking reaction of the resin material more complete, so that the mold reaches its final mechanical properties (such as hardness, strength and dimensional stability); It should be noted that through the above printing and post-processing procedures, a physical mold can be obtained; the cavity surface of the mold is not an ideal optical surface, but carries "reverse" geometric information for compensating physical deformation; S5, inject liquid flexible material into the physical mold for curing and molding, and then demold to obtain a flexible optical lens array; It should be noted that this step aims to use the physical mold prepared in step S4, which carries pre-distortion compensation information, to manufacture a flexible optical lens array that infinitely approximates the three-dimensional geometric model in geometric appearance through an accurately controlled replication molding process; Further, before the material is injected, in order to ensure that the subsequent demolding can be successfully and damage-free, the cavity surface of the physical mold obtained in step S4 needs to be demolded, and in this embodiment, a preferred method is to perform surface silanization treatment, the specific operation is: The mold is placed in a vacuum plasma cleaning machine for several minutes of oxygen plasma treatment to activate its surface; then, it is exposed to a saturated steam environment containing a volatile release agent (for example, (1H, 1H, 2H, 2H-Perfluorooctyltrichlorosilane)), the silane molecules will chemically bond with the activated sites on the surface of the mold, forming a dense, low-surface-energy monomolecular release layer, which can greatly reduce the adhesion between the flexible material (such as polydimethylsiloxane (PDMS)) and the mold resin; Further, taking the two-component polydimethylsiloxane (PDMS) as an example, according to the recommended mass ratio of the manufacturer (usually main agent: curing agent = 10:1), the main agent and curing agent are accurately weighed using a high-precision electronic balance, and both are placed in a clean container for thorough and uniform mechanical stirring; at the same time, in order to completely eliminate the air bubbles introduced during stirring (since air bubbles are fatal defects in optical elements), the mixed liquid PDMS needs to be vacuum degassed, and the container containing the PDMS is placed in a vacuum dryer or a planetary centrifugal mixer, vacuumed to below -0.1 MPa, and maintained for tens of minutes until no air bubbles escape is observed by naked eye; Further, the thoroughly degassed liquid flexible material is slowly poured along one side of the mold or injected into the surface-treated physical mold cavity through a low-pressure injection system, it needs to be noted that this operation process should avoid generating new turbulence and air entrainment, in addition, in order to ensure that the material completely fills all the microstructures, the entire mold can be placed again in a vacuum environment for a short air pumping after injection to remove the small air bubbles that may remain in the corners of the cavity; Further, the mold filled with material is placed horizontally in a programmable precision oven, since any deviation in temperature or time will lead to deviation of the compensation effect, therefore, it is necessary to strictly follow the same solidification temperature curve set in the multi-physical field coupling simulation of S2 step (for example, heating at a constant temperature of 70°C for 2 hours) to reproduce the simulated thermal history, so as to ensure that the pre-distortion design on the mold can correctly compensate for the actual physical deformation (solidification shrinkage, thermal shrinkage, etc.); Further, after the solidification is completed, the mold is taken out of the oven and naturally cooled to room temperature, since the material of the flexible optical lens array is soft and the structure is delicate, the demolding process must be extremely careful, usually starting from one corner of the flexible optical lens array, using a non-sharp tool (such as a Teflon tweezers or a soft spatula) to gently lift it up, then, at a gentle angle, slowly and steadily peeling off the entire flexible optical lens array from the physical mold (wherein, thanks to the previous surface release treatment, the resistance during this process will be very small), thereby avoiding stretching deformation or tearing damage to the elements during demolding; It should be noted that, at this point, the final product - a flexible optical lens array - is obtained, since the cavity surface of the physical mold was previously subjected to an accurate "reverse" pre-distortion, and, at the same time, the liquid flexible material, after having undergone a series of complex physical deformations (solidification shrinkage, thermal shrinkage, elastic springback), its final stable configuration "self-corrects" and accurately recovers the three-dimensional geometric model defined in step S1.

[0025] Those skilled in the art will appreciate that embodiments of the present application can be readily used as a method, a system or a computer program product. Accordingly, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) embodying computer-readable program code. Embodiments of the present application are also directed to computer program products comprising computer- readable program code embodied in a computer-usable storage medium.

[0026] The present application is described below with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded processing system or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart Figure 1 one or more functions specified in the flowchart block or blocks. Figure 1 one or more functions specified in the flowchart block or blocks.

[0027] These computer program instructions can also be stored in a computer- readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the function specified in the flowchart block or blocks. Figure 1 one or more functions specified in the flowchart block or blocks. Figure 1 one or more functions specified in the flowchart block or blocks.

[0028] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 ​one or more processes and / or blocks Figure 1 the steps of the functions specified in the one or more blocks.

[0029] While the preferred embodiments of the application have been described, additional variations and modifications can be employed by those skilled in the art once they learn of the basic inventive concepts. Therefore, the appended claims are intended to cover all such variations and modifications as fall within the scope of the application.

[0030] It is apparent that many modifications and variations of this application can be effected although only a few have been chosen for purposes of illustrative discussion. Thus, it is contemplated to cover the present application as broadly as the prior art allows and the following claims are intended to cover all such modifications and variations as fall within the scope of the application.

Claims

1. A method for 3D printing-assisted arc-shaped mirror wall forming of a flexible optical lens array, characterized in that, include: An ideal optical model of the flexible optical lens array is established, and the physical parameters of the liquid flexible material used for molding are obtained; Based on the ideal optical model, by performing multi-physics field coupling simulation on the entire molding process of the liquid flexible material, an initial deviation field generated by the physical deformation of the liquid flexible material during the molding process is predicted and quantified. Based on the initial deviation field, an iterative calculation is performed using an inverse topology optimization algorithm to generate a three-dimensional model of a pre-distorted mold that can compensate for the physical deformation. Using additive manufacturing equipment, a physical mold is printed based on the three-dimensional model of the pre-distorted mold. The liquid flexible material is injected into the physical mold for solidification and molding, and then demolded to obtain a flexible optical lens array.

2. The 3D printing-assisted arc-shaped mirror wall forming method for flexible optical lens arrays as described in claim 1, characterized in that, The physical parameters include: The rheological properties, thermophysical properties, curing kinetics, curing shrinkage, and mechanical properties of the liquid flexible material after curing are described.

3. The 3D printing-assisted arc-shaped mirror wall forming method for flexible optical lens arrays as described in claim 1 or 2, characterized in that, The multiphysics coupling simulation includes: Fluid dynamics simulation of the injection and filling process of liquid flexible materials; The curing process of materials is simulated by coupled simulation of heat transfer and chemical reaction kinetics; Furthermore, structural mechanics simulations were performed on the stress, strain, and elastic rebound of the material during the curing and demolding processes.

4. The 3D printing-assisted arc-shaped mirror wall forming method for flexible optical lens arrays as described in claim 3, characterized in that, The process of predicting and quantifying the initial bias field includes: Generate an initial mold model that is complementary to the geometry of the ideal optical model; The multiphysics coupling simulation is performed using the initial mold model to obtain a simulated deformed product model. The normal displacement of a point on the surface of the simulated deformed product model relative to the corresponding point of the ideal optical model is calculated to form the initial deviation field.

5. The 3D printing-assisted arc-shaped mirror wall forming method for flexible optical lens arrays as described in claim 1, characterized in that, The objective function of the reverse topology optimization algorithm is to minimize the geometric error between the surface of the final molded product and the surface of the ideal optical model.

6. The 3D printing-assisted arc-shaped mirror wall forming method for flexible optical lens arrays as described in claim 5, characterized in that, The gradient of the objective function with respect to the geometry of the mold surface is calculated using the adjoint state method, and the three-dimensional model of the pre-distorted mold is iteratively updated based on the gradient until the objective function converges.

7. The 3D printing-assisted arc-shaped mirror wall forming method for flexible optical lens arrays as described in claim 1, characterized in that, After generating the pre-distorted mold 3D model and before printing the physical mold, the process also includes: The actual morphological data of the flexible optical lens array obtained in actual production is collected and compared with the morphological data predicted by multi-physics field coupling simulation to obtain a residual field. Based on the historical accumulation of multiple residual fields and the corresponding process parameters and mold geometric features, an artificial intelligence model for predicting residuals is trained. The generated three-dimensional model of the pre-distorted mold is corrected using the residuals predicted by the artificial intelligence model.

8. The 3D printing-assisted arc-shaped mirror wall forming method for flexible optical lens arrays as described in claim 7, characterized in that, The artificial intelligence model is a graph convolutional network model, which is used to directly process lens array surface data characterized by a non-Euclidean grid structure.

9. The 3D printing-assisted arc-shaped mirror wall forming method for flexible optical lens arrays as described in claim 1, characterized in that, The additive manufacturing equipment is a photopolymer 3D printer with a resolution better than 50 micrometers.

10. The 3D printing-assisted arc-shaped mirror wall forming method for flexible optical lens arrays as described in claim 1, characterized in that, After obtaining the flexible optical lens array, the process further includes: The obtained flexible optical lens array is subjected to three-dimensional morphological measurement; The measured three-dimensional topography data is used as a new sample to update the artificial intelligence model for predicting residuals, thereby achieving adaptive optimization of the molding method.

Images