A method and system for curing shrinkage compensation and forming accuracy control in 3D printing

Abstract

This invention discloses a method and system for curing shrinkage compensation and forming accuracy control in 3D printing, relating to the field of 3D printing additive manufacturing technology. The method includes: acquiring 3D model data of the part to be formed and printing process parameters; establishing and calling a compensation model; generating layered data and target exposure information for each layer based on the 3D model; correcting the target exposure information based on the compensation model and process parameters during printing to obtain execution exposure information and complete curing; acquiring in-layer images and comparing them with layered data to calculate the current layer's deviation field; updating the compensation model based on the deviation field increment, generating the compensation amount for the next layer, and repeating this process until forming is complete. The method employs a deviation field-driven closed-loop update to achieve adaptive convergence to working condition drift, improving the consistency of key dimensions, boundary fidelity, and shape stability, reducing reliance on trial participation and experience, and supporting collaborative correction based on contour, dosage, and cycle time, thereby improving the repeatability of mass production of precision parts.

Description

Technical Field

[0001] This invention relates to the field of 3D printing additive manufacturing technology, specifically to a method and system for curing shrinkage compensation and forming accuracy control in 3D printing. Background Technology

[0002] In recent years, additive manufacturing has gradually moved from rapid prototyping to small-batch and even large-scale manufacturing. Among them, photopolymer 3D printing, represented by SLA, DLP, and MSLA, has been widely used in precision molds, medical devices, microstructured devices, and dental restorations due to its advantages in high resolution and surface quality. With advancements in projection optical engines, resin formulations, and motion platform control, single-layer exposure and peeling cycles have achieved higher repeatability. Slicing software has also evolved from simple contour generation to integrated control platforms that support grayscale control, zoned exposure, and process parameter management. Forming accuracy control has gradually become an important competitive indicator for photopolymer equipment.

[0003] However, current dimensional compensation methods for photopolymer printing still rely primarily on offline calibration and empirical correction. These methods often employ overall scaling, fixed contour offsets, or lookup table compensation for a small number of typical structures. This makes it difficult to characterize the nonlinear coupling relationship between "local geometry—energy input—shrinkage stress," leading to non-uniform shrinkage, systematic reduction in pore size, and accumulated boundary warping in structures with thin walls / thick solids, dense pores, or abrupt cross-sections. Simultaneously, batch variations in resin, environmental temperature fluctuations, release film aging, and light source attenuation introduce operational drift. Fixed parameters are difficult to adaptively adjust with each layer, and errors are often amplified through interlayer aggregation. While some solutions incorporate camera monitoring or final inspection, these often only address defect alarms or post-printing overall comparisons, lacking calculations of intra-layer deviation fields consistent with layer benchmarks, and even more so, lacking stable update mechanisms that can be completed between layers. This makes them susceptible to compensation oscillations caused by measurement noise, hindering the achievement of "online, continuous, and convergent" curing shrinkage compensation and closed-loop control of forming accuracy. In addition, existing compensation methods typically only correct geometric boundaries and lack coordinated control of exposure dose and interlayer timing, making it difficult to reduce internal stress lock-in without sacrificing forming strength; at the same time, the lack of associated storage and traceability of process data limits model iteration and batch reproduction. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by this invention is that existing methods for compensating for curing shrinkage and controlling forming accuracy in photopolymer 3D printing rely on offline static calibration, are difficult to adapt to non-uniform shrinkage and working condition drift, lack an intra-layer deviation field feedback mechanism consistent with the layer benchmark, and are susceptible to noise during online updates, leading to compensation oscillations and difficulty in convergence. The problem also lies in how to update the compensation model based on the intra-layer image during the layer-by-layer printing process and generate executable exposure correction amounts to achieve continuous closed-loop accuracy control.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, embodiments of the present invention provide a method for compensating for curing shrinkage and controlling forming accuracy in 3D printing, characterized by comprising the following steps:

[0008] Obtain the 3D model data of the part to be formed and the corresponding printing process parameters, and establish and call a compensation model to characterize the relationship between curing shrinkage forming deviation and compensation amount;

[0009] Based on the 3D model data, layered data is generated, and target exposure information for each layer is generated;

[0010] During the printing process, the target exposure information is corrected based on the compensation model and the printing process parameters for the current layer to obtain the execution exposure information for the current layer, and the current layer is cured and formed accordingly.

[0011] Acquire the solidified intralayer image data of the current layer, and calculate the deviation field of the current layer based on the intralayer image data and the layer data;

[0012] The compensation model is updated based on the deviation field, and the compensation amount of the next layer is generated. The target exposure information of the next layer is corrected based on the compensation amount and the process is repeated until the curing of all layers is completed, thereby obtaining a three-dimensional molded part with curing shrinkage compensation.

[0013] As a preferred embodiment of the curing shrinkage compensation and forming accuracy control method for 3D printing described in this invention, the step of acquiring the three-dimensional model data of the part to be formed and the corresponding printing process parameters includes:

[0014] The three-dimensional model data is subjected to geometric consistency processing and coordinate unification processing. The geometric consistency processing includes closure checks and repair of holes and broken surfaces. The coordinate unification processing includes unifying the three-dimensional model data to the printer coordinate system and recording its placement posture on the forming platform.

[0015] Obtain the printing process parameters corresponding to the current printing task. The printing process parameters include energy input parameters and forming cycle parameters that affect curing shrinkage, and further include material and environmental parameters that characterize the current working conditions.

[0016] The printing process parameters are obtained by reading from the equipment control interface, equipment operation log, integrated sensors and material management module, and by user input, and are bound to the current printing task for later retrieval.

[0017] As a preferred embodiment of the curing shrinkage compensation and forming accuracy control method for 3D printing described in this invention, the establishment and invocation of the compensation model for characterizing the correspondence between curing shrinkage forming deviation and compensation amount includes:

[0018] Geometric description information related to curing shrinkage sensitivity is extracted based on the three-dimensional model data;

[0019] A compensation model is constructed and invoked, which is configured to output an exposure compensation amount for correcting the target exposure information based on the layer data of the current printing layer, the geometric description information, and the printing process parameters when performing subsequent calculations for the current printing layer.

[0020] The exposure compensation amount includes contour boundary offset amount, local exposure dose correction amount, and exposure cycle parameter correction amount;

[0021] The compensation model is implemented in the form of lookup tables and rule bases, parameterized mapping models, and data-driven models.

[0022] Furthermore, the compensation model is configured to update its internal parameters and rules based on the deviation field.

[0023] As a preferred embodiment of the curing shrinkage compensation and forming accuracy control method for 3D printing described in this invention, the step of generating layered data based on the three-dimensional model data includes:

[0024] Based on the layer thickness in the printing process parameters, the layer spacing is set along the printing construction direction, and multiple sets of mutually parallel cross-sectional planes are established;

[0025] Intersection operations are performed on each of the said cross-sectional planes and the three-dimensional model data to obtain one or more closed two-dimensional contours corresponding to each layer, and when there are holes and cavities, the outer contour and the inner contour are organized into a set of layered contours of the same layer.

[0026] Perform contour consistency processing on the hierarchical contour set. The contour consistency processing includes contour smoothing and denoising, removal of isolated islands, merging of adjacent contours, and topology consistency verification.

[0027] Furthermore, when the contour size is close to the projection pixel resolution, the corresponding contour is subjected to connectivity repair and boundary reconstruction processing based on grid mask to form layered data that can be stably shaped.

[0028] As a preferred embodiment of the curing shrinkage compensation and forming accuracy control method for 3D printing described in this invention, wherein: the generation of target exposure information for each layer includes:

[0029] Based on the resolution and pixel size of the projection system, the set of layered contours of each layer is transformed into the projection coordinate system and rasterized to generate the exposure pattern of the corresponding layer. The area to be formed is marked as the exposure area, and the hole and cavity areas are marked as the non-exposure area.

[0030] Based on the printing process parameters, each layer is configured with target exposure parameters corresponding to the exposure pattern. The target exposure parameters include exposure time, gray level, and light intensity level.

[0031] Anti-aliasing is performed on the outline boundary of the exposure pattern, the anti-aliasing including boundary grayscale transition and subpixel boundary reconstruction;

[0032] Furthermore, the target exposure information is structured and encapsulated, including the layer's exposure pattern, target exposure parameters, contour data representation for subsequent exposure correction, attribute labels for exposure area partitioning control, and reference information for coordinate registration with the image data within the layer.

[0033] As a preferred embodiment of the curing shrinkage compensation and forming accuracy control method for 3D printing described in this invention, the step of obtaining the execution exposure information of the current layer and completing the curing and forming of the current layer accordingly includes:

[0034] The compensation model is invoked, and based on the printing process parameters and the layer data and geometric description information corresponding to the current layer, the exposure compensation amount used to correct the target exposure information is calculated.

[0035] The exposure compensation amount includes contour boundary offset amount, local exposure dose correction amount, and exposure cycle parameter correction amount;

[0036] The target exposure information of the current layer is corrected according to the exposure compensation amount to obtain the execution exposure information of the current layer. The correction includes the correction of the exposure pattern and exposure parameters.

[0037] The printing equipment is controlled to complete the curing and forming of the current layer based on the execution exposure information, wherein the execution exposure information includes the execution exposure pattern and its corresponding execution exposure parameters;

[0038] Furthermore, the execution exposure information of the current layer is associated with and stored along with the exposure compensation amount and the key operating condition information corresponding to the current layer, so as to be used for deviation field calculation and compensation model update of subsequent layers.

[0039] As a preferred embodiment of the curing shrinkage compensation and forming accuracy control method for 3D printing described in this invention, wherein: the calculation of the deviation field of the current layer includes:

[0040] The imaging module installed in the printing device acquires the in-layer image data after the current layer has been cured.

[0041] Based on preset and associated reference information, the image data within the layer is coordinate registered with the layered data of the corresponding layer. The reference information includes reference markers, alignment marker areas, and the calibration relationship between the projected coordinate system and the platform coordinate system.

[0042] The actual shaped contour and features are extracted from the in-layer image data after coordinate registration, and compared with the expected contour and expected features in the layered data to calculate the deviation field.

[0043] The deviation field is used to characterize contour boundary offset, feature size error, and region scale variation;

[0044] Furthermore, the deviation field is subjected to confidence assessment and stabilization processing to form deviation information for updating the compensation model.

[0045] As a preferred embodiment of the curing shrinkage compensation and forming accuracy control method for 3D printing described in this invention, the step of updating the compensation model according to the deviation field and generating the compensation amount for the next layer, and correcting the target exposure information of the next layer based on the compensation amount and performing the process iteratively includes:

[0046] The deviation field is regionalized and characterized to obtain error information for characterizing systematic deviations. The regionalization and characteristic representation includes statistical processing of contour boundary offset, feature size error and regional scale change according to preset region type, geometric category and attribute label.

[0047] The update basis for the compensation model is constructed based on the error information, so that the update direction of the compensation model can reduce the expected deviation of the subsequent printing layer, and the incremental update of the compensation model is performed when the preset update trigger condition is met. The preset update trigger condition includes the error amplitude threshold and the error direction consistency condition.

[0048] Based on the updated compensation model, combined with the layer data and geometric description information corresponding to the next printing layer and the printing process parameters, the compensation amount of the next printing layer is calculated. The compensation amount includes the contour boundary offset amount, the local exposure dose correction amount, and the exposure cycle parameter correction amount.

[0049] After applying amplitude and rate of change constraints to the compensation amount, the compensation amount is used to correct the target exposure information of the next printing layer to generate the execution exposure information of the next printing layer, and the process is repeated until all layers are cured and formed.

[0050] Furthermore, version management is implemented for the incremental update process of the compensation model, so that when deviation anomalies and imaging quality anomalies are detected, it can revert to the previous stable version of the compensation model and continue to execute using the previous stable compensation amount.

[0051] Secondly, embodiments of the present invention provide a 3D printing curing shrinkage compensation and forming accuracy control system, including:

[0052] Data acquisition and compensation model preparation module: acquire the three-dimensional model data of the part to be formed and the corresponding printing process parameters, and establish and call the compensation model to characterize the relationship between curing shrinkage forming deviation and compensation amount;

[0053] Layering and Target Exposure Information Generation Module: Generates layered data based on the 3D model data, and generates target exposure information for each layer;

[0054] Layer-by-layer exposure correction and curing execution module: During the printing process, the target exposure information of the current layer is corrected based on the compensation model and the printing process parameters to obtain the execution exposure information of the current layer, and the curing of the current layer is completed accordingly;

[0055] Intra-layer image acquisition and deviation field calculation module: acquires intra-layer image data after solidification of the current layer, and calculates the deviation field of the current layer based on the intra-layer image data and the layer data;

[0056] Compensation model update and next layer compensation amount generation module: Update the compensation model according to the deviation field and generate the compensation amount of the next layer. Based on the compensation amount, correct the target exposure information of the next layer and repeat the process until the curing and forming of all layers is completed, thereby obtaining a three-dimensional formed part with curing shrinkage compensation.

[0057] The beneficial effects of this invention are as follows: By updating the model incrementally layer by layer through “target exposure information – execution exposure information – intra-layer image – deviation field – model update” and driven by the deviation field, the compensation strategy can adapt to resin, temperature, peeling and optical closed loops, shifting the curing shrinkage error from offline correction at the printing back end to online suppression during the printing process; adaptive convergence of working conditions such as state, thereby improving the consistency of key dimensions, boundary fidelity and overall shape stability under complex geometry and variable working conditions, while reducing the dependence on operational experience and repeated trial and error. Attached Figure Description

[0058] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort, wherein:

[0059] Figure 1 The overall flowchart of a method for curing shrinkage compensation and forming accuracy control in 3D printing provided in the first embodiment of the present invention;

[0060] Figure 2 This is a module connection diagram of a 3D printing curing shrinkage compensation and forming accuracy control system provided in the third embodiment of the present invention. Detailed Implementation

[0061] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0062] Example 1, referring to Figure 1 This invention provides a method for curing shrinkage compensation and forming accuracy control in 3D printing, as one embodiment of the present invention.

[0063] S1: Obtain the 3D model data of the part to be formed and the corresponding printing process parameters, and establish and call the compensation model to characterize the relationship between curing shrinkage forming deviation and compensation amount.

[0064] The purpose of step S1 is to transform the geometry to be formed and the actual working conditions into structured inputs that can be directly used for exposure correction and closed-loop updates. Based on this, a compensation model capable of outputting exposure compensation amounts is established and invoked, providing a calculable basis for generating exposure information for subsequent layers. To ensure this step can be directly implemented in engineering, the following examples illustrate this with specific scenarios.

[0065] The acquisition of the 3D model data of the part to be formed can be the process of reading a standard 3D model file and importing it into the printing control system. For example, when the imported model is in STL format, the control system can first identify whether the model has holes and broken surfaces. For example, for a part with a thin-walled cavity, if the STL has missing facets at the corners of the cavity, the contour will break after layering, resulting in a lack of target exposure information and thus creating gaps at the curing boundary. To avoid such problems, the control system can perform automatic repair: filling in missing facets, deleting duplicate faces, unifying face normals, and merging nearest vertices, so that the model forms a closed contour when truncated at any height. It should also be noted that, to ensure that the subsequent in-layer image data and layered data can be aligned, the control system can unify the model to the printer coordinate system, for example, with the platform center as the origin and record the model's posture. For example, if the same part is rotated on the platform, the position of holes in the in-layer image will be completely offset. If the posture is not recorded, it will lead to errors in the deviation calculation. Therefore, recording the placement posture in this step helps to stabilize the subsequent deviation field calculation.

[0066] Furthermore, to enable the compensation model to respond to the non-uniformity of curing shrinkage, the control system can extract geometric description information related to shrinkage sensitivity from the 3D model data as model input and index information. For example, for a structure containing both thick solid regions and thin-walled cantilever regions, the heat release and cross-linking degree of the thick solid region during curing are more likely to cause local stress concentration, while the thin-walled cantilever region is more prone to warping accumulation during the peeling stage. Therefore, the control system can label categories such as "thick solid region," "thin-walled region," "long cantilever region," and "pore boundary region" in the layer pre-analysis and store these categories with each layer. As another example, for a screen structure with dense micropores, pore size deviation usually manifests as local contour shrinkage. The control system can record descriptive information such as the number of pores, the proportion of pore boundary length, and the minimum spacing between pores, so that the compensation model can output more targeted exposure compensation. It should be noted that this type of geometric description information can be obtained through conventional algorithms such as mesh traversal, cross-sectional projection, contour extraction, and neighborhood thickness estimation, without relying on complex formulas.

[0067] It should be noted that the printing process parameters are used to characterize the energy input conditions and operating status of the current printing task, and this set of parameters should be able to be called by the compensation model. For example, in MSLA and DLP equipment, the single-layer exposure time and grayscale level jointly determine the energy dose. When the ambient temperature is low and the resin viscosity is high, leveling slows down and the curing rate may change. If the same exposure strategy is still used, over-curing at the boundary and under-curing in some areas often occur, resulting in interlayer drift of shrinkage error. Therefore, this step records the ambient temperature, resin type, and batch identification together, which helps the compensation model output differentiated compensation amounts under different operating conditions. For another example, the increased number of times the release film is used will lead to an increase in peel force. Long, thin-walled areas are more likely to generate small displacements at the moment of peeling, which accumulate into dimensional deviations. Recording the release film status and number of uses as process parameters helps to explain and correct such deviations. It should also be noted that parameter acquisition can be achieved by reading preset values ​​from the device control interface, reading actual execution values ​​from the device log, collecting data from temperature and humidity sensors, reading batch information from the resin management module, and inputting material identifiers by the user; the control system preferably binds and stores the parameters with the printing task for subsequent model updates and traceability.

[0068] The compensation model characterizes the relationship between curing shrinkage forming deviation and compensation amount. Its inputs include layer data, geometric description information, and printing process parameters related to the current layer. The output is the exposure compensation amount used to correct the target exposure information. For example, the compensation amount can manifest as an adjustment to the outward expansion and inward contraction of the current layer's exposure pattern boundary, or as an adjustment to the increase or decrease of exposure grayscale in certain areas. For instance, for the boundary region of a hole, the compensation model can output an outward expansion compensation amount in the hole diameter direction to offset the reduction in hole diameter caused by curing shrinkage. For long, thin-walled cantilever regions, the compensation model can output a compensation amount that reduces the local exposure dose to reduce stress locking caused by excessively rapid curing, thereby reducing the risk of warpage. It should also be noted that, to avoid the compensation model becoming an unimplementable abstract concept, the compensation model in this embodiment has an executable implementation form: Firstly, a lookup table and rule base form, for example, selecting the corresponding boundary bias and dosage correction amounts based on resin type, layer thickness range, and geometric category labels; secondly, a parametric regression form, for example, outputting compensation amounts based on exposure time, temperature, and geometric category; and thirdly, a data-driven model form, for example, calling a trained model file and inputting the aforementioned features to output compensation amounts. Furthermore, the compensation model is preferably configured as an updatable model, meaning that after obtaining the deviation field in subsequent steps, the lookup table entries and model parameters can be adjusted. For example, when several consecutive layers exhibit a consistent inward contour deviation in the same thin-walled region, the control system can write this deviation information into the update queue, causing the compensation model to output a larger outward expansion compensation amount in the next layer, thereby achieving convergent correction within the same printing task.

[0069] Furthermore, step S1 processes the 3D model data into layerable, comparable, and calibrable benchmark data, and organizes the printing process parameters into a working condition description that can be called by the model. This enables the compensation model to have the necessary input conditions to generate exposure compensation amounts, thus providing a clear control object for subsequent exposure correction. Especially in scenarios where curing shrinkage exhibits nonlinear differences due to local thickness, cross-sectional variations, and energy input distribution, this step introduces geometric description information and process parameters to enable compensation amounts to be generated layer by layer and region by region, rather than scaling the entire model proportionally. This mechanism provides an implementable data interface for subsequent closed-loop control based on in-layer image calculation of the deviation field and model update, allowing the invention to maintain the stability and repeatability of forming accuracy control even when working condition drift exists.

[0070] S2: Generate layered data based on the three-dimensional model data, and generate target exposure information for each layer.

[0071] The purpose of step S2 is to convert the 3D model data obtained in step S1 into discrete layered data according to the printing direction, and to form target exposure information that can be directly executed by the photopolymerization device based on the layered data, thereby providing a unified expected benchmark for the subsequent closed-loop control of "exposure correction - exposure execution - deviation feedback". The layered data is used both to generate target exposure information and for subsequent alignment and comparison with image data within the layer. Therefore, this step preferably simultaneously completes the consistency processing of layer contours, the preservation of small features, and the structured encapsulation of exposure information.

[0072] The generation of layered data can be achieved using a cross-sectional discretization method based on layer thickness. Specifically, the control system sets the layer spacing according to the layer thickness in the printing process parameters and extracts the cross-sectional contours layer by layer from the 3D model data. Specifically, the control system establishes a series of equally spaced cross-sectional planes along the printing direction, performs intersection operations between each plane and the 3D model, and obtains one or more closed 2D contours for that layer. When the part has cavities and holes, the outer and inner contours can be obtained simultaneously in the same layer. The control system organizes the outer contour and the corresponding inner contour into a set of layered contours for the same layer. It should also be noted that the layering process should strictly follow the printer coordinate system and model placement posture determined in step S1 to ensure that the layered data can serve as the sole geometric reference for subsequent deviation field calculations. For example, for parts with threaded and inclined features, if the layering reference and model posture are inconsistent, the obtained cross-sectional contours will introduce projection distortion, leading to non-shrinkage factors being mixed into the deviations obtained in subsequent comparisons. Therefore, this step ensures the stability and reliability of the layering reference through coordinate unification and posture locking.

[0073] Furthermore, to ensure the manufacturability and boundary stability of subsequent exposure patterns, the control system can simplify and unify the two-dimensional contours. For example, it can perform noise reduction and smoothing on the contour point sequence, remove isolated islands smaller than a preset area threshold, merge adjacent contours with a distance smaller than a preset threshold, and perform topological consistency checks on the contours to avoid contour non-closure, self-intersection, and pseudo-small features caused by mesh discretization errors. Furthermore, addressing the common problems of lost fine features and easily broken thin walls in photopolymerization molding, the control system can perform feature preservation processing at the layered data level. For example, when a layer contains fine line structures and micropore structures with dimensions close to the projected pixel resolution, the control system can first convert the cross-sectional contour into a pixel mask, and then perform connectivity repair and boundary reconstruction processing to ensure continuous fine lines and closed hole boundaries, thereby ensuring that the subsequently generated target exposure information can be stably formed. It should also be noted that when the part exhibits abrupt changes in cross-section in certain layers (e.g., transitioning from a thin column to a large plane), the control system can record cross-sectional area change markers and thickness category markers in the layered data for subsequent compensation models to identify stress-induced risk layers, providing a priori basis for exposure correction.

[0074] It should be noted that the target exposure information represents the planned exposure content and conditions for each layer before closed-loop correction, including the exposure pattern and its corresponding exposure parameters. Specifically, the exposure pattern can be a binary or multi-grayscale bitmap of the layer, or a combination of vectorized contours and filling strategies; the exposure parameters include exposure time, grayscale level, and light intensity level, and may further include different exposure strategies for the bottom layer and ordinary layers, inter-layer waiting time, and lift-off and fall-off cycle parameters related to stripping. Furthermore, the target exposure information is preferably stored in a structured manner that associates layer number with exposure pattern and exposure parameters, so that subsequent steps can clearly locate the expected benchmark of the current layer and make corrections to obtain the execution exposure information.

[0075] Furthermore, the generation of target exposure information for each layer can be achieved as follows: the control system converts the set of layered contours of the layer into an exposure pattern in a projection coordinate system. Specifically, the control system rasterizes the contours on the projection plane to generate a bitmap mask based on the projection system resolution and pixel size; for areas requiring solid forming, the mask area is marked as the exposure area; for areas with holes and cavities, the mask area is marked as the non-exposure area; for example, for parts with internal cavities, the outer and inner contours of the same layer form a mask structure with a solid outer surface and a hollow inner surface after rasterization, so that the projected light field only exposes the solid outer area, thereby reducing the risk of the internal cavity being mistakenly solidified and blocked. It should also be noted that for projection devices such as DLP and MSLA, in order to reduce the impact of jagged boundaries on dimensional accuracy, the control system can perform anti-aliasing processing on the contour boundaries, such as introducing grayscale transitions and sub-pixel boundary reconstruction at the boundary pixels, so that the boundary energy distribution is closer to the continuous contour; the target exposure information generated by this processing can still be further corrected by the compensation model.

[0076] To support the subsequent step S3's correction of the target exposure information, the target exposure information is preferably represented using a parameterizable data structure. For example, the control system can retain the vector contour data of this layer before generating the bitmap, so that when the compensation model outputs the contour offset, geometric offset operations can be directly performed on the vector contour and then rasterized, thus avoiding boundary distortion and resolution loss caused by direct morphological transformation of the bitmap. As another example, the control system can divide the same layer's exposure area into a near-boundary region and an internal core region, and into a support adjacency region and a free boundary region, and attach attribute labels to different regions, allowing subsequent compensation amounts to apply different exposure dose corrections according to region, thereby achieving zoned exposure control and providing a data interface for internal stress distribution regulation.

[0077] Furthermore, to facilitate reliable comparison between subsequent intra-layer image data and layered data, this embodiment preferably saves reference information for alignment synchronously in the target exposure information. For example, reference points and alignment mark areas that do not participate in part forming can be set at the edge of the printing area, enabling the intra-layer images acquired by the camera to quickly complete coordinate registration based on the reference points; and the calibration relationship between the platform coordinate system and the projected coordinate system is written into the target exposure information as task-level metadata, so that the image pixel coordinates can be accurately mapped to the layered contour coordinates during subsequent deviation field calculations. It should be noted that this reference information does not change the forming geometry of the part, but is only used to improve the stability of measurement and comparison, thereby supporting the feasibility of subsequent closed-loop updates.

[0078] Furthermore, step S2, on the one hand, extracts layers of the 3D model and performs consistency processing and feature preservation on the contours to form layered data that can be directly used as the expected geometric benchmark, providing a unified reference for subsequent deviation field calculations; on the other hand, by converting the layered data into structured, parameterizable target exposure information, the compensation amount output by the subsequent compensation model can be efficiently applied in the form of contour offset and zoned dose correction, thereby avoiding the roughness and boundary distortion caused by traditional methods that rely solely on overall bitmap deformation. Thus, this step provides a stable data benchmark and an executable control blank for subsequent exposure correction and closed-loop compensation, making curing shrinkage compensation and forming accuracy control feasible for engineering implementation.

[0079] S3: During the printing process, the target exposure information is corrected based on the compensation model and the printing process parameters for the current layer to obtain the execution exposure information of the current layer, and the current layer is cured and formed accordingly.

[0080] The purpose of step S3 is to transform the target exposure information obtained in step S2 into execution exposure information that can achieve curing shrinkage compensation under the current working conditions, and to complete the curing and forming of the current layer accordingly. Unlike the conventional approach of directly exposing according to fixed exposure parameters, this step introduces a compensation model and printing process parameters during the layer-by-layer printing process, so that the exposure pattern and exposure parameters can be controllably adjusted with each layer, with each region, and with each working condition. This allows for feedforward suppression of dimensional deviations caused by non-uniform shrinkage during curing and provides convergent initial control quantities for subsequent deviation field feedback updates.

[0081] When correcting the target exposure information for the current layer, the control system preferably first determines the input set for this layer and then calls the compensation model to output the compensation amount accordingly. The input set includes the layer data of the current layer and its corresponding target exposure pattern, target exposure parameters, and printing process parameters such as layer thickness, exposure time, grayscale level and light intensity level, resin type and batch identifier, resin tank temperature and ambient temperature, and lift-and-fall cycle parameters related to peeling. Furthermore, to improve the targeting of the compensation, the control system can also introduce the regional attribute labels formed in step S2, such as near-boundary area, core area, support adjacent area, free boundary area, thin-walled area, and hole boundary area, enabling the compensation model to output differentiated compensation amounts for different areas. It should also be noted that the above inputs can all be directly obtained by the equipment or generated by the slicing process; therefore, this step is engineering-feasible.

[0082] Furthermore, the compensation amount output by the compensation model is used to generate the execution exposure information. This compensation amount includes a boundary offset for the exposure pattern contour, a correction for the local exposure dose, and a correction for the exposure cycle parameters. Specifically, when the compensation amount is a contour boundary offset, the control system can perform a geometric offset operation on the vector contour retained in the target exposure information to form an execution contour after expansion and contraction. The execution contour is then rasterized to generate the execution exposure pattern. For example, when the hole boundary region exhibits a stable decreasing aperture trend in historical printing, the compensation model can output an expansion offset in the aperture direction, making the holes in the execution exposure pattern slightly larger than the target holes, so that the aperture approaches the design value after curing and shrinkage. It should be noted that using a vector contour for offsetting and then rasterizing avoids the stepped jagged distortion and resolution loss caused by directly performing morphological processing such as dilation and erosion on the bitmap, thereby improving the geometric fidelity of the compensation execution.

[0083] When the compensation amount is a local exposure dose correction amount, the control system can set different gray levels and exposure times for different areas in the executed exposure pattern to achieve zoned dose control. For example, for thin-walled areas with free boundaries, the control system can reduce the local dose to decrease the peak curing rate and reduce the risk of stress lock-in and warping accumulation. For adjacent support areas and thick solid areas, the control system can appropriately increase the local dose to enhance early curing stiffness and improve resistance to peeling disturbances. It should also be noted that when the equipment adopts a binary exposure mode, dose correction can be achieved through fine-tuning of exposure time and accumulation of multiple short exposures, making the local dose still adjustable, thus allowing for implementation without relying on specific hardware.

[0084] When the compensation amount involves exposure cycle parameters, the control system can modify the interlayer waiting time, lift speed, fall speed, lift height, and segmented lift strategy during the execution of exposure parameters, enabling pattern compensation and process cycle to be controlled in a coordinated manner. For example, when the ambient temperature is low and the resin viscosity is high, resulting in slower leveling, the control system can appropriately increase the interlayer waiting time to ensure sufficient resin reflow and reduce localized material shortages and boundary defects caused by insufficient leveling. Similarly, when the release film condition leads to increased peel force, the control system can reduce the lift speed and adopt a segmented lift method to reduce the disturbance to thin-walled areas during peeling, thereby reducing cumulative interlayer deviation. It should be noted that the above-mentioned cycle parameters fall under the category of printing process parameters, and modifying them in this step enhances the robustness of the compensation strategy against process drift.

[0085] It should be noted that the obtained execution exposure information for the current layer refers to the final instruction set that can directly drive the printing equipment after the above corrections are completed. It includes the execution exposure pattern and the corresponding execution exposure parameters, and is used to replace the target exposure information for manufacturing this layer. Furthermore, to ensure process traceability and reliable subsequent closed-loop updates, the control system preferably associates and stores the execution exposure information of this layer with the compensation amount of this layer and the key process status information of this layer, including the actual exposure time, actual grayscale and light intensity level, resin tank temperature, actual lifting and falling cycle execution value, etc., and binds them to the layer number, providing data basis for the deviation field calculation in the subsequent step S4 and the model update in step S5.

[0086] It should also be noted that, when completing the current layer curing, the control system loads the exposure pattern onto the projection system and controls the light source output and platform movement according to the exposure parameters. Specifically, for DLP and MSLA devices, the control system maps the exposure pattern onto the projection pixel array and triggers exposure; for SLA scanning devices, the control system can convert the modified execution contour into a scanning path and determine the scanning speed and laser power in conjunction with the exposure parameters, so that the curing trajectory and energy input meet the requirements of the exposure information, thereby completing the curing of the current layer and bonding it to the lower layer.

[0087] Furthermore, step S3 applies the compensation amount output by the compensation model to the three executable dimensions of the target exposure information—contour, dosage, and beat—before each layer is formed, thus forming the execution exposure information. This elevates the compensation from traditional overall scaling to fine control oriented towards layers and regions, enabling feedforward suppression of dimensional deviations caused by non-uniform shrinkage while curing occurs. Simultaneously, by incorporating printing process parameters into the correction calculation and allowing coordinated adjustment of beat parameters, this step can respond to the drift caused by changes in temperature, resin state, and peeling conditions, reducing the tendency of error accumulation between layers and improving the stability and repeatability of forming accuracy control.

[0088] S4: Obtain the image data of the current layer after solidification, and calculate the deviation field of the current layer based on the image data of the current layer and the layer data.

[0089] The purpose of step S4 is to obtain, in a quantifiable manner, the difference between the actual and expected geometry of the current layer after it has cured, and to characterize this difference as a deviation field that can be directly utilized in subsequent step S5. Unlike overall measurement only after printing, this step obtains deviation information at the layer scale, allowing deviations to be identified and used for compensation updates of subsequent layers before they accumulate, thereby improving the convergence speed and stability of closed-loop control. The deviation field is preferably two-dimensional spatial distribution data consistent with the projection plane coordinates of the current layer, used to describe deviations related to curing shrinkage, such as outer contour boundary offset, aperture error, linewidth error, and local area scale changes.

[0090] The acquisition of in-layer image data after the current layer has cured can be accomplished by an imaging module installed in the printing device. The imaging module can be a top-view camera, a side-view camera, and a coaxial imaging component refracted by a reflector, preferably in conjunction with an illumination unit to improve image contrast and contour discernibility. Furthermore, the illumination unit can employ coaxial light and low-angle ring light to enhance the brightness gradient of the contour edges and suppress reflection interference from the liquid resin surface; the control system can automatically adjust the illumination intensity and camera exposure time, and perform white balance and background correction before each batch of printing to obtain stable and repeatable imaging quality. Furthermore, the imaging triggering timing can be set after the current layer exposure is complete and the platform has completed one lift-off, or after the platform has fallen back and the resin has re-spread and stabilized; preferably, the imaging triggering timing should be kept consistent within the same printing task to avoid introducing non-geometric factors due to liquid surface disturbance and reflection differences.

[0091] Furthermore, the imaging module preferably performs basic calibration and registration preparation, enabling image pixel coordinates to be mapped to projection coordinates and platform coordinates. For example, a reference mark area that does not participate in part forming can be set at the edge of the printing area. The reference mark can be a crosshair target, concentric circles, or a specific arrangement of dots. The actual position of the reference mark is synchronously acquired each time an image within a layer is acquired, and matched with the theoretical position of the reference mark stored in step S2, thereby calculating the coordinate mapping relationship between the image and the layered data, ensuring a consistent coordinate reference for subsequent deviation calculations. It should also be noted that, in cases where setting reference marks is inconvenient, the control system can also use an overall shape matching method of the outer contour to estimate translation and rotation deviations to complete registration, thus ensuring that the deviation field calculation does not rely on a single alignment method.

[0092] Furthermore, before calculating the deviation field, the control system preferably preprocesses the in-layer image data to reduce interference from factors such as uneven illumination, liquid surface reflection, resin residue, and motion blur on boundary extraction. Specifically, preprocessing may include noise reduction filtering, brightness normalization, background subtraction, and contrast enhancement processing between exposed and unexposed areas. When there are local highly reflective areas in the image, the control system can use multi-frame acquisition and select stable frames, and perform threshold clipping on bright areas to avoid false detections. When the resin surface is still flowing at the time of imaging, the control system can reduce motion blur by delaying acquisition and using a short exposure shutter speed, thereby improving the stability of boundary positioning.

[0093] It should be noted that the core of calculating the deviation field of the current layer based on the intra-layer image data and the layered data lies in aligning the actual forming geometry extracted from the intra-layer image with the expected geometric reference represented by the layered data generated in step S2, and calculating the difference in the same coordinate system after alignment. Specifically, after registration, the control system extracts the actual contour and feature information from the intra-layer image. The feature information includes the outer contour boundary, hole boundary, key size boundary, and edge lines within the preset detection area. Furthermore, to improve the robustness of contour extraction, the control system can combine threshold segmentation and gradient detection to determine the boundary position. For example, when the solidified boundary has a solidified halo due to light scattering, the image may show a low-contrast diffusion area outside the boundary. In this case, the control system can take the center of the high gradient position as the true contour position to remove the influence of scattered light and reduce false outward expansion error. For micro-holes and fine lines with near pixel resolution, the control system can also use a sub-pixel edge positioning method to refine the boundary position to improve the measurement accuracy of aperture and line width.

[0094] Furthermore, the calculation of the deviation field can be performed separately according to feature type. Specifically, for contour boundary deviation, the control system can sample boundary points at preset intervals on the expected contour and search for the corresponding actual edge position in the image in the normal direction of each boundary point to obtain the normal offset of that point, thereby forming a boundary offset deviation distributed along the boundary; for the size deviation of holes and local features, the control system can calculate the difference between the actual aperture and the expected aperture and map it to the neighborhood of the hole boundary to form a local deviation distribution; for regional scale variation deviation, the control system can perform regional overlap analysis on the actual exposure area and the expected exposure area in the image to obtain the local area difference and local shape difference, and convert them into a deviation distribution on the projection plane. Furthermore, to improve the usable correlation between the deviation field and the compensation amount, the control system can aggregate and statistically analyze the deviation field according to the regional attribute labels in step S2, for example, calculating the average boundary offset and average size error of the near boundary area, core area, support adjacent area, thin wall area and hole boundary area respectively, so that the subsequent compensation model update obtains a more stable error signal.

[0095] To avoid compensation oscillations caused by measurement noise, this embodiment preferably performs effectiveness screening and stabilization processing on the deviation field. Specifically, the control system can set a reliable threshold rule. For example, when the deviation amplitude is less than the minimum resolvable size corresponding to the imaging resolution, it is regarded as measurement noise and no model update is triggered. When the deviation in a certain local area is abnormally large and accompanied by abnormal image quality, it is marked as an abnormal layer and a conservative update strategy is adopted. The control system can also perform moving average and weighted fusion on the deviation fields of several adjacent layers to reduce the impact of occasional disturbances in a single layer on the model update, thereby improving the stability and repeatability of closed-loop control.

[0096] Furthermore, step S4 acquires and quantifies the difference between the actual forming geometry and the expected layering reference at the hierarchical scale, forming a deviation field that can be used to generate compensation amounts and update the compensation model. This allows the closed-loop control of the present invention to no longer rely on overall measurements after printing, thereby enabling earlier identification and suppression of error accumulation caused by curing shrinkage, peeling disturbances, and operating condition drift. Simultaneously, through reference mark registration, robust contour extraction, and deviation stabilization processing, this step can still output a reliable deviation signal in the actual environment of resin surface reflection and noise, providing a stable data basis for the adaptive update in the subsequent step S5, thereby improving the stability and repeatability of forming accuracy control.

[0097] S5: Update the compensation model according to the deviation field and generate the compensation amount of the next layer. Based on the compensation amount, correct the target exposure information of the next layer and repeat the process until the curing of all layers is completed, thereby obtaining a three-dimensional molded part with curing shrinkage compensation.

[0098] The purpose of step S5 is to transform the deviation field obtained in step S4 into an adaptive correction basis for the compensation model, and based on this, generate the next layer of directly executable compensation amount, so that the compensation strategy can continuously converge as the working conditions drift and error accumulation trend during the printing process. Unlike the approach that relies solely on initial calibration and fixed empirical parameters, this step uses a recursive mechanism of "deviation field - model update - next layer compensation amount" to make the compensation model conform to the current resin state, temperature conditions, peeling conditions and optical state layer by layer within the same printing task, thereby improving the stability and repeatability of forming accuracy control, and obtaining a three-dimensional formed part with curing shrinkage compensation after all layers have been cured and formed.

[0099] The process of updating the compensation model based on the deviation field preferably includes extracting usable error signals from the deviation field, constructing an error metric for updating and determining the adjustment direction of model parameters accordingly, executing the update, and managing the version. Specifically, the control system can first perform regional summarization and feature representation of the deviation field to reduce the impact of measurement noise on the update and improve the update's targeting. For example, the average boundary offset, aperture deviation, and linewidth deviation of the near-boundary region, hole boundary region, thin-walled free boundary region, support adjacent region, and core region can be calculated separately, and the consistency of deviation direction and the number of continuous layers can be recorded to determine whether the deviation is a random disturbance or a systematic drift. Furthermore, when the deviation of the same feature region shows a stable trend with the same direction and similar amplitude within several consecutive layers, the control system can determine it as a compensable systematic deviation and trigger a model update; when the deviation shows unstable fluctuations and is associated with abnormal image quality, the control system can adopt a conservative update strategy and postpone the update to avoid compensation oscillations.

[0100] Furthermore, the error metric is used to guide the updates of the compensation model towards reducing the system bias. It can be constructed based on the degree of deviation between the compensation model output and the bias field, enabling the adjustment of model parameters to reduce the expected bias of the next layer. For example, when the bias field indicates a persistent contour shrinkage bias in a certain region, the error metric will guide the compensation model to increase the outward contour bias of that region in the next prediction, and simultaneously adjust the dose and beat corrections for that region, so that the actual contour of subsequent layers reverts to the design baseline. When the bias field indicates a general shrinkage of the pore boundaries, the error metric will guide the compensation model to increase the outward bias of the pore boundaries, thereby reducing the tendency for the pore size to decrease with cumulative layer growth. It should be noted that the above guiding relationship can be achieved through rule weight adjustment, parameter recursive correction, and model fine-tuning, without requiring precise inversion of the material mechanism.

[0101] The update method of the compensation model can be set according to the implementation form of the compensation model, and preferably adopts a lightweight update mechanism that can be executed in real time within the printing cycle. It should be noted that the lightweight update mechanism aims to ensure that the model update calculation can be completed within a finite time interval between two printing layers, thereby not interrupting the continuous printing process and maintaining the established forming cycle. Furthermore, when the compensation model is in the form of a lookup table and rule base, the update can be manifested as incremental correction of the corresponding entries, such as adding or subtracting adjustments to the boundary bias, dosage correction, and cycle correction amounts indexed by "resin type—layer thickness range—region label," and setting an upper limit for the update step size and a trigger threshold to avoid overcompensation. Furthermore, when the compensation model is in the form of a parametric mapping, the update can be manifested as recursive correction of the model parameters, such as adjusting the boundary bias coefficient, dosage coefficient, and cycle correction coefficient with small steps, so that the model output tends to offset the systematic errors reflected in the deviation locations in the next layer. It should also be noted that when the compensation model is a data-driven model, the update can adopt online fine-tuning and incremental learning: the control system forms an online sample with the input set of the current layer and the deviation field, and uses the error metric as the update signal to fine-tune the model parameters in a limited small step size; in order to avoid model drift leading to instability, online updates are preferably limited to relevant parameters at the output end and a small number of controllable parameters, and a model version rollback mechanism is set to restore to the previous stable version when abnormal deviations occur.

[0102] Furthermore, after updating the compensation model, the control system generates the compensation amount for the next layer based on the updated compensation model. The compensation amount preferably matches the executable control dimension used in step S3, including contour boundary offset, local dose correction, and cycle time parameter correction. For example, when the deviation field shows that the hole boundaries are generally shrinking inward, the compensation model outputs a larger outward offset for the hole boundaries in the next layer; when the deviation field shows abnormal boundary offset in the free boundary thin-walled region accompanied by peeling disturbance characteristics, the compensation model can output a correction amount that reduces the dose and increases the interlayer waiting time for that region in the next layer to reduce the peak curing rate and improve leveling, thereby suppressing further error propagation. Furthermore, to improve the stability and safety of the compensation amount application, the control system can set constraint rules for the compensation amount, including the maximum offset amplitude, the maximum dose adjustment range, and the maximum cycle adjustment range, and can implement layer-by-layer gradual restriction on the compensation amount according to the layer number; for example, in the transition layer between the abrupt cross-section layer and the thin-walled to thick solid, the control system can limit the rate of change of the compensation amount to avoid introducing new boundary distortion and local under-curing risks due to excessive single-layer compensation.

[0103] It should be noted that the step of correcting the target exposure information of the next layer based on the compensation amount and executing it cyclically means that the control system uses the compensation amount generated in step S5 as the input to step S3, so that the target exposure information of the next layer is corrected to the execution exposure information and solidified. Then, the image within the layer is acquired again and the deviation field is calculated. Subsequently, the compensation model is updated and the compensation amount of the next layer is generated, thus forming a closed-loop iterative process that runs through all layers. Furthermore, the control system preferably establishes a closed-loop log for each layer, and stores the target exposure information, execution exposure information, compensation amount, deviation field summary and model version number of that layer in association to achieve process traceability and anomaly rollback. When the deviation of a certain layer exceeds the confidence threshold and the image quality is abnormal, the control system can mark the layer as a sample that does not participate in the update and continue to execute using the compensation amount generated by the previous stable model version, thereby ensuring system stability.

[0104] Furthermore, step S5 transforms the deviation field into an error signal that can be used to update the compensation model, and applies the update result as a compensation amount to the next layer's exposure correction in real time. This allows the compensation model to adapt layer by layer to time-varying factors such as resin batch variations, temperature fluctuations, release condition changes, and optical state drift within the same printing task, suppressing the accumulation trend of errors between layers. Simultaneously, through mechanisms such as regional aggregation, trigger thresholds, conservative updates, amplitude constraints, and version rollback, this step maintains the stability of closed-loop updates in real-world environments with measurement noise and occasional disturbances, making the compensation process exhibit convergence characteristics. In summary, step S5 transforms the precision control of 3D printing from a mode relying on pre-process offline, static calibration to an online, dynamic, self-optimizing process throughout the entire manufacturing process. This reduces the reliance on repeated manual parameter testing and relaxes the requirements for environmental constancy and material consistency, thus providing crucial support for the application of photopolymerization technology in stable manufacturing scenarios for high-precision parts.

[0105] Example 2 is the second embodiment of the present invention, which differs from the previous embodiment in that:

[0106] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art or the current technical solution, can be embodied in the form of a software product. This current computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0107] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0108] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0109] Example 3, referring to Figure 2 As an embodiment of the present invention, a curing shrinkage compensation and forming accuracy control system for 3D printing is provided, which includes a data acquisition and compensation model preparation module, a layer and target exposure information generation module, a layer-by-layer exposure correction and curing forming execution module, an intra-layer image acquisition and deviation field calculation module, and a compensation model update and next layer compensation amount generation module.

[0110] Data acquisition and compensation model preparation module: acquire the three-dimensional model data of the part to be formed and the corresponding printing process parameters, and establish and call the compensation model to characterize the relationship between curing shrinkage forming deviation and compensation amount;

[0111] Layering and Target Exposure Information Generation Module: Generates layered data based on the 3D model data, and generates target exposure information for each layer;

[0112] Layer-by-layer exposure correction and curing execution module: During the printing process, the target exposure information of the current layer is corrected based on the compensation model and the printing process parameters to obtain the execution exposure information of the current layer, and the curing of the current layer is completed accordingly;

[0113] Intra-layer image acquisition and deviation field calculation module: acquires intra-layer image data after solidification of the current layer, and calculates the deviation field of the current layer based on the intra-layer image data and the layer data;

[0114] Compensation model update and next layer compensation amount generation module: Update the compensation model according to the deviation field and generate the compensation amount of the next layer. Based on the compensation amount, correct the target exposure information of the next layer and repeat the process until the curing and forming of all layers is completed, thereby obtaining a three-dimensional formed part with curing shrinkage compensation.

[0115] Example 4 is an embodiment of the present invention, which provides a method for curing shrinkage compensation and forming accuracy control in 3D printing. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation / comparative experiments.

[0116] This embodiment uses a 405nm DLP photopolymerization printer (projection resolution 3840×2160, pixel size approximately 50μm) to comparatively print seven typical structural parts under constant temperature conditions: stepped perforated plate calibration parts, thin-walled cantilever structures, precision threaded joints, microporous screens, fine-module gears, internal cavity flow channel manifolds, and multi-pin connector seats. Each type of part is mounted on the same platform and uses the same resin batch, with consistent basic process windows (layer thickness 35–50μm, exposure time approximately 2.18–2.72s, light intensity level approximately 75.6%–80.2%, and lift-off speed approximately 4.3–5.4mm / s). Two strategies are implemented: one is a conventional "overall scaling + fixed exposure / beat" comparison strategy (scaling factor set empirically); the other is the closed-loop strategy of the present invention. Under the closed-loop strategy, firstly, 3D model files such as STL are imported and geometric consistency processing is performed (closure check, hole / damage repair, normal unification, and nearest neighbor vertex merging). Then, the model is unified to the printer coordinate system and the placement posture is recorded. Further, the model is pre-analyzed to extract geometric description information related to shrinkage sensitivity (e.g., thin-walled / thick solid classification, free boundary / support adjacency zone labels, hole boundary and feature boundary sets, cross-sectional abrupt layer markings, etc.). A compensation model is then established and invoked as the initial mapping relationship, combined with printing process parameters. Subsequently, slices are generated based on layer thickness, generating layered data and target exposure information for each layer. In addition to exposure patterns and parameters, the target exposure information also includes structured encapsulation of vector contour representation, partition attribute labels, and image registration reference information. During the printing process, a compensation model is invoked before each layer exposure. This model integrates the current layer's layer data, geometric description information, and real-time / task-level process parameters to calculate compensation amounts such as contour boundary offset, local dose correction, and cycle time correction. This generates the layer's exposure information and completes curing. After curing, the in-machine imaging module acquires image data within the layer, performs coordinate registration using reference information, extracts the actual contour and feature boundaries, and compares them with the layer benchmark to calculate the deviation field. After confidence evaluation and stabilization, the deviation field forms usable deviation information, used for incremental updates to the compensation model and generating the compensation amount for the next layer. To ensure online cycle time continuity, update calculations are limited to the interval between two layers. When imaging quality anomalies and deviation anomalies are detected, anomaly filtering is triggered, and the model can be rolled back to the previous stable model version. After the part is printed and cleaned, a tool microscope and coordinate measuring machine are used to verify key dimensions, aperture / equivalent aperture, and warpage. The comparison results shown in the table below are statistically analyzed (the terms "traditional" and "invention" in the table refer to the statistical values ​​of the same part under the two strategies).

[0117] The key data recorded is shown in Table 1:

[0118] Table 1: Experimental Data Recording Table

[0119]

[0120] As can be seen from the mean absolute error (MAE) of critical dimensions, the conventional "overall scaling + fixed exposure / beat" comparison strategy exhibits obvious non-uniform error characteristics on parts with different geometric categories, and the error amplitude increases significantly with thin walls, small features, and internal cavity structures. For example, the conventional MAE for microporous screens and flow channel manifolds reaches 95.4 μm and 104.8 μm, respectively, while that for connectors and thin-walled cantilever structures is 89.6 μm and 83.1 μm, respectively. This reflects that the conventional method is difficult to simultaneously account for multiple sources of error such as "hole boundary shrinkage," "free boundary shrinkage / warping," and "stress locking in abrupt cross-section layers." In comparison, the closed-loop method of this invention reduces the MAE of the aforementioned parts to 31.2 μm, 36.7 μm, 29.8 μm, and 27.6 μm, respectively, under the same equipment and material window. For structures with boundary accuracy as the primary concern, such as stepped hole plates, gears, and threaded joints, the MAE is also reduced from 62.7 μm, 58.3 μm, and 71.8 μm to 18.9 μm, 17.4 μm, and 22.1 μm. This result indicates that the layer-by-layer update mechanism driven by the deviation field can suppress the interlayer accumulation of errors and enable the compensation amount to adaptively converge with the structure and operating conditions, rather than relying on a single empirical scaling factor.

[0121] Further observation of the maximum boundary offset index revealed that traditional strategies exhibited a boundary drift phenomenon of "coexistence of local over-curing and local under-curing" on most parts, with the maximum boundary offset fluctuating within the range of 126.6–221.4 μm. Particularly, the offset reached 206.7 μm and 221.4 μm for microporous screens and internal manifolds, respectively, indicating that boundary control is more prone to instability near small features and complex cavities. The method of this invention reduces the corresponding index to the range of 41.8–93.5 μm, with the stepped perforated plate and gears showing offsets of only 44.7 μm and 41.8 μm, respectively. This demonstrates that the executable control dimension of "contour boundary offset" can directly affect the boundary normal error, avoiding the boundary distortion caused by traditional coarse-grained processing using bitmap morphology and overall scaling. Meanwhile, the aperture / equivalent aperture error data further corroborates this point: for orifice plates, threaded joints, microporous screens, gear holes, manifold interfaces, and connector hole groups, the traditional error ranges from 47.6 to 88.9 μm, while the present invention reduces it to the range of 14.2 to 28.6 μm, demonstrating that the model has a stronger directional compensation capability in the highly sensitive area of ​​"hole boundaries"; although the aperture error of the manifold is reduced to 28.6 μm, it is still higher than that of the gear / orifice plate, which is consistent with the engineering fact that the internal cavity structure is affected by both light scattering and peeling disturbance, and the error mechanism is more complex.

[0122] Regarding shape stability, the peak-to-peak warpage is a typical weakness of traditional methods, reaching 268.6 μm for thin-walled cantilever and 312.7 μm for internal manifold, indicating that fixed exposure and overall scaling alone cannot manage the amplification effect of curing stress and peeling disturbance on the free boundary of thin walls. The method of this invention reduces these peaks to 112.4 μm and 146.5 μm, respectively, a significant reduction. This is because the compensation in this invention does not merely adjust the geometric boundary, but simultaneously introduces two types of synergistic control measures: "local dose correction" and "exposure beat parameter correction." This reduces the peak curing rate in the thin-walled region of the free boundary, maintains necessary stiffness in the adjacent support region and thick solid region, and improves leveling and peeling load through beat fine-tuning, thereby suppressing the conditions for warpage formation at the process level. This three-dimensional coupled compensation of "geometry-energy-beat" provides a control granularity and path that traditional overall scaling strategies cannot achieve.

[0123] From the perspective of online feasibility and robustness, the additional inter-layer computation time of this invention is 52–76 ms, and the total printing time increases by only 0.8%–1.7%, indicating that model inference, deviation field calculation, and incremental updates can be completed within the interval between two layer cycles without disrupting printing continuity. Simultaneously, the number of abnormal layer filtering triggers is 0–2 times, and the number of model version rollbacks is 0–1 times, demonstrating that in the presence of actual imaging noise and occasional disturbances, the confidence assessment / stabilization and version management mechanisms can effectively block the "erroneous injection" of abnormal data into model updates, ensuring closed-loop convergence rather than oscillation. In summary, the comparative results of this embodiment objectively demonstrate that compared to existing compensation methods that rely primarily on empirical scaling and fixed parameters, this invention, through a recursive closed loop of "intra-layer imaging deviation field—model update—next layer compensation amount," elevates forming accuracy control from static calibration to an online adaptive optimization process, achieving significant and reproducible comprehensive improvements in key dimensions, boundary offset, aperture preservation, and warpage suppression.

[0124] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications and equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications and substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for compensating for curing shrinkage and controlling forming accuracy in 3D printing, characterized in that, include: Obtain the 3D model data of the part to be formed and the corresponding printing process parameters, and establish and call a compensation model to characterize the relationship between curing shrinkage forming deviation and compensation amount; Based on the 3D model data, layered data is generated, and target exposure information for each layer is generated; During the printing process, the target exposure information is corrected based on the compensation model and the printing process parameters for the current layer to obtain the execution exposure information for the current layer, and the current layer is cured and formed accordingly. Acquire the solidified intralayer image data of the current layer, and calculate the deviation field of the current layer based on the intralayer image data and the layer data; The compensation model is updated according to the deviation field and the compensation amount of the next layer is generated. The target exposure information of the next layer is corrected based on the compensation amount and the process is repeated until the curing of all layers is completed, thereby obtaining a three-dimensional molded part with curing shrinkage compensation. The step of updating the compensation model based on the deviation field and generating the compensation amount for the next layer, and correcting the target exposure information of the next layer based on the compensation amount and performing the process iteratively includes: The deviation field is regionalized and characterized to obtain error information for characterizing systematic deviations. The regionalization and characteristic representation includes statistical processing of contour boundary offset, feature size error and regional scale change according to preset region type, geometric category and attribute label. The update basis for the compensation model is constructed based on the error information, so that the update direction of the compensation model can reduce the expected deviation of the subsequent printing layer, and the incremental update of the compensation model is performed when the preset update trigger condition is met. The preset update trigger condition includes the error amplitude threshold and the error direction consistency condition. Based on the updated compensation model, combined with the layer data and geometric description information corresponding to the next printing layer and the printing process parameters, the compensation amount of the next printing layer is calculated. The compensation amount includes the contour boundary offset amount, the local exposure dose correction amount, and the exposure cycle parameter correction amount. After applying amplitude and rate of change constraints to the compensation amount, the compensation amount is used to correct the target exposure information of the next printing layer to generate the execution exposure information of the next printing layer, and the process is repeated until all layers are cured and formed. Furthermore, version management is implemented for the incremental update process of the compensation model, so that when deviation anomalies and imaging quality anomalies are detected, it can revert to the previous stable version of the compensation model and continue to execute using the previous stable compensation amount.

2. The method for compensating for curing shrinkage and controlling forming accuracy in 3D printing as described in claim 1, characterized in that, The process of obtaining the 3D model data of the part to be formed and the corresponding printing process parameters includes: The three-dimensional model data is subjected to geometric consistency processing and coordinate unification processing. The geometric consistency processing includes closure checks and repair of holes and broken surfaces. The coordinate unification processing includes unifying the three-dimensional model data to the printer coordinate system and recording its placement posture on the forming platform. Obtain the printing process parameters corresponding to the current printing task. The printing process parameters include energy input parameters and forming cycle parameters that affect curing shrinkage, and further include material and environmental parameters that characterize the current working conditions. The printing process parameters are obtained by reading from the equipment control interface, equipment operation log, integrated sensors and material management module, and by user input, and are bound to the current printing task for later retrieval.

3. The method for compensating for curing shrinkage and controlling forming accuracy in 3D printing as described in claim 2, characterized in that, The establishment and invocation of the compensation model used to characterize the relationship between curing shrinkage forming deviation and compensation amount includes: Geometric description information related to curing shrinkage sensitivity is extracted based on the three-dimensional model data; A compensation model is constructed and invoked, which is configured to output an exposure compensation amount for correcting the target exposure information based on the layer data of the current printing layer, the geometric description information, and the printing process parameters when performing subsequent calculations for the current printing layer. The exposure compensation amount includes contour boundary offset amount, local exposure dose correction amount, and exposure cycle parameter correction amount; The compensation model is implemented in the form of lookup tables and rule bases, parameterized mapping models, and data-driven models. Furthermore, the compensation model is configured to update its internal parameters and rules based on the deviation field.

4. The method for compensating for curing shrinkage and controlling forming accuracy in 3D printing as described in claim 3, characterized in that, The step of generating layered data based on the three-dimensional model data includes: Based on the layer thickness in the printing process parameters, the layer spacing is set along the printing construction direction, and multiple sets of mutually parallel cross-sectional planes are established; Intersection operations are performed on each of the said cross-sectional planes and the three-dimensional model data to obtain one or more closed two-dimensional contours corresponding to each layer, and when there are holes and cavities, the outer contour and the inner contour are organized into a set of layered contours of the same layer. Perform contour consistency processing on the hierarchical contour set. The contour consistency processing includes contour smoothing and denoising, removal of isolated islands, merging of adjacent contours, and topology consistency verification. Furthermore, when the contour size is close to the projected pixel resolution, the corresponding contour is subjected to connectivity repair and boundary reconstruction based on grid mask to form stable and well-formed layered data.

5. The method for compensating for curing shrinkage and controlling forming accuracy in 3D printing as described in claim 4, characterized in that, The generated target exposure information for each layer includes: Based on the resolution and pixel size of the projection system, the set of layered contours of each layer is transformed into the projection coordinate system and rasterized to generate the exposure pattern of the corresponding layer. The area to be formed is marked as the exposure area, and the hole and cavity areas are marked as the non-exposure area. Based on the printing process parameters, each layer is configured with target exposure parameters corresponding to the exposure pattern. The target exposure parameters include exposure time, gray level, and light intensity level. Anti-aliasing is performed on the outline boundary of the exposure pattern, the anti-aliasing including boundary grayscale transition and subpixel boundary reconstruction; Furthermore, the target exposure information is structured and encapsulated, including the layer's exposure pattern, target exposure parameters, contour data representation for subsequent exposure correction, attribute labels for exposure area partitioning control, and reference information for coordinate registration with the image data within the layer.

6. The method for compensating for curing shrinkage and controlling forming accuracy in 3D printing as described in claim 5, characterized in that, The step of obtaining the current layer's exposure information and completing the current layer's curing process accordingly includes: The compensation model is invoked, and based on the printing process parameters and the layer data and geometric description information corresponding to the current layer, the exposure compensation amount used to correct the target exposure information is calculated. The exposure compensation amount includes contour boundary offset amount, local exposure dose correction amount, and exposure cycle parameter correction amount; The target exposure information of the current layer is corrected according to the exposure compensation amount to obtain the execution exposure information of the current layer. The correction includes the correction of the exposure pattern and exposure parameters. The printing equipment is controlled to complete the curing and forming of the current layer based on the execution exposure information, wherein the execution exposure information includes the execution exposure pattern and its corresponding execution exposure parameters; Furthermore, the execution exposure information of the current layer is associated with and stored along with the exposure compensation amount and the key operating condition information corresponding to the current layer, so as to be used for deviation field calculation and compensation model update of subsequent layers.

7. The method for compensating for curing shrinkage and controlling forming accuracy in 3D printing as described in claim 6, characterized in that, The calculated deviation field for the current layer includes: The imaging module installed in the printing device acquires the in-layer image data after the current layer has been cured. Based on preset and associated reference information, the image data within the layer is coordinate registered with the layered data of the corresponding layer. The reference information includes reference markers, alignment marker areas, and the calibration relationship between the projected coordinate system and the platform coordinate system. The actual shaped contour and features are extracted from the in-layer image data after coordinate registration, and compared with the expected contour and expected features in the layered data to calculate the deviation field. The deviation field is used to characterize contour boundary offset, feature size error, and region scale variation; Furthermore, the deviation field is subjected to confidence assessment and stabilization processing to form deviation information for updating the compensation model.

8. A 3D printing curing shrinkage compensation and forming accuracy control device, used to implement the 3D printing curing shrinkage compensation and forming accuracy control method as described in any one of claims 1 to 7, characterized in that, include: Data acquisition and compensation model preparation module: acquire the three-dimensional model data of the part to be formed and the corresponding printing process parameters, and establish and call the compensation model to characterize the relationship between curing shrinkage forming deviation and compensation amount; Layering and Target Exposure Information Generation Module: Generates layered data based on the 3D model data, and generates target exposure information for each layer; Layer-by-layer exposure correction and curing execution module: During the printing process, the target exposure information of the current layer is corrected based on the compensation model and the printing process parameters to obtain the execution exposure information of the current layer, and the curing of the current layer is completed accordingly; Intra-layer image acquisition and deviation field calculation module: acquires intra-layer image data after solidification of the current layer, and calculates the deviation field of the current layer based on the intra-layer image data and the layer data; Compensation model update and next layer compensation amount generation module: Update the compensation model according to the deviation field and generate the compensation amount of the next layer. Based on the compensation amount, correct the target exposure information of the next layer and repeat the process until the curing and forming of all layers is completed, thereby obtaining a three-dimensional formed part with curing shrinkage compensation.

Images