A continuous fiber pre-impregnation system, a multi-material proportioning controllable 3D printing method and system

Abstract

The present application belongs to the technical field of composite material additive manufacturing, and discloses a continuous fiber prepreg system, a 3D printing method and system with adjustable multi-material proportioning, comprising resin and fiber material conveying modules, a multi-chamber composite material impregnation module and an extrusion control and detection module connected in sequence, wherein the multi-chamber composite material impregnation module is provided with a mixed resin chamber and an impregnation chamber in communication, and the mixed resin chamber and the impregnation chamber are distributed in a ladder type in series from top to bottom; after multi-component resin is uniformly mixed in the mixed resin chamber, the resin enters the impregnation chamber to be mixed with the continuous fiber to obtain the prepreg tow. The present application improves impregnation quality, mixing uniformity and efficiency.

Description

Technical Field

[0001] This invention belongs to the technical field of additive manufacturing of composite materials, and more specifically, relates to a continuous fiber prepreg system and a 3D printing method and system with adjustable multi-material ratio. Background Technology

[0002] Continuous fiber reinforced composites, due to their extremely high specific strength, specific stiffness, and excellent fatigue resistance, have shown great application potential in aerospace, automotive lightweighting, robotics, and high-end medical devices. In recent years, combining continuous fiber composites with 3D printing technology has enabled the moldless, rapid, and free-form production of complex composite structural components, becoming a research hotspot in advanced manufacturing.

[0003] However, despite the progress made in existing continuous fiber 3D printing technology, it still faces significant technical bottlenecks and shortcomings in practical engineering applications. Currently, most mainstream continuous fiber 3D printing methods utilize pre-prepared commercial filaments or simple composites of resin and fiber on the print head. In these technologies, the fiber volume fraction and resin matrix type remain constant throughout the printing process. The resulting components exhibit homogeneous material properties across all regions, making it impossible to continuously adjust local material strength based on the actual stress conditions. Furthermore, under actual service conditions, the stress distribution within three-dimensional structural components is often highly uneven, exhibiting distinct stress concentration and low-stress areas. Existing homogeneous printing methods, when dealing with complex stressed components, typically require increasing the overall fiber content or using high-performance resins to meet the strength requirements of the highest stress areas. This not only results in significant waste of expensive materials such as carbon fiber and high-performance specialty resins but also leads to excessive overall component weight, contradicting the principles of lightweight design. Conversely, prioritizing weight reduction may result in insufficient strength in stress concentration areas, leading to failure.

[0004] While some existing technologies attempt to improve structural strength by altering fiber arrangement paths through topology optimization, this only optimizes the spatial path and does not intervene at the intrinsic material composition level. Currently, the market lacks a technological architecture that deeply couples the pre-processing software results with the underlying hardware of the printer extrusion and prepreg system. It also lacks online prepreg hardware devices capable of dynamically and continuously changing resin types or resin / fiber ratios in real-time during printing based on algorithmic instructions. Summary of the Invention

[0005] In view of the above-mentioned defects or improvement needs of the existing technology, the present invention provides a continuous fiber prepreg system, a 3D printing method and system with adjustable multi-material ratio, which aims to solve the problem of how to improve the uniformity and efficiency of mixing multi-component resin and fiber, and optimize the impregnation process structure.

[0006] To achieve the above objectives, according to one aspect of the present invention, a continuous fiber prepreg system is provided. The continuous fiber prepreg system includes a resin and fiber material conveying module, a multi-chamber composite material impregnation module, and an extrusion control and detection module connected in sequence. The resin and fiber material conveying module provides a predetermined flow rate of multi-component resin and continuous fibers with constant tension to the multi-chamber composite material impregnation module. The multi-chamber composite material impregnation module mixes the multi-component resin and then mixes it with the continuous fibers to obtain a prepreg tow, and transmits the prepreg tow to the extrusion control and detection module. The extrusion control and detection module controls the extrusion of the prepreg tow. The multi-chamber composite material impregnation module is provided with a connected mixing resin chamber and an impregnation chamber, which are arranged in a stepped series configuration. After the multi-component resin is uniformly mixed in the mixing resin chamber, it enters the impregnation chamber and is mixed with the continuous fibers to obtain the pre-impregnated fiber bundle.

[0007] Furthermore, the resin and fiber material conveying module includes a continuous fiber bundle feeding unit and a multi-source resin independent servo feeding unit connected in parallel and both connected to the multi-chamber composite material impregnation module; the continuous fiber bundle feeding unit is used to provide continuous fibers with constant tension to the multi-chamber composite material impregnation module; the multi-source resin independent servo feeding unit is used to provide multi-component resin with a predetermined flow rate to the multi-chamber composite material impregnation module.

[0008] Furthermore, the multi-source resin independent servo feeding unit is provided with at least two independent resin feeding and melting channels in parallel, and a third anti-backflow check valve is provided at the junction of the end of each resin feeding and melting channel and the multi-chamber composite material impregnation module; the resin feeding and melting channel includes a resin extrusion servo motor, a micro screw extruder and a third anti-backflow check valve connected in sequence.

[0009] Furthermore, the multi-chamber composite material impregnation module includes a stepped body, a first anti-backflow one-way valve, and a second anti-backflow one-way valve. The mixing resin chamber and the impregnation chamber are respectively opened in the body and are located at opposite ends of the body. The body has a first resin inlet and a second resin inlet, both of which are connected to the mixing resin chamber. The multi-component resins enter the mixing resin chamber through the first resin inlet and the second resin inlet, respectively. The first resin inlet and the second resin inlet are respectively equipped with the first anti-backflow one-way valve and the second anti-backflow one-way valve.

[0010] Furthermore, the extrusion control and inspection module includes a temperature sensor, a pressure sensor, and a tapered wear-resistant nozzle. The temperature sensor and the pressure sensor are respectively disposed on the tapered wear-resistant nozzle, which is connected to the multi-chamber composite material impregnation module.

[0011] The present invention also provides a 3D printing system with adjustable multi-material ratios. The 3D printing system includes a prepreg system, a printing actuator, and a main controller as described above. The main controller is used to control the prepreg system and the printing actuator. The prepreg system is connected to the printing actuator.

[0012] This invention also provides a 3D printing method with adjustable multi-material ratios. The 3D printing method is implemented using the 3D printing system with adjustable multi-material ratios as described above, and includes the following steps: (1) Based on the boundary conditions and load conditions given by the actual service conditions of the structural component to be printed, the finite element analysis method is used to simulate the three-dimensional stress field of the three-dimensional structural model of the structural component to be printed, extract the stress tensor and principal stress distribution characteristics inside the structural component, and then combine the spatial coordinates of the nodes and their corresponding stress values ​​into data pairs. (2) Using data pairs as input and the optimal material ratio of fiber to different resins as output, a data-driven neural network ratio mapping model is constructed. Then, the material ratio data output by the neural network ratio mapping model, the three-dimensional slice model of the structure to be printed, and the printing path planning are collaboratively planned to compile and generate control G code. (3) Input the generated G code into the 3D printing system to achieve spatiotemporal synchronization and closed-loop control between the continuous fiber prepreg system and the printing actuator, and complete the 3D printing of the structural parts to be printed.

[0013] Furthermore, the neural network proportioning mapping model is a multi-objective optimization model based on NSGA-II. During its iterative solution process, a neural network surrogate model is used to calculate the objective function value of the neural network proportioning mapping model under the corresponding material proportion. The neural network surrogate model is a feedforward neural network based on a backpropagation neural network or a deep multilayer perceptron with residual connections, and its input is a data pair.

[0014] Furthermore, the neural network matching mapping model is defined as follows: Decision variable: Material proportion vector ,in This represents the volume fraction of continuous fibers. The percentage of resin A is given by [formula], and the percentage of resin B is given by [formula]. Sure; The decision space constraint is: ; Objective function: Maximize local mechanical properties Minimize material weight ;in, and The output of the trained neural network proxy model.

[0015] Furthermore, spatial continuity compensation is performed on the material ratio data of local nodes output by the neural network ratio mapping model. Specifically, based on the comparison between the distance between the node and its neighboring points and a set threshold, if the distance is less than the set threshold, the material ratio corresponding to the node is compensated.

[0016] In summary, compared with the prior art, the continuous fiber prepreg system, multi-material ratio adjustable 3D printing method and system provided by the present invention have the following advantages: 1. The multi-chamber composite material impregnation module is provided with a connected mixing resin chamber and an impregnation chamber, which are arranged in a stepped, vertically connected series. After the multi-component resin is uniformly mixed in the mixing resin chamber, it enters the impregnation chamber and mixes with the continuous fibers to obtain the pre-impregnated fiber bundle. This facilitates the full pre-mixing of the resin before it comes into contact with the fibers, thereby improving the impregnation quality, mixing uniformity, and efficiency.

[0017] 2. Using data pairs as input and the optimal material ratio of fibers to different resins as output, a data-driven neural network-based material ratio mapping model is constructed. Leveraging the powerful nonlinear mapping capability of neural networks, adaptive prediction of multiple material ratios under arbitrarily complex stress states is achieved. This significantly reduces the amount of high-performance resin used while ensuring structural safety, truly realizing the adaptive stress distribution of material components in three-dimensional space.

[0018] 3. By collaboratively planning the material ratio data output by the neural network ratio mapping model, the 3D slice model of the structural component to be printed, and the printing path plan, control G-code is generated and printed. This enables real-time and continuous adjustment of the mixing ratio of different types of resins and the volume fraction of resin and fiber during continuous fiber prepreg. This breaks the limitation of fixed types and ratios of prepreg resins in existing technologies, and truly realizes the on-demand formulation of continuous fiber prepregs with gradient mechanical properties at the printing front end. This provides a flexible supply capability at the material level for the differentiated performance requirements of different areas of the component.

[0019] 4. Construct a data-driven neural network proportioning mapping model to achieve efficient and high-precision conversion from three-dimensional stress field data to dynamic printing control parameters. This invention abandons the traditional fixed threshold mapping method and constructs a multi-objective optimization model based on NSGA-II—a neural network proportioning mapping model. A neural network surrogate model is introduced as a surrogate evaluation model to accelerate fitness calculation, enabling rapid adaptive solution of the optimal multi-material proportions under arbitrarily complex stress states.

[0020] 5. The neural network proportioning mapping model is trained using a large amount of sample data. By leveraging the powerful nonlinear mapping capability of the neural network, the optimal mixing proportion under any complex stress state can be predicted quickly and adaptively.

[0021] 6. Considering the stability of the overall mechanical properties of the structural components, compensation is made for the material ratio output by the neural network ratio mapping model. Spatial interpolation and smooth transition processing are performed in the stress gradient abrupt change region or the boundary region of different material ratios to ensure the continuous and gradual change of material composition in the macrostructure, fundamentally avoiding stress concentration, interface delamination or interlaminar shear failure caused by abrupt changes in material ratio. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the structure of a continuous fiber prepreg system provided in an embodiment of the present invention; Figure 2 yes Figure 1 A cross-sectional view of the continuous fiber prepreg system in the image; Figure 3 This is a flowchart of the three-dimensional stress field numerical simulation and feature extraction steps for the target structural component proposed in this invention; Figure 4 This is a schematic diagram of the network architecture of the neural network proxy model in the data-driven neural network matching mapping model proposed in this invention; Figure 5 This is a flowchart illustrating the NSGA-II algorithm in the data-driven neural network matching mapping model proposed in this invention. Figure 6 This is a schematic diagram of the variable strength continuous fiber composite material 3D printing molding process and collaborative control step module proposed in this invention; Figure 7 This is a diagram illustrating the manufacturing process of a frustum structure, which is an embodiment of the stress-adaptive 3D printing method with real-time control of multi-material ratio proposed in this invention.

[0023] In all the accompanying drawings, the same reference numerals are used to denote the same elements or structures, wherein: 11-continuous fiber feed spool, 12-AC servo unwinding motor, 13-tension sensor, 14-resin extrusion servo motor, 15-micro screw extruder, 16-third anti-backflow check valve, 21-spiral stirring impeller, 22-mixing resin chamber, 23-heating jacket, 24-impregnation chamber, 25-staggered display pins, 26-continuous fiber, 27-first anti-backflow check valve, 28-second anti-backflow check valve, 31-gradient wear-resistant nozzle, 32-pressure sensor, 33-temperature sensor. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0025] This invention provides a continuous fiber prepreg system that breaks through the limitation of fixed material composition in traditional continuous fiber 3D printing, and achieves precise matching between material strength and component stress state, taking into account both structural performance and lightweight objectives.

[0026] Please see Figure 1 and Figure 2 The continuous fiber prepreg system, as a material preparation unit of the 3D printing system, aims to transform material proportioning instructions into a continuous fiber composite material at the physical level. By precisely adjusting the resin extrusion pressure, impregnation temperature, or fiber traction speed, the continuous fiber prepreg system achieves real-time continuous control of the resin volume fraction or the mixing ratio of different types of resins in the continuous fiber bundle, thereby obtaining a continuous fiber prepreg bundle with variable mechanical properties at the printing front end.

[0027] The continuous fiber prepreg system includes a resin and fiber material conveying module, a multi-chamber composite material impregnation module, and an extrusion control and detection module connected in sequence, and achieves coordinated operation by relying on a distributed closed-loop control system.

[0028] The resin and fiber material conveying module provides a predetermined flow rate of multi-component resin and continuous fibers 26 with constant tension to the multi-chamber composite material impregnation module. The multi-chamber composite material impregnation module mixes the multi-component resin and then mixes it with the continuous fibers 26 to obtain a prepreg tow, and then transmits the prepreg tow to the extrusion control and detection module. The extrusion control and detection module controls the extrusion of the prepreg tow. The multi-chamber composite material impregnation module is provided with a connected mixing resin chamber 22 and an impregnation chamber 24, which are arranged in a stepped, vertically connected series configuration. The multi-component resin is uniformly mixed in the mixing resin chamber 22 and then enters the impregnation chamber 24 to mix with the continuous fibers 26 to obtain the prepreg tow.

[0029] The resin and fiber material delivery module includes a continuous fiber bundle feeding unit and a multi-source resin independent servo feeding unit connected in parallel to the multi-chamber composite material impregnation module. The continuous fiber bundle feeding unit provides continuous fibers with constant tension to the multi-chamber composite material impregnation module. The multi-source resin independent servo feeding unit provides a predetermined flow rate of multi-component resin to the multi-chamber composite material impregnation module.

[0030] The continuous fiber bundle feeding unit includes a continuous fiber feed reel 11 and an AC servo unwinding motor 12. A tension sensor 13 is disposed near the AC servo unwinding motor 12, and the tension sensor 13 is used to detect the tension of the continuous fiber filament. Continuous fiber filaments are disposed on the continuous fiber feed reel 11, and the AC servo unwinding motor 12 is used to pull the continuous fiber filaments to the multi-chamber composite material impregnation module.

[0031] The AC servo unwinding motor 12 is used to continuously feed the continuous fibers on the continuous fiber feed spool 11 to the multi-cavity solid composite material impregnation module; the tension sensor 13 is used to collect the tension of the continuous fiber bundle in real time and feed it back to the controller. The controller controls the AC servo unwinding motor 12 in real time according to the received tension value to prevent the fibers from loosening and knotting or from breaking or fuzzing due to excessive tension.

[0032] The multi-source resin independent servo feeding unit has at least two independent resin feeding and melting channels arranged in parallel. Each resin feeding and melting channel includes a micro-screw extruder 15 driven by a high-precision servo motor. A third anti-backflow check valve 16 is provided at the junction of each resin feeding and melting channel and the multi-chamber composite material impregnation module to prevent resin cross-flow when a chamber is pressurized. In one embodiment, the resin feeding and melting channel includes a resin extrusion servo motor 14, a micro-screw extruder 15, and a third anti-backflow check valve 16 connected in sequence.

[0033] The multi-chamber composite material impregnation module includes a stepped body, a spiral stirring impeller 21, a heating jacket 23, a first anti-backflow one-way valve 27, a second anti-backflow one-way valve 28, and staggered display pins 25. The mixing resin chamber 22 and the impregnation chamber 24 are respectively opened in the body and are located at opposite ends of the body. The body has a first resin inlet and a second resin inlet, both of which are connected to the mixing resin chamber 22. The multi-component resins enter the mixing resin chamber 22 through the first resin inlet and the second resin inlet, respectively. The first resin inlet and the second resin inlet are respectively equipped with the first anti-backflow one-way valve 27 and the second anti-backflow one-way valve 28. The heating jacket 23 is located on the outer side of the body area corresponding to the impregnation chamber 24.

[0034] The mixing resin chamber 22 is equipped with a spiral stirring impeller 21, which is used to mix the molten multi-component resin from the first resin inlet and the second resin inlet. The impregnation chamber 24 also has a fiber inlet and a fiber outlet, both of which are connected to the impregnation chamber 24. The uniformly mixed molten resin in the mixing resin chamber 22 enters the impregnation chamber 24 and mixes with the continuous fiber bundle entering the impregnation chamber 24 from the fiber inlet to form an impregnated fiber bundle. The impregnated fiber bundle exits from the fiber outlet and enters the extrusion control and inspection module. The impregnation chamber 24 is equipped with staggered display pins 25, which are used to support the continuous fiber bundle.

[0035] The extrusion control and detection module is used to extrude the prepreg tow and detect the temperature and pressure inside the nozzle. The extrusion control and detection module includes a temperature sensor 33, a pressure sensor 32, and a tapered abrasion-resistant nozzle 31, with the temperature sensor 33 and the pressure sensor 32 respectively mounted on the tapered abrasion-resistant nozzle 31. Relying on the tapered abrasion-resistant nozzle 31, which integrates the pressure sensor 32 and the temperature sensor 33, the extrusion control and detection module dynamically fine-tunes the heating temperature and back pressure of the impregnation chamber 24, stabilizing the output of high-quality, variable-strength continuous fiber prepreg tow with a regular cross-section and no defects.

[0036] The present invention also provides a 3D printing method with adjustable multi-material ratio, wherein the 3D printing method uses the continuous fiber prepreg system as described above as the feeding unit.

[0037] Please see Figure 3 , Figure 4 , Figure 5 and Figure 6 The printing method mainly includes the following steps: Step 1: Numerical simulation and feature extraction of the three-dimensional stress field of the target structural component.

[0038] For the three-dimensional structural model of the structural component to be printed, combined with the boundary conditions and load conditions given by the actual service conditions of the structural component, the finite element analysis method is used to perform three-dimensional stress field simulation of the three-dimensional structural model of the structural component to be printed, extract the stress tensor and principal stress distribution characteristics inside the structural component, and then combine the spatial coordinates of the nodes and their corresponding stress values ​​into data pairs.

[0039] In one specific implementation, the stress data of discrete nodes obtained from finite element analysis of the structural component are bound and integrated with their corresponding three-dimensional spatial coordinates. This unstructured node data is then reconstructed into a high-dimensional data matrix structure, generating a high-quality three-dimensional mapping model of spatial coordinates and stress distribution. This step obtains the global mechanical response state of the structural component under service conditions and transforms it into a digital, structured data matrix, thereby providing a high-precision stress input benchmark for neural network intelligent decision-making. This implementation mainly includes two steps: acquiring the stress field of the structural component to be printed and establishing a three-dimensional mapping model of spatial coordinates and stress distribution. Figure 3 As shown.

[0040] To obtain the stress field of the structural component to be printed, the three-dimensional structural model of the component needs to be imported into the finite element simulation software. Then, according to the actual working conditions of the component in actual engineering service, constraint boundary conditions and load boundary conditions are applied to the three-dimensional structural model to obtain the stress tensor and principal stress distribution characteristics inside the structure, and to identify the high stress concentration area and low stress bearing area in the component.

[0041] Step 2: Intelligent mapping of stress-multi-material ratio and variable parameter path planning based on neural networks.

[0042] Using data pairs as input and the optimal material ratio of fiber to different resins as output, a data-driven neural network ratio mapping model is constructed. Then, the material ratio data output by the neural network ratio mapping model, the 3D slice model of the structure to be printed, and the printing path plan are collaboratively planned to compile and generate control G code.

[0043] The neural network proportioning mapping model is a multi-objective optimization model based on NSGA-II. In its iterative solution process, a neural network surrogate model is used to calculate the objective function value of the neural network proportioning mapping model under the corresponding material proportion. The neural network surrogate model is a feedforward neural network based on a backpropagation neural network or a deep multilayer perceptron with residual connections, and its input is a data pair.

[0044] This invention constructs a data-driven neural network proportioning mapping model to achieve efficient and high-precision conversion from three-dimensional stress field data to dynamic printing control parameters. It abandons the traditional fixed threshold mapping method and constructs a multi-objective optimization model based on NSGA-II—a neural network proportioning mapping model—and introduces a neural network surrogate model. This surrogate model acts as a proxy evaluation model to accelerate fitness calculation, enabling rapid adaptive solutions for the optimal multi-material proportions under arbitrarily complex stress states.

[0045] Specifically, the following steps are included: S2.1, Construction of a data-driven neural network matching mapping model.

[0046] Using local node data-data pairs as network input and the optimal blending ratio of fibers to different resins that satisfies the local mechanical requirements and achieves lightweighting as output, a data-driven neural network blending mapping model is constructed. This neural network blending mapping model is based on the NSGA-II multi-objective optimization model. The neural network blending mapping model is trained using a large amount of sample data, leveraging the powerful nonlinear mapping capability of the neural network to achieve fast-adaptive prediction of the optimal blending ratio under arbitrary complex stress states.

[0047] S2.2, considering the stability of the overall mechanical properties of the structural components, a spatial continuity compensation algorithm is introduced for the optimal mixing ratio (optimal material ratio) output by the neural network ratio mapping model. Spatial interpolation and smooth transition processing are performed in stress gradient abrupt change regions or at the boundaries of different material ratios to ensure continuous and gradual changes in material composition on the macroscopic structure, fundamentally avoiding stress concentration, interface delamination, or interlaminar shear failure caused by abrupt changes in material ratio.

[0048] S2.3, Material-Trajectory Co-operational G-code Generation Technology: The material ratio data predicted by the neural network ratio mapping model, the three-dimensional slice model of the structure to be printed, and the printing path planning are deeply co-planned. Finally, customized control G-code containing spatial pose coordinates, print head movement speed, and prepreg system driving instructions is generated, providing instruction-level support for real-time control of the system.

[0049] In one implementation, a feedforward neural network based on a backpropagation neural network (BPNN) or a deep multilayer perceptron with residual connections (Res-MLP) is constructed as a neural network surrogate model to replace time-consuming finite element calculations. The input to this neural network surrogate model is data pairs, and the output is the mechanical response parameters under the corresponding material mix design. See [link to relevant documentation]. Figure 4 .

[0050] For each node Construct high-dimensional input feature variables :

[0051] In the formula, For spatial coordinates, The magnitude of the principal stress, For a complete stress tensor with six independent components, The stress gradient is the stress gradient. The output of the neural network surrogate model is the local yield strength. Weight per unit volume of material .

[0052] The neural network surrogate model was trained using over 5000 pre-collected finite element simulation samples, each sample containing a node feature vector and its corresponding mechanical response label. Training employed the Adam optimizer with a mean squared error loss function, a batch size of 128, and 500 training epochs, while early stopping was used to prevent overfitting. After training, the neural network surrogate model can output mechanical performance predictions for arbitrary alloy composition schemes within milliseconds, achieving a speedup of more than four orders of magnitude compared to finite element calculations.

[0053] For each local region in the structural component, the NSGA-II algorithm is run for multi-objective optimization. The neural network matching mapping model is defined as follows: Decision variable: Material proportion vector ,in This represents the volume fraction of continuous fibers. The percentage of resin A is given by [formula], and the percentage of resin B is given by [formula]. Sure.

[0054] The decision space constraint is:

[0055] Objective function: Maximize local mechanical properties Minimize material weight .in, and The output is quickly generated by the trained neural network proxy model.

[0056] Constraints: To ensure that continuous fibers form a continuous network, manufacturability constraints are set. To prevent interface delamination failure due to excessively low proportion of a single resin, interface compatibility constraints are set. .

[0057] For each local region, perform NSGA-II optimization and set the population size. Maximum number of iterations ; Employing simulated binary crossover SBX, distribution index Crossover probability Multinomial variation and distribution index were used. Probability of mutation Number of elite individuals retained per generation The execution process steps are as follows: Figure 5 As shown, specifically: (1) Initialization: 100 initial individuals (initial ratio of fiber to resin) are randomly generated within the volume ratio range of fiber to resin to form the parent population. .

[0058] (2) Fitness assessment: For each individual in the population, the objective function value is calculated using a neural network surrogate model: local yield strength. Weight per unit volume of material .

[0059] (3) Non-dominated ranking: The population individuals are ranked according to Pareto dominance and the non-dominated level of each individual is calculated.

[0060] (4) Crowding distance calculation: For individuals within the same non-dominated layer, calculate their crowding distance to maintain population diversity.

[0061] (5) Selection operation: A binary tournament selection mechanism is adopted, giving priority to individuals with low non-dominance level and large crowding distance to enter the mating pool.

[0062] (6) Crossover and mutation: Individuals in the mating pool are classified according to the crossover probability. Perform SBX crossover, based on mutation probability. Perform polynomial mutation to generate offspring population .

[0063] (7) Merging and Elite Preservation: Merging the parent generation with offspring Merge the individuals and select the 100 best individuals as the next generation population based on non-dominated ranking and crowding distance. .

[0064] (8) Termination judgment: When the maximum number of iterations is reached. The algorithm terminates when the Pareto front change is less than the threshold for 20 consecutive generations; otherwise, it returns to step (2) to continue iterating.

[0065] After convergence, the Pareto optimal solution set for the local region is obtained. The optimal multi-material ratio vector is automatically selected based on a set safety factor, using the following selection strategy: (1) Set a safety factor Calculate the minimum yield strength required by the local design. .

[0066] (2) Select solutions from the Pareto set that satisfy the following conditions. All solutions.

[0067] (3) Among the selected solutions, choose the material weight. The solution with the smallest minimum value is taken as the final material ratio.

[0068] (4) If no solution satisfies the safety factor requirement, then it degenerates into a choice. The largest solution is used as the final material ratio.

[0069] Ultimately, each local node obtains the optimal allocation vector:

[0070] in, Through the above process, a rapid and adaptive solution for the optimal multi-material ratio under any complex stress state is achieved.

[0071] Subsequently, spatial continuity compensation is performed on the local node proportion data output by the neural network proportioning mapping model to ensure the continuous and gradual change of material composition in the macroscopic structure and avoid interface failure caused by abrupt changes in proportion. Specifically, based on the comparison between the distance between a node and its neighboring points and a set threshold, if the distance is less than the threshold, the material proportion corresponding to the node is compensated to avoid abrupt changes in the material proportion.

[0072] Discretize the design space of the structural component into side lengths A uniform voxel mesh was constructed. Using the material proportion data of the nodes output by the neural network proportioning mapping model as known sample points, ordinary kriging interpolation was employed, with a search neighborhood radius set to 3 mm ~ 5 mm. , , The three material parameters are spatially interpolated independently. In regions of abrupt stress gradient changes or at the interface between high and low strength materials, the interpolation algorithm automatically smooths the proportion distribution based on spatial correlation and outputs a smoothed voxel proportion matrix.

[0073] Traverse all voxel nodes and calculate the spatial variation rate of material composition between each voxel and its 26 neighboring voxels:

[0074] in, , , , , , , They are respectively Point and The continuous fiber volume fraction, the proportion of resin A, and the proportion of resin B at the point. The Euclidean distance between the centers of the two voxels is given.

[0075] Set the maximum safe threshold for the rate of change of material proportions in space. .when When a sudden change risk is detected in a local region, a three-dimensional Gaussian smoothing filter is triggered to perform a secondary correction on the corresponding local proportion matrix. This correction process is iteratively executed until the rate of change across the entire region meets the constraint or the maximum number of iterations is reached. After correction, [the following steps are performed]. Perform boundary clipping on and Perform normalized projection to ensure This compensation treatment fundamentally avoids stress concentration, interfacial delamination, or interlaminar shear failure caused by abrupt changes in mix proportions.

[0076] Based on this, the material ratio data, the three-dimensional slice model of the structural component to be printed, and the printing path planning are deeply collaboratively planned, and compiled to generate customized control G-code containing multi-dimensional information.

[0077] The slicing engine is invoked to read the 3D model, and the layer thickness parameter is set to 0.2 mm ~ 0.4 mm to generate the internal filling path of each slice outline.

[0078] The generated continuous filling path is discretized into micro-segments with a fixed arc length of 1 mm to obtain a discrete sequence of trajectory endpoints. For each trajectory endpoint... According to its spatial coordinates Extract the corresponding material ratio parameters from the output smooth material ratio. .

[0079] Based on the printhead movement speed and nozzle cross-sectional area The multi-channel extrusion amount for each trajectory point is calculated using the following formula: Total volumetric flow rate of matrix resin: ; Volumetric flow rate of resin A: ; Resin B volumetric flow rate: ; Continuous fiber volumetric flow rate: .

[0080] Further converted into extrusion step instructions for multiple servo motors:

[0081] Finally, the post-processing algorithm converts the spatial pose coordinates Print head movement speed Multi-channel extrusion rate of prepreg system The code is encapsulated at the instruction level and compiled to generate customized control G-code. This code is then directly sent to the main controller, providing high-precision instruction-level support for the synchronous speed regulation of multiple servo motors and the real-time control of multiple material ratios in the underlying prepreg system.

[0082] This invention fully realizes the efficient and high-precision conversion from three-dimensional stress field data to dynamic printing control parameters, forming the core software driver module of the entire multi-material ratio real-time control stress adaptation 3D printing system.

[0083] Step 3: 3D printing molding process and collaborative control of variable strength continuous fiber composite materials.

[0084] The generated G-code is input into the 3D printing system to achieve spatiotemporal synchronization and closed-loop control between the continuous fiber prepreg system and the printing actuator, thereby completing the 3D printing of the structural part to be printed.

[0085] The generated customized control code is imported into the 3D printing system to achieve spatiotemporal synchronization and closed-loop control between the prepreg system and the printing actuator. Based on the temporal relationship between the spatial motion trajectory of the print head and the material delivery of the prepreg system, the spatial motion commands and prepreg drive commands are time-synchronized and matched. For motion change areas such as start-stop, acceleration, and deceleration, the execution sequence of material ratio commands is set to advance or delay to ensure the precise correspondence between material components and deposition positions. In the multi-material interface transition area, the interface bonding quality between layers with different material ratios is improved by means of local laser preheating of the deposited layer surface and dynamic adjustment of the compaction force of the pressure rollers, ultimately completing the collaborative printing and molding of variable strength continuous fiber composite material components.

[0086] In one implementation, the generated variable parameter G code file is imported into the main controller, which then parses the code line by line. Figure 6 As shown, the process is decomposed into spatial motion commands, material proportioning commands, and auxiliary motion commands. The parsed commands are then sent to the motion control module, prepreg control module (controller), and auxiliary function control module, respectively, with each module sharing a unified time reference signal. Following a preset slicing sequence, the variable-strength continuous fiber composite material is printed layer by layer. During the printing of each layer, the main controller coordinates motion execution and material delivery in real time based on the material proportioning distribution and spatial trajectory of that layer, until the entire structural component is printed. After printing, the structural component, along with the printing platform, is removed.

[0087] The spatial motion command includes the displacement coordinates and movement speed of the printhead on each axis; the material ratio command includes the rotation speed or extrusion pressure setting value and impregnation temperature setting value of the resin extrusion servo motor 14 in the prepreg system; the auxiliary action command includes control commands such as laser heating power and pressure roller pressure for heating the material (prepreg filaments) coming out of the prepreg system.

[0088] The motion control module drives the print head to move along the planned trajectory according to the spatial motion instructions; the prepreg control module adjusts the prepreg system according to the material ratio instructions; and the auxiliary function control module performs interface enhancement operations according to the auxiliary instructions inserted in the code.

[0089] Because the prepreg system has inherent response delays, including the delivery delay of resin from extrusion to impregnation chamber 24, the time required for the pressure in impregnation chamber 24 to reach steady state, and the tension response lag caused by changes in fiber traction speed, it is necessary to set timing compensation parameters during execution so that the execution time of the material ratio command is triggered earlier than the spatial motion command.

[0090] Because the material ratios used at different locations vary during the printing process, keeping the laser heating power and roller pressure constant can lead to problems such as poor interlayer bonding. Therefore, the auxiliary function control module needs to adjust the laser heating power and the pressure of the roller on the deposited fiber bundles in real time based on preset curves showing the relationship between laser heating power and material ratio, and roller pressure and material ratio. This improves the molecular activity and wetting ability of the surfaces to be bonded, thereby enhancing interlayer bonding performance.

[0091] This invention also provides a 3D printing system with adjustable multi-material ratios. The 3D printing system includes a prepreg system, a printing actuator, and a main controller as described above. The main controller is connected to both the prepreg system and the printing actuator, and the prepreg system is connected to the printing actuator. This 3D printing system is used to implement the 3D printing method with adjustable multi-material ratios as described above.

[0092] The present invention will be further described in detail below with reference to specific embodiments.

[0093] This embodiment uses a frustum structure as an example to illustrate the variable strength continuous fiber composite material 3D printing method proposed in this invention. The overall process is as follows: Figure 7 As shown, the process includes: 3D model creation, finite element mesh generation, stress field analysis, node data extraction and reconstruction, material ratio optimization, G-code generation, and online prepreg continuous fiber composite material printing.

[0094] First, a 3D model of the component to be printed is created and imported into finite element analysis software for mesh generation. Tetrahedral elements are used to discretize the model, resulting in the finite element mesh model of the component. Then, boundary conditions and load conditions are applied according to the actual service conditions of the component, and the finite element model is solved using finite element methods to obtain the stress field distribution cloud map of the component.

[0095] Furthermore, the discrete node data in the finite element solution results are extracted to obtain the spatial coordinate information and corresponding stress information of each node, and node feature vectors are constructed. Subsequently, the node data is cleaned and normalized to remove singular points and outliers, and then reconstructed into a standardized high-dimensional data matrix for subsequent neural network surrogate model training and material ratio optimization.

[0096] Furthermore, the standardized node feature data is input into a pre-trained surrogate evaluation neural network model, and combined with the NSGA-II multi-objective optimization algorithm, multi-material ratio optimization is performed on different regions of the component. The optimization objectives include maximizing local mechanical properties and minimizing material weight, and the decision variables include the continuous fiber volume fraction, the proportion of resin A, and the proportion of resin B. Through non-dominated sorting and Pareto optimal solution screening, the optimal material ratio scheme corresponding to each local region of the component is obtained.

[0097] Furthermore, the continuous material ratio matrix is ​​mapped to the component slicing path. During slicing, the component printing path is planned. Subsequently, the slicing path is discretized according to a fixed arc length, and the corresponding material ratio parameters are extracted by trilinear interpolation based on the spatial coordinates of the trajectory points, generating a G-code file containing spatial motion information, multi-material ratio information, and auxiliary process parameters.

[0098] Furthermore, the generated G-code is imported into the printing control system. The online prepreg system dynamically mixes and impregnates continuous fibers with multi-component resin in the impregnation chamber 24, and continuously extrudes them through a tapered nozzle to form prepreg bundles. Under the action of laser-assisted heating and pressure roller compaction, the prepreg bundles are deposited layer by layer on the surface of the printing platform, ultimately completing the molding and manufacturing of variable strength continuous fiber composite material components.

[0099] The stress-adaptive 3D printing method and system proposed in this invention, which achieves spatial adaptive distribution of material composition and fiber arrangement inside complex components through finite element stress field-driven material ratio optimization, continuous fiber path collaborative planning, and online prepreg dynamic control, can effectively improve the load-bearing efficiency, interlayer bonding performance, and lightweight level of the components.

[0100] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A continuous fiber prepreg system, characterized in that: The continuous fiber prepreg system includes a resin and fiber material conveying module, a multi-chamber composite material impregnation module, and an extrusion control and detection module connected in sequence. The resin and fiber material conveying module provides a predetermined flow rate of multi-component resin and continuous fibers with constant tension to the multi-chamber composite material impregnation module. The multi-chamber composite material impregnation module mixes the multi-component resin and then mixes it with the continuous fibers to obtain a prepreg tow, and then transmits the prepreg tow to the extrusion control and detection module. The extrusion control and detection module controls the extrusion of the prepreg tow. The multi-chamber composite material impregnation module is provided with a connected mixing resin chamber and an impregnation chamber, which are arranged in a stepped series configuration. After the multi-component resin is uniformly mixed in the mixing resin chamber, it enters the impregnation chamber and is mixed with the continuous fibers to obtain the pre-impregnated fiber bundle.

2. The continuous fiber prepreg system as described in claim 1, characterized in that: The resin and fiber material delivery module includes a continuous fiber bundle feeding unit and a multi-source resin independent servo feeding unit connected in parallel and both connected to the multi-chamber composite material impregnation module; the continuous fiber bundle feeding unit is used to provide continuous fibers with constant tension to the multi-chamber composite material impregnation module; the multi-source resin independent servo feeding unit is used to provide multi-component resin with a predetermined flow rate to the multi-chamber composite material impregnation module.

3. The continuous fiber prepreg system as described in claim 2, characterized in that: The multi-source resin independent servo feeding unit has at least two independent resin feeding and melting channels arranged in parallel. A third anti-backflow check valve is provided at the junction of the end of each resin feeding and melting channel and the multi-chamber composite material impregnation module. The resin feeding and melting channel includes a resin extrusion servo motor, a micro screw extruder and a third anti-backflow check valve connected in sequence.

4. The continuous fiber prepreg system according to any one of claims 1-3, characterized in that: The multi-chamber composite material impregnation module includes a stepped body, a first anti-backflow one-way valve, and a second anti-backflow one-way valve. The mixing resin chamber and the impregnation chamber are respectively opened in the body and are located at opposite ends of the body. The body has a first resin inlet and a second resin inlet, both of which are connected to the mixing resin chamber. The multi-component resins enter the mixing resin chamber through the first resin inlet and the second resin inlet, respectively. The first resin inlet and the second resin inlet are respectively equipped with the first anti-backflow one-way valve and the second anti-backflow one-way valve.

5. The continuous fiber prepreg system according to any one of claims 1-3, characterized in that: The extrusion control and inspection module includes a temperature sensor, a pressure sensor, and a tapered wear-resistant nozzle. The temperature sensor and the pressure sensor are respectively installed on the tapered wear-resistant nozzle, which is connected to the multi-chamber composite material impregnation module.

6. A 3D printing system with adjustable multi-material ratio, characterized in that: The 3D printing system includes a prepreg system, a printing actuator, and a main controller as described in any one of claims 1-5. The main controller is used to control the prepreg system and the printing actuator, and the prepreg system is connected to the printing actuator.

7. A 3D printing method with adjustable multi-material ratio, characterized in that: The 3D printing method is implemented using the multi-material ratio adjustable 3D printing system described in claim 6, and includes the following steps: (1) Based on the boundary conditions and load conditions given by the actual service conditions of the structural component to be printed, the finite element analysis method is used to simulate the three-dimensional stress field of the three-dimensional structural model of the structural component to be printed, extract the stress tensor and principal stress distribution characteristics inside the structural component, and then combine the spatial coordinates of the nodes and their corresponding stress values ​​into data pairs. (2) Using data pairs as input and the optimal material ratio of fiber to different resins as output, a data-driven neural network ratio mapping model is constructed. Then, the material ratio data output by the neural network ratio mapping model, the three-dimensional slice model of the structure to be printed, and the printing path planning are collaboratively planned to compile and generate control G code. (3) Input the generated G code into the 3D printing system to achieve spatiotemporal synchronization and closed-loop control between the continuous fiber prepreg system and the printing actuator, and complete the 3D printing of the structural parts to be printed.

8. The 3D printing method with adjustable multi-material ratio as described in claim 7, characterized in that: The neural network proportioning mapping model is a multi-objective optimization model based on NSGA-II. In its iterative solution process, a neural network surrogate model is used to calculate the objective function value of the neural network proportioning mapping model under the corresponding material proportion. The neural network surrogate model is a feedforward neural network based on a backpropagation neural network or a deep multilayer perceptron with residual connections, and its input is a data pair.

9. The 3D printing method with adjustable multi-material ratio as described in claim 7, characterized in that: The neural network matching mapping model is defined as follows: Decision variable: Material proportion vector ,in This represents the volume fraction of continuous fibers. The percentage of resin A is given by [formula], and the percentage of resin B is given by [formula]. Sure; The decision space constraint is: ； Objective function: Maximize local mechanical properties Minimize material weight ;in, and The output of the trained neural network proxy model.

10. The 3D printing method with adjustable multi-material ratio as described in claim 7, characterized in that: Spatial continuity compensation is performed on the material ratio data of local nodes output by the neural network ratio mapping model. Specifically, based on the comparison between the distance between the node and its neighboring points and a set threshold, if the distance is less than the set threshold, the material ratio corresponding to the node is compensated.

Images