A yarn let-off tension self-adaptive control method for manufacturing a complex structure preform

Abstract

The present application relates to a kind of complex structure preform manufacturing yarn let-off tension self-adaptive control method, comprising: according to the curvature of different regions of target three-dimensional preform surface, obtain the yarn density required in each region, obtain the first tension required for weaving different regions, and the first tension is the tension after yarn winding guide bar;First tension and the related parameters under current working condition are input into proxy model, and the initial tension predicted is obtained, and the initial tension is the tension before yarn winding guide bar;According to the initial tension predicted by the tension controller in device controls yarn;The related parameters include the contact angle between yarn and guide bar, the friction coefficient in contact process, the structural parameters of guide bar and yarn;The proxy model is constructed by training set to neural network model training.This application can accurately control the initial tension of yarn in real time, ensure that yarn always maintains the best tension state in the weaving process of preform.

Description

Technical Field

[0001] This invention relates to the field of yarn tension control technology, and in particular to an adaptive control method for warp tension of yarns used in the manufacture of complex prefabricated structures. Background Technology

[0002] Three-dimensional preforms, characterized by their compact structure, excellent mechanical properties, high strength, and high toughness, have been widely applied in high-tech fields such as aerospace. The fabrication process of three-dimensional preforms is highly complex, with yarn tension being a key parameter affecting the molding quality and directly influencing the final product's mechanical properties and structural stability. Compared to traditional planar fabrics, three-dimensional preforms possess a significant three-dimensional morphology and complex curved surface structures, exhibiting certain curvatures and undulations. This necessitates stricter control over yarn tension during manufacturing. Specifically, different regions of the three-dimensional preform have varying yarn tension requirements. Some regions, due to their curvature and undulations, exhibit complex curvatures, requiring higher yarn densities to enhance their strength and load-bearing capacity; while other regions require relatively lower densities. Therefore, the warp tension of the yarn during preform manufacturing needs to be rationally adjusted based on the different curvatures of the three-dimensional preform surface and the density differences in different regions to ensure the functionality and structural stability of the three-dimensional preform.

[0003] During the weaving process, the yarn is drawn from the yarn bobbin, its direction is changed by the yarn guide frame, and then it is transferred to the production line of the target 3D preform. Because the yarn's direction changes after being transferred by the guide frame, the angle changes, and the yarn tension changes accordingly. Therefore, the initial warp tension required during the actual weaving process is unclear. Traditional tension control methods determine the initial yarn tension setting range through experience-based adjustments, resulting in low accuracy. Furthermore, the complex local surface curvature within the 3D preform causes the yarn weaving density to be spatially uneven. How to rationally set the initial tension range of the yarn in each area based on the changes in the surface curvature of the 3D preform, ensuring that the yarn is always in an optimal tension state throughout the weaving process of the target 3D preform, and guaranteeing the excellent mechanical properties, structural stability, and product quality of the woven 3D preform, is a pressing problem that needs to be solved. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an adaptive control method for the warp tension of yarns used in the manufacture of complex prefabricated structures. The aim is to ensure that the yarn remains in optimal tension throughout the entire weaving process of the target three-dimensional prefabricated structure.

[0005] The technical solution adopted in this invention is as follows:

[0006] This invention provides an adaptive control method for warp tension of yarns used in the manufacture of complex prefabricated structures, comprising:

[0007] Based on the curvature of different regions on the surface of the target three-dimensional preform, the required yarn density for each region is obtained, and the first tension required for weaving different regions is obtained. The first tension is the tension of the yarn after it passes through the guide bar.

[0008] Input the first tension and relevant parameters under the current working condition into the surrogate model to obtain the predicted initial tension, which is the tension of the yarn before it passes through the guide bar;

[0009] The yarn is controlled according to the predicted initial tension to meet the weaving requirements of the target three-dimensional preform;

[0010] The relevant parameters include the contact angle between the yarn and the guide bar, the coefficient of friction during the contact process, and the structural parameters of the guide bar and the yarn.

[0011] The proxy model is constructed by training a neural network model using a training set.

[0012] The further technical solution is as follows:

[0013] The construction of the training set includes:

[0014] Establish a three-dimensional model of the yarn winding around the warp guide bar;

[0015] Set the calculation conditions, perform numerical simulation on the three-dimensional model, and calculate the simulated values ​​of the initial tension under different working conditions;

[0016] The simulated values ​​were experimentally verified, and the simulated values ​​were corrected based on the experimental results;

[0017] The corrected initial tension and the corresponding model parameters and structural parameters of the three-dimensional model under each working condition are used to form sample data. The sample data corresponding to each working condition are then summarized into the first training set.

[0018] The proxy model is constructed by training the neural network model using the first training set;

[0019] The calculation conditions include numerical model parameters and solver analysis conditions.

[0020] The construction of the training set also includes:

[0021] Using the data from the first training set, with the yarn tension theoretical model as the target fitting function, the undetermined coefficients of the target fitting function are determined through nonlinear curve fitting.

[0022] By using the controlled variable method and the yarn tension theoretical model, theoretical values ​​of initial tension under more working conditions can be calculated.

[0023] The theoretical values ​​and their corresponding variable values ​​are used to form a second training set, which is then combined with the first training set to form a total training set.

[0024] The proxy model is constructed by training the neural network model using the total training set;

[0025] The yarn tension theoretical model is as follows:

[0026]

[0027] In the formula, F b F a θ represents the tension of the yarn before and after it passes over the guide bar, respectively; θ represents the arc-shaped contact angle between the yarn and the guide bar; r is the radius of the guide bar; a is the nonlinear friction coefficient between the yarn and the guide bar; and a and n are the undetermined coefficients, both of which are constants related to the materials of the two objects at the point of contact.

[0028] The numerical model parameters include:

[0029] Yarn density, elastic modulus, Poisson's ratio, and tensile strength;

[0030] The material of the yarn guide bar;

[0031] The initial tension of the yarn after it has passed through the guide bar;

[0032] The contact relationship and coefficient of friction between the yarn and the guide wheel;

[0033] The loads and boundary conditions of the three-dimensional model.

[0034] The solver analysis conditions are set, including:

[0035] Set two load steps and set the calculation substeps for each load step;

[0036] Gravity is applied to the entire 3D model during the calculation time of the first load step, and the first tension is applied to the yarn during the calculation time of the second load step.

[0037] The setting of the boundary conditions includes:

[0038] Fixed supports are applied to the two end faces of the yarn guide bar to ensure that it does not move;

[0039] A fixed support is applied to one end of the yarn, and a force is applied to the other end along the direction of yarn transmission.

[0040] The step of experimentally verifying the simulated values ​​and correcting the simulated values ​​based on the experimental results includes:

[0041] An experimental platform for yarn winding around a warp guide bar was built, and experimental tests were conducted using experimental conditions corresponding to those in the numerical simulation calculations to obtain the test values ​​of the initial tension under the corresponding working conditions.

[0042] The test value is compared with the simulated value. If the error is not within the set range, the test value is used to replace the simulated value, thereby correcting the simulated value.

[0043] The structural parameters of the three-dimensional model include the yarn radius, the guide rod radius, and the contact angle between the yarn and the guide rod.

[0044] The construction of the proxy model includes:

[0045] The training set is divided into a training dataset and a validation dataset;

[0046] The neural network model parameters are optimized using the training dataset.

[0047] The neural network model was evaluated using the validation dataset.

[0048] Training is completed when the mapping relationship between the corrected initial tension and the corresponding model parameters and structural parameters of the three-dimensional model meets the set requirements.

[0049] The neural network model includes convolutional neural networks, recurrent neural networks, or multilayer perceptrons.

[0050] The method controls the yarn according to the predicted initial tension using a yarn feeding device, the yarn feeding device comprising:

[0051] A yarn tube frame, on which several yarn tubes are mounted;

[0052] The guide frame is composed of a grid structure of several yarn guide bars of the same diameter arranged in a crisscross pattern. After a single yarn exits from the yarn bobbin, it passes through a single grid and around the yarn guide bars, thereby being transferred to the production line of the target three-dimensional preform.

[0053] A yarn tension controller is used to adjust the initial tension of the single yarn to match the predicted initial tension.

[0054] The distance between the yarn tube frame and the guide frame is adjustable.

[0055] The beneficial effects of this invention are as follows:

[0056] This invention achieves accurate prediction of yarn tension changes before and after winding around a guide bar by constructing and applying a surrogate model. Based on the complex curvature surface and non-uniform yarn density differences of the three-dimensional preform, the surrogate model obtains the predicted initial tension in real time using measured parameters under the current working conditions. This allows for precise control of the initial yarn tension, ensuring that the yarn remains in optimal tension throughout the weaving process. This meets the differentiated tension requirements of different regions of the target three-dimensional preform, guaranteeing excellent mechanical properties, structural stability, and product quality.

[0057] The numerical simulation results of this invention have been experimentally verified and are more accurate. The constructed first training set contains sample data under different working conditions, which fully meets the basic requirements for constructing a proxy model.

[0058] This invention employs a combination of numerical simulation and theoretical model calculations to construct a high-quality second training set. This expands the operating conditions while simultaneously improving computational accuracy and efficiency, further enhancing the prediction accuracy of the surrogate model.

[0059] This invention is applicable to the initial tension control of high-volume yarns, with a wide range of applications and strong versatility.

[0060] Other features and advantages of the invention will be set forth in the following description or may be learned by practicing the invention. Attached Figure Description

[0061] Figure 1 This is a schematic diagram of the yarn feeding device according to an embodiment of the present invention.

[0062] Figure 2 This is a schematic diagram of the geometric structure of the yarn tension theoretical model in an embodiment of the present invention.

[0063] Figure 3 This is a schematic diagram of the structure of the target three-dimensional prefabricated body in an embodiment of the present invention.

[0064] Figure 4 The figure shows the numerical simulation results under the working conditions of the embodiment of the present invention.

[0065] Figure 5 The figure shows the calculation results of the yarn tension theoretical model under the working conditions of the embodiment of the present invention.

[0066] Figure 6 This is a comparison chart of the calculation results of the yarn tension theoretical model and the simulation results under the working conditions of the present invention.

[0067] In the diagram: 1. Yarn bobbin; 2. Yarn bobbin holder; 3. Yarn tension controller; 4. Single yarn; 5. Guide frame; 51. Yarn guide bar. Detailed Implementation

[0068] The specific embodiments of the present invention are described below with reference to the accompanying drawings.

[0069] This embodiment provides an adaptive control method for yarn warp tension in the manufacturing of complex prefabricated structures. This method can control the operation of the yarn warp feeding device. (See also...) Figure 1 The yarn feeding device includes:

[0070] Yarn tube frame 2, on which several yarn tubes 1 are provided;

[0071] The guide frame 5 is composed of a grid structure of several yarn guide bars of the same diameter that are crisscrossed. After a single yarn 4 exits from the yarn tube 1, it passes through a single grid and around the yarn guide bar, thereby being transferred to the production line of the target three-dimensional preform.

[0072] Yarn tension controller 3, which is used to adjust the initial tension of the single yarn 4.

[0073] In a specific implementation, the distance between the yarn tube frame 2 and the guide frame 5 is adjustable.

[0074] In a preferred embodiment, the grid formed by the crisscrossing yarn guides is arranged in a uniform array.

[0075] In a preferred embodiment, the yarn tubes 1 are evenly distributed in an array on the yarn tube frame 2.

[0076] In a preferred embodiment, the yarn tension controller 3 is mounted on the surface of the yarn bobbin 2.

[0077] As a preferred embodiment, the yarn guide bar is made of stainless steel.

[0078] Based on the nonlinear friction theorem, the process of yarn changing direction as it wraps around an external object can be considered as the process of yarn wrapping around the surface of a cylinder. Therefore, it is possible to target... Figure 2 The following is a theoretical model of yarn tension based on the geometric structure of a single yarn 4 passing around the yarn guide bar 51:

[0079]

[0080] In the formula, F b The initial tension, F, represents the tension of the yarn before it passes over the guide bar. a The tension of the yarn after it passes over the guide bar is the first tension; θ represents the contact angle between the yarn and the guide bar in arc; r is the radius of the cylindrical guide bar; a is the nonlinear friction coefficient between the yarn and the guide bar. a and n are both constants related to the materials of the two objects at the contact point, and are both undetermined coefficients that can be obtained through nonlinear fitting.

[0081] The adaptive control method includes the following steps:

[0082] S1. Based on the curvature of different regions on the surface of the target three-dimensional preform, the required yarn density for each region is obtained, and the first tension required for weaving different regions is obtained.

[0083] As a specific implementation method, such as Figure 3 As shown in the figure, A, B, and C represent three different regions of the target 3D prefabricated body. Due to the specific application fields and scenarios of the target 3D prefabricated body, in order to meet specific mechanical performance requirements, such as high load-bearing capacity, high impact resistance, high strength, and high wear resistance, the target 3D prefabricated body needs to maintain a specific shape and surface curvature. For example, the edges, corners, and recessed parts of the 3D prefabricated body need to exhibit higher curvature; while other regions are close to a plane, so the corresponding curvature values ​​are lower. Therefore, different regions need to exhibit different surface curvatures. The yarn density and surface curvature are positively correlated in local regions. For high-density fabric regions, greater yarn tension is required to prevent yarn slippage or deformation during weaving, while in low-density regions, the distance between yarns is larger, requiring less yarn tension to prevent yarn breakage or fabric deformation. For regions exhibiting higher curvature, due to the bending and undulation of the surface, the yarns in that region are compressed and arranged more tightly, resulting in an increased local weaving density coefficient; while regions with lower curvature values ​​have a smaller weaving density coefficient. Figure 3 The required yarn weaving density coefficients corresponding to the surface curvature of regions A, B, and C are 0.107–0.493, 0.051–0.222, and 0.061–0.224, respectively. Based on the yarn density, the required yarn tension, i.e. the first tension, for the three different regions of the target three-dimensional preform are 1.30N–2.50N, 0.60N–1.20N, and 0.70N–1.25N, respectively.

[0084] S2. Input the first tension and relevant parameters under the current working condition into the surrogate model to obtain the predicted initial tension.

[0085] The relevant parameters include the contact angle between the yarn and the guide bar, the coefficient of friction during the contact process, and the structural parameters of the guide bar and the yarn.

[0086] The proxy model is constructed by training a neural network model using a training set.

[0087] The construction of the training set includes:

[0088] S21. Establish a three-dimensional model of the yarn winding around the warp guide bar.

[0089] S22. Set calculation conditions and perform numerical simulation on the three-dimensional model to calculate the simulated values ​​of the initial tension under different working conditions.

[0090] The calculation conditions include numerical model parameters and solver analysis conditions.

[0091] The numerical model parameters include:

[0092] Yarn density, elastic modulus, Poisson's ratio, and tensile strength;

[0093] The material of the yarn guide bar;

[0094] The initial tension of the yarn after it has passed through the guide bar;

[0095] The contact relationship and coefficient of friction between the yarn and the yarn guide bar;

[0096] The loads and boundary conditions of the three-dimensional model.

[0097] S23. Verify the simulated values ​​experimentally, and correct the simulated values ​​based on the experimental results;

[0098] S24. The corrected initial tension and the corresponding model parameters and structural parameters of the three-dimensional model under each working condition are combined to form sample data. The sample data corresponding to each working condition are summarized into a first training set. The surrogate model can then be constructed by training the neural network model using the first training set.

[0099] The structural parameters of the three-dimensional model include the yarn radius, the guide rod radius, and the contact angle between the yarn and the guide rod.

[0100] In a specific implementation, steps S21 and S22 include the following steps:

[0101] (1) Three-dimensional model construction:

[0102] Based on the actual working conditions of the yarn, an arbitrary circle is drawn on the front reference plane with the origin as the center using SolidWorks 3D modeling software. The diameter of the circle is set to 40mm using the Smart Dimension function, and the extrusion depth is set to 40mm, completing the construction of a cylindrical 3D model with a diameter and height of 40mm. Next, a new reference plane is created that coincides with the middle plane of the cylinder. On this reference plane, an arc is drawn with the center of the cylinder's end face as the center, and the radius of curvature of the arc segment is specified as 20.5mm, indicating that the cross-section of the yarn is a circle with a diameter of 1mm. Then, using the outer surface of the cylinder as a reference object, two path lines are drawn on the new reference plane, with the endpoints of the two path lines coinciding with the two endpoints of the arc.

[0103] Add tangential mating relationships to the created path lines and arcs, and set the contact angle between the yarn and the cylinder (90°~180°) using intelligent dimensional settings to make it conform to the actual path. After verifying that the path relationship between the cylinder and the yarn is correct, save the constructed 3D model as a step file, and finally generate a 3D model file that can be imported into the Ansys simulation tool.

[0104] (2) Calculation condition settings:

[0105] Import the constructed 3D model file into the Workbench module of Ansys simulation software, select the transient structure module, and set the yarn density to 1800 kg / m² according to the actual working conditions. 3 The cylindrical body is made of stainless steel with an elastic modulus of 80 GPa, a Poisson's ratio of 0.25, and a tensile strength of 3530 MPa. The mesh cell type of the yarn is set to cable cell, the contact relationship between the yarn and the cylinder is defined as frictional contact, and the coefficient of friction is specified as 0.2.

[0106] Boundary condition settings: Apply fixed supports to both end faces of the cylinder to ensure it does not move; apply fixed supports to one end of the yarn.

[0107] A fixed support is applied, and a force is set at the other end along the direction of yarn transmission to simulate the first tension of 1.35N after the yarn comes into frictional contact with the external cylinder.

[0108] Load condition setting: Apply standard gravity to the entire system simultaneously to approximate the stress state under real production conditions.

[0109] Mesh generation: The yarn is meshed with a fine unit size of 0.5mm to ensure the accuracy of the stress results, and the cylinder is meshed with a unit size of 2mm to shorten the calculation and solution time.

[0110] Solver analysis settings: In the solver analysis settings, set two load steps. Apply gravity to the entire system during the calculation time of the first load step, and apply initial tension to the yarn during the calculation time of the second load step. Set the number of calculation substeps for all load steps to 200 to ensure the convergence of the solution process. The entire simulation process lasts 4 seconds.

[0111] This specific implementation method only sets the relevant parameters and friction coefficient of the yarn according to specific production conditions, and sets the load conditions and analysis conditions. It is understood that the relevant parameters of the yarn, the friction coefficient between the yarn and the cylinder, and the initial tension of the yarn can be adjusted according to the actual working conditions.

[0112] (3) Data acquisition and analysis:

[0113] After setting the load and analysis conditions for the simulation, the solution calculation is performed, outputting the equivalent stress results of the yarn before and after it passes over the cylinder, and calculating the initial tension of the yarn based on the cross-sectional area of ​​the yarn. The numerical simulation results are as follows: Figure 4 As shown.

[0114] This specific implementation constructs a three-dimensional model of a circular yarn with a diameter of 1 mm in contact with a cylinder with a radius of 20 mm based on actual working conditions, setting the contact angle between the yarn and the cylinder to 90° to 180°. Using a contact angle of 90° as a reference, simulation is performed every 10° until 180° is reached, obtaining a series of data and a dataset of yarn tension changes under the current conditions.

[0115] By using the above steps, yarns with different cross-sectional dimensions and cylinders with different radii are combined, and a series of friction coefficients and initial tensions are set to simulate the initial tension of the yarn under various working conditions, obtaining a large amount of simulation data.

[0116] In a specific implementation, step S23 specifically includes:

[0117] An experimental platform for yarn winding around a warp guide bar was built, and experimental tests were conducted using experimental conditions corresponding to those in the numerical simulation calculations to obtain the test values ​​of the initial tension under the corresponding working conditions.

[0118] The test value is compared with the simulated value. If the error is not within the set range, the test value is used to replace the simulated value, thereby correcting the simulated value.

[0119] Because numerical simulations are time-consuming and the simulated values ​​need to be experimentally verified to ensure accuracy, the amount of simulation data obtained in actual operation is limited, i.e., the amount of data in the first training set is limited. Therefore, this embodiment further optimizes the construction of the training set by the following method: The construction of the training set further includes:

[0120] S25. Using the data from the first training set, the yarn tension theoretical model described above is applied... The undetermined coefficients a and n of the target fitting function are determined by nonlinear curve fitting.

[0121] S26. Using the controlled variable method, theoretical values ​​of initial tension under more working conditions are calculated through the yarn tension theoretical model.

[0122] S27. The theoretical values ​​and corresponding variable values ​​are used to form a second training set, which is then combined with the first training set to form a total training set.

[0123] In a specific implementation, step S25 includes:

[0124] Using the nonlinear fitting function of MATLAB software, the contact angle θ, the radius r of the cylindrical guide bar, and the tension F of the yarn before it passes over the guide bar were selected under the corresponding working conditions in the first training set. b and the tension F of the yarn after it passes over the guide bar a Nonlinear curve fitting is performed on a and n to obtain their specific values. Substituting these values ​​into the target fitting function, a nonlinear yarn tension theoretical model is obtained for a yarn with a friction coefficient of 0.2, a cross-section diameter of 1 mm, and a cylinder with a radius of 20 mm.

[0125] In a specific implementation, step S26 includes:

[0126] Based on the yarn tension theoretical model, the contact angle θ between the yarn and the guide bar, the radius r of the cylindrical guide bar, and the friction coefficient a between the yarn and the guide bar are changed sequentially by controlling the variables to calculate the corresponding results, i.e., the theoretical values, which constitute the second training set.

[0127] This embodiment can analyze the variation law of yarn tension with relevant parameters based on a large amount of data obtained by numerical simulation and theoretical model, providing a basis for research and application in related fields.

[0128] The calculation results using the yarn tension theoretical model are as follows: Figure 5 As shown.

[0129] Comparison of calculation results from the theoretical model of yarn tension with numerical simulation results. Figure 6 As shown, the fitting results between the two methods are good, verifying the accuracy of the calculation results in this implementation.

[0130] The data in the second training set was obtained directly from calculations on the theoretical model, resulting in high accuracy and a large dataset. Figure 5 It is known that the tension under any contact angle condition can be theoretically calculated, covering a vast number of conditions. Therefore, training the neural network model using the aforementioned total training set can further improve the accuracy of the surrogate model.

[0131] The construction of the proxy model involves training and optimizing the neural network model using factors such as contact angle, friction coefficient, cylinder radius, and initial tension after the yarn transmission angle change as inputs, and the initial tension before the yarn transmission angle change as output. A specific implementation includes the following steps:

[0132] (1) Selecting a neural network model: Analyze the size and characteristics of the training set to determine a suitable neural network model. Common model types include convolutional neural networks (CNN), recurrent neural networks (RNN), and multilayer perceptrons (MLP). Since the input parameters of the training set usually do not have significant temporal or spatial characteristics, this embodiment prefers the MLP model, which has a simple structure and is suitable for processing scalar inputs.

[0133] In addition, the structural parameters of the neural network model, such as the number of layers and the number of neurons per layer, are adjusted according to the data characteristics of the training set in order to balance the model's expressive power and training efficiency.

[0134] (2) Training set partitioning: The training set is divided into a training dataset and a validation dataset in a 7:3 ratio. The training dataset is used to optimize the parameters of the neural network model. The validation dataset is used to evaluate the performance of the neural network model. During the partitioning process, the consistency of data distribution is ensured to avoid the impact of data bias on the model training effect.

[0135] (3) Neural Network Model Training: The neural network is trained by taking X from the training dataset as input and the corresponding Y as output. The training set contains sample pairs (X1, Y1), (X2, Y2), ..., (Xs, Ys). X represents several input parameter vectors, including but not limited to contact angle, friction coefficient, and the radius of the cylinder; Y represents the initial yarn tension corresponding to the input parameters. Subscripts 1, 2, ..., s represent the sample data numbers corresponding to different working conditions.

[0136] During training, appropriate loss functions (such as mean squared error, MSE) and optimization algorithms (such as Adam or SGD) are selected, and training efficiency and effectiveness are improved by setting hyperparameters such as learning rate and batch size. Early stopping can also be employed to prevent overfitting of the model.

[0137] (4) Evaluation of the neural network model: The model is trained using X input from the validation dataset. Error metrics between the model output and Y input from the validation dataset are calculated, such as mean squared error (MSE), mean absolute error (MAE), and coefficient of determination (R²). 2 These metrics are used to comprehensively evaluate the model's fit and prediction accuracy in order to determine whether it meets performance requirements.

[0138] (5) Optimization of the neural network model: The model is optimized based on the evaluation results. Optimization methods include, but are not limited to, the following: 1. Adjusting the network structure, such as increasing the number of hidden layers or the number of neurons per layer, to improve the model's learning ability; 2. Adjusting hyperparameters, such as optimizing the learning rate and regularization coefficient, to balance the training speed and generalization ability of the model; 3. Increasing the amount of data or introducing data augmentation techniques to expand the diversity of training data; 4. Introducing cross-validation methods to further verify the stability and robustness of the model. The optimized model needs to repeat steps (3) and (4) to ensure that the mapping relationship between the modified initial tension and the corresponding model parameters and the structural parameters of the three-dimensional model meets the requirements, ensuring that the model meets expectations and completing the training.

[0139] The trained neural network model is used as a surrogate model in a real-world scenario. By inputting the first tension and related parameter data X measured at the current site, the surrogate model can quickly output the predicted initial tension value Y.

[0140] S3. The initial tension of the yarn is controlled by the yarn tension controller 3 to make it consistent with the initial tension value Y, thereby meeting the weaving requirements of the target three-dimensional preform.

[0141] In summary, this invention derives the required initial tension for different regions of the target 3D preform based on the surface curvature required for each region. Using 3D simulation or a combination of 3D simulation and theoretical models, the variation of the initial tension with the initial tension, contact angle, coefficient of friction, and structural parameters of the yarn and guide bar is obtained, forming a training set of data under different working conditions. The neural network model is then trained and optimized to obtain a surrogate model capable of accurately predicting yarn tension under different working conditions. By inputting relevant measured parameters into the surrogate model, the predicted initial yarn tension is obtained in real time. Finally, a yarn tension controller adjusts the yarn according to the predicted initial tension, ensuring it is delivered to the production line of the target 3D preform at the optimal tension. This specifically meets the weaving requirements of the target 3D preform, ensuring the yarn remains at its optimal tension throughout the weaving process, guaranteeing the excellent mechanical properties, structural stability, and product quality of the target 3D preform. This invention is applicable to the initial tension control of high-volume yarns.

[0142] It will be understood by those skilled in the art that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for adaptive control of warp tension in the manufacturing of complex prefabricated structures, characterized in that, include: Based on the curvature of different regions on the surface of the target three-dimensional preform, the required yarn density for each region is obtained, and the first tension required for weaving different regions is obtained. The first tension is the tension of the yarn after it passes through the guide bar. Input the first tension and relevant parameters under the current working condition into the surrogate model to obtain the predicted initial tension, which is the tension of the yarn before it passes through the guide bar; The yarn is controlled according to the predicted initial tension to meet the weaving requirements of the target three-dimensional preform; The relevant parameters include the contact angle between the yarn and the guide bar, the coefficient of friction during the contact process, and the structural parameters of the guide bar and the yarn. The proxy model is constructed by training a neural network model using a training set; The construction of the training set includes: Establish a three-dimensional model of the yarn winding around the warp guide bar; Set the calculation conditions, perform numerical simulation on the three-dimensional model, and calculate the simulated values ​​of the initial tension under different working conditions; The simulated values ​​were experimentally verified, and the simulated values ​​were corrected based on the experimental results; The corrected initial tension and the corresponding neural network model parameters and the structural parameters of the three-dimensional model under each working condition are used to form sample data. The sample data corresponding to each working condition are then summarized into the first training set. The proxy model is constructed by training the neural network model using the first training set; The calculation conditions include numerical model parameters and solver analysis conditions; The construction of the training set also includes: Using the data from the first training set, with the yarn tension theoretical model as the target fitting function, the undetermined coefficients of the target fitting function are determined through nonlinear curve fitting. By using the controlled variable method and the yarn tension theoretical model, theoretical values ​​of initial tension under more working conditions can be calculated. The theoretical values ​​and their corresponding variable values ​​are used to form a second training set, which is then combined with the first training set to form a total training set. The proxy model is constructed by training the neural network model using the total training set; The yarn tension theoretical model is as follows: ， In the formula, F b , F a These represent the tension of the yarn before and after it passes over the guide bar, respectively. The arc-shaped contact angle between the yarn and the guide bar; r The radius of the yarn guide bar; a The coefficient of friction between the yarn and the guide bar is the nonlinear friction coefficient. a and n The undetermined coefficients are all constants related to the materials of the two objects at the junction.

2. The method according to claim 1, characterized in that, The numerical model parameters include: Yarn density, elastic modulus, Poisson's ratio, and tensile strength; The material of the yarn guide bar; The initial tension of the yarn after it has passed through the guide bar; The contact relationship and coefficient of friction between the yarn and the guide wheel; The loads and boundary conditions applied to the three-dimensional model.

3. The method according to claim 2, characterized in that, The solver analysis conditions are set, including: Set two load steps and set the calculation substeps for each load step; Gravity is applied to the entire 3D model during the calculation time of the first load step, and the first tension is applied to the yarn during the calculation time of the second load step.

4. The method according to claim 2, characterized in that, The setting of the boundary conditions includes: Fixed supports are applied to the two end faces of the yarn guide bar to ensure that it does not move; A fixed support is applied to one end of the yarn, and a force is applied to the other end along the direction of yarn transmission.

5. The method according to claim 1, characterized in that, The step of experimentally verifying the simulated values ​​and correcting the simulated values ​​based on the experimental results includes: An experimental platform for yarn winding around a warp guide bar was built, and experimental tests were conducted using experimental conditions corresponding to those in the numerical simulation calculations to obtain the test values ​​of the initial tension under the corresponding working conditions. The test value is compared with the simulated value. If the error is not within the set range, the test value is used to replace the simulated value, thereby correcting the simulated value.

6. The method according to claim 1, characterized in that, The structural parameters of the three-dimensional model include the yarn radius, the guide rod radius, and the contact angle between the yarn and the guide rod.

7. The method according to claim 1, characterized in that, The construction of the proxy model includes: The training set is divided into a training dataset and a validation dataset; The parameters of the neural network model are optimized using the training dataset. The neural network model was evaluated using the validation dataset. Training continues until the mapping relationship between the corrected initial tension and the corresponding neural network model parameters and the structural parameters of the three-dimensional model meets the set requirements. The neural network model includes convolutional neural networks, recurrent neural networks, or multilayer perceptrons.

8. The method according to claim 1, characterized in that, The method controls the yarn according to the predicted initial tension using a yarn feeding device, the yarn feeding device comprising: Yarn tube frame (2), on which several yarn tubes (1) are provided; The guide frame (5) is composed of a grid structure of several yarn guide bars of the same diameter and intersecting horizontally and vertically; after a single yarn (4) comes out from the yarn tube (1), it passes through a single grid and goes around the yarn guide bar, thereby being transferred to the production line of the target three-dimensional preform. Yarn tension controller (3), which is used to adjust the initial tension of the single yarn (4) to match the predicted initial tension; The distance between the yarn tube frame (2) and the guide frame (5) is adjustable.

Images