Digital forming process for carbon fiber composite part

Abstract

The invention belongs to the technical field of carbon fiber composite material forming, and discloses a digital forming process of a carbon fiber composite material part, which accurately captures the influence of anisotropy and weaving structure of carbon fibers on forming through digital modeling and multi-scale simulation driven by material characteristics, and solves the problem of quality fluctuation caused by dependence on artificial experience in a traditional process. Real-time closed-loop control and a defect self-healing system are combined, forming deviation can be dynamically corrected, quality defects caused by improper fiber orientation, insufficient resin infiltration and the like are avoided, and the consistency of the quality of parts in different batches is ensured; through a parameter pre-judgment optimization and virtual verification platform verification technology of a neuromorphic chip, the process breaks through the limitation of a traditional trial and error method, and an optimal scheme can be quickly determined without repeatedly adjusting parameters; real-time linkage of forming data and process optimization is achieved through a digital thread and a closed loop feedback mechanism, efficiency loss caused by disconnection of the forming link and the detection link is avoided, and the process research and development period is greatly shortened.

Description

Technical Field

[0001] This invention belongs to the field of carbon fiber composite molding technology, specifically a digital molding process for carbon fiber composite parts. Background Technology

[0002] Carbon fiber composite materials are indispensable in aerospace, new energy vehicles, and high-end equipment manufacturing due to their high strength, lightweight, and fatigue resistance. In the aerospace industry, they are used to reduce aircraft weight and increase range; in the new energy vehicle industry, they are used to reduce vehicle weight and extend range; and in high-end equipment, they are used to enhance the corrosion resistance and lifespan of components, making them a key material driving the upgrading of high-end manufacturing.

[0003] Existing processes rely on manual experience to determine parameters, which involves a cumbersome trial-and-error process, resulting in poor part quality stability and difficulty in meeting high-precision requirements.

[0004] A search revealed that the invention patent with publication number CN117124616A discloses a rectangular carbon fiber composite honeycomb panel mold and molding method. The method uses FiberSIM software to generate layup information and project positioning, which improves layup efficiency and angle accuracy, but still has limitations: it only focuses on macroscopic layup positioning and does not consider the anisotropy of carbon fiber, the coupling effect of the braided structure with temperature field and pressure field, resulting in a large deviation between simulation and reality.

[0005] Furthermore, neither this patent nor similar technologies have established a closed-loop feedback mechanism for the molding process and quality inspection. The adjustment of process parameters lacks dynamic data support, resulting in low production efficiency and high costs, making it difficult to meet the high-precision manufacturing requirements of complex structural parts. Summary of the Invention

[0006] The purpose of this invention is to provide a digital molding process for carbon fiber composite parts to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a digital forming process for carbon fiber composite parts, the specific steps of which are as follows:

[0008] S1. Bio-inspired dynamic digital modeling: Simulate the interaction between carbon fiber and resin through artificial neural networks, construct a model that dynamically adjusts with humidity and prepreg storage time, embed fiber growth trajectory algorithm, integrate geometric and material information, and provide a realistic digital foundation for step S2.

[0009] S2. Quantum computing-assisted multi-field coupling simulation: Based on the dynamic model in step S1, multi-scale multi-field coupling simulation is carried out using the parallel capabilities of quantum computing. Quantum characteristics are incorporated to analyze field changes, capture fiber stress patterns, and output key parameters.

[0010] S3. Neuromorphic chip driving parameter optimization: Receive quantum simulation parameters, construct the system with neuromorphic computing, convert historical data into synaptic weights, predict the difficulties in forming and generate strategies, and provide parameters for step S4.

[0011] S4. Magnetoelectric Fusion Adaptive Molding Control: The magnetoelectric device is activated according to the optimized parameters. Through the collaboration of magnetostrictive material and piezoelectric crystal, dual closed-loop control is achieved, and simulation data is exchanged in real time to ensure that molding proceeds according to the optimized path.

[0012] S5. Photothermal Response Type Defect Self-Healing Control: In magnetoelectric molding, the digital twin system monitors and locates the position and type of fiber defects, activates the photothermal system to trigger microcapsule repair, and links mechanical reinforcement to form a digital closed loop for defect repair.

[0013] S6, Brain-Computer Interactive Virtual Verification System: Maps the molded data and digital model after defect repair to the virtual verification platform. It collects neural electrical signals through brain-computer interface devices and converts them into quantitative parameters. Combined with the force feedback interaction module, it realizes the simulation of physical characteristics. Researchers can transmit performance expectations through preset parameter commands, and the platform generates the corresponding virtual molding scheme in real time based on the input parameters.

[0014] S7. Carbon Footprint-Performance Dual-Objective Optimization: Through the verification scheme of the virtual verification platform, a carbon footprint and performance coupled model is constructed to optimize the parameter balance between performance and energy consumption, forming a synergistic closed loop and providing a sustainable benchmark for subsequent production.

[0015] Preferably, the specific steps of the bio-inspired dynamic digital modeling in step S1 are as follows:

[0016] S11. Technical basis of dynamic modeling: The interface mechanism between carbon fiber and resin is simulated by artificial neural network, focusing on capturing the mutual influence between the two under different environments. The model can respond to variables such as humidity fluctuation and prepreg storage time in real time. By embedding the carbon fiber bundle "growth trajectory" algorithm, the fiber weaving method and material distribution law are accurately mapped to construct a dynamic digital carrier that integrates geometric features and material properties.

[0017] The anisotropic constitutive equation for carbon fiber:

[0018] σ ij =C ijkl ε kl

[0019] In the formula: σ ij Let i be the stress tensor (i,j = 1, 2, 3, representing the stress components in three-dimensional space);

[0020] C ijkl The stiffness tensor reflects the anisotropy of carbon fiber and is related to the fiber weaving angle and the number of layers.

[0021] ε kl For strain tensor (k,|=1,2,3, representing the three-dimensional spatial strain components);

[0022] Formula source: Based on the extension of the classic Hooke's law and combined with the characteristics of carbon fiber braided structures, it is used for the coupling modeling of material properties and digital model in step S1.

[0023] S12. The model's connecting value: This model integrates the structural parameters of the parts with the dynamic characteristics of the materials, providing high-precision initial data for subsequent quantum computing simulations. Its dynamic adjustment capability ensures that the digital model always fits the actual state of the materials.

[0024] Preferably, the specific steps of quantum computing-assisted multi-field coupling simulation in step S2 are as follows:

[0025] S21. Technical Implementation of Multi-Scale Simulation: Based on a bio-inspired dynamic model, leveraging the parallel processing advantages of quantum computing, it breaks through the limitations of traditional computing power and realizes cross-scale analysis from the molecular scale to the macroscopic structure; it incorporates the quantum tunneling effect of carbon fiber and the quantum entangled state of resin molecules into the calculation, accurately reproducing the dynamic coupling process of multiple physical fields such as temperature field and pressure field;

[0026] Fiber orientation evolution equation:

[0027]

[0028] In the formula: θ is the instantaneous orientation angle of the fiber (°);

[0029] t represents time (s);

[0030] k is the orientation adjustment coefficient, which is related to fiber stiffness;

[0031] θ0 is the target orientation angle, in degrees, derived from the design value of the digital model;

[0032] μ is the pressure sensitivity coefficient;

[0033] The pressure gradient is expressed in Pa / m.

[0034] Formula source: Combining particle orientation theory in fluid mechanics with carbon fiber slip characteristics, used for dynamic prediction of fiber orientation in multi-scale simulation step S2;

[0035] Formula for correction factor of quantum tunneling effect:

[0036]

[0037] Where: λ is the corrected fiber interface stress transfer coefficient;

[0038] λ0 is the transfer coefficient without quantum effects;

[0039] α is the quantum tunneling coefficient (experimental fit value);

[0040] m is the fiber mass density (kg / m³) 3 );

[0041] ΔE is the energy barrier at the fiber-resin interface (eV);

[0042] Formula source: Based on the quantum mechanical tunneling effect theory, used to correct the deviation in the calculation of interface stress in the macroscopic simulation in step S2;

[0043] S22. Application and connection of simulation data: The simulation results not only capture the stress transmission law of the fiber interface, but also output key parameters such as fiber orientation and resin flow rate, providing data support for step S3. Through dynamic interaction with the preceding model, it is ensured that the simulation data is highly matched with the actual material state.

[0044] Preferably, the specific steps for optimizing the neuromorphic chip driving parameters in step S3 are as follows:

[0045] S31. Construction of the parameter optimization system: Receive key parameters output by quantum simulation, construct an adaptive optimization system with neuromorphic computing as the core. The system transforms historical molding data into synaptic weights of a spiking neural network, and establishes the correlation between process parameters and molding quality by simulating the associative learning mechanism of the human brain.

[0046] Reinforcement learning reward function:

[0047] R = w1Δσ + w2Δt - w3ΔC

[0048] In the formula: R is the reward value (used for parameter optimization iteration);

[0049] w1, w2, and w3 are weighting coefficients (corresponding to performance, efficiency, and cost, respectively);

[0050] Δσ is the deviation (MPa) between the actual strength and the target strength;

[0051] Δt is the molding time deviation (s);

[0052] ΔC represents the production cost deviation (in yuan);

[0053] Formula source: Based on the reinforcement learning Q-Learning algorithm extension, used for constructing the objective function for process parameter optimization in step S3;

[0054] S32. Practical integration of prediction results: The system can identify potential problems such as fiber accumulation at the corners of parts in advance, and autonomously generate optimization strategies such as pressure gradient adjustment. Its output of precise parameter schemes directly provides operational basis for magneto-electric fusion molding control, realizing an intelligent transition from simulation analysis to actual production parameter setting.

[0055] Preferably, the specific steps of the magnetoelectric fusion adaptive shaping control in step S4 are as follows:

[0056] S41. Synergistic mechanism of magnetoelectric control: The magnetoelectric composite drive device is activated according to the optimized parameters. Through the magnetic field response characteristics of magnetostrictive materials and the electric field vibration effect of piezoelectric crystals, a synergistic control mode of "magnetically controlled fiber orientation + electrically controlled resin flow" is formed to solve the molding problem of complex structural regions.

[0057] Magnetically controlled fiber orientation equation:

[0058]

[0059] Where: M is the magnetization intensity (A / m, reflecting the magnetic response of the fiber);

[0060] χ represents magnetic susceptibility (related to the carbon content of carbon fiber);

[0061] H is the magnetic field strength (A / m);

[0062] κ is the orientation coupling coefficient;

[0063] The fiber orientation gradient (° / m);

[0064] Formula source: Combining magnetostriction effect and materials mechanics, used for dynamic control calculation of magnetic field strength and fiber orientation in step S4;

[0065] S42. Dynamic control connection logic: During the control process, the dynamic data of quantum computing simulation is called in real time. By comparing the actual forming state with the simulation expectation, the magnetic field strength and electric field frequency are dynamically adjusted. This real-time interaction with the simulation system ensures that the forming process strictly follows the optimized path.

[0066] Preferably, the specific steps for controlling the self-healing of photothermal responsive defects in step S5 are as follows:

[0067] S51. Digital foundation for defect monitoring: During the magneto-electric fusion molding process, the digital twin system synchronously maps the molding state of the parts. By fusing data from multiple sensors, it accurately locates the position and type of defects such as fiber breakage and resin bubbles, providing real-time digital basis for repair decisions.

[0068] Defect identification criterion formula:

[0069]

[0070] In the formula: δ is the defect judgment coefficient (a defect is judged when ≥1);

[0071] ε 实测 The strain value (με) acquired by the sensor;

[0072] ε 模拟 The strain value (με) predicted by the digital twin;

[0073] ε 阈值 Maximum allowable strain deviation (μ, set based on material properties);

[0074] Formula source: Based on statistical process control theory, used for the quantitative identification of defects such as fiber breakage and resin bubbles in step S5;

[0075] S52. Closed-loop connection of self-healing repair: For different defects, the photothermal response system is activated: the fiber fracture area is triggered by a specific laser to release nanofibers from microcapsules for bridging, while the bubble area is heated by laser to promote the release of curing accelerator; the repair process is linked with mechanical devices for reinforcement, forming a digital closed loop from defect identification to repair.

[0076] Preferably, the specific steps of the brain-computer interactive virtual verification system in step S6 are as follows:

[0077] S61. Technical Architecture of Virtual Verification: The molding data after defect repair (including fiber orientation, resin distribution, and mechanical performance parameters) and dynamic digital model are mapped in real time to the virtual verification platform (built based on a 3D modeling engine), integrating an EEG signal acquisition module (such as an EEG device) and a force feedback interaction module (such as a tactile glove); R&D personnel output preset performance parameter instructions (such as target strength and stiffness thresholds) through the EEG signal acquisition module, and the platform converts the instructions into virtual molding parameters through a neural signal decoding algorithm, generating a matching virtual solution in real time;

[0078] S62. Application and connection of verification results: In the virtual environment, the physical properties of the parts (such as compressive strength and surface hardness) are simulated through the force feedback interaction module, and the rationality of the process parameters is verified through the virtual-real parameter comparison algorithm. The verification results are output as a quantitative parameter table (including the deviation values ​​of virtual and measured parameters) as the basis for confirming the final process scheme and providing a data benchmark for carbon footprint-performance dual-objective optimization.

[0079] Formula for virtual-real mapping error:

[0080]

[0081] In the formula: e represents the relative error of the virtual-to-real mapping of the virtual verification platform;

[0082] n is the number of sampling points;

[0083] X 虚拟,i For the i-th parameter value of the virtual verification platform, such as strength or size;

[0084] X 物理,i Let be the measured value of the i-th parameter of the physical part;

[0085] Formula source: Based on the root mean square error (RMSE) improvement, used for the consistency assessment of virtual verification and physical entity in step S6.

[0086] Preferably, the specific steps of the carbon footprint-performance dual-objective optimization in step S7 are as follows:

[0087] S71. Model basis for dual-objective optimization: Based on the process scheme verified by the virtual verification platform, a dynamic coupling model of the whole life cycle carbon footprint and part performance is constructed. Carbon emission data such as raw material production and equipment energy consumption are transformed into quantifiable performance constraints. The mechanical performance and environmental protection indicators are balanced through a multi-objective evolutionary algorithm.

[0088] Carbon footprint-performance coupling equation:

[0089] C = γE + βt + ξ, where E ≥ E min

[0090] In the formula: C represents the total life cycle carbon emissions (kgCO2);

[0091] γ is the carbon emission factor per unit intensity (kgCO2 / MPa);

[0092] E represents the mechanical strength of the part (MPa);

[0093] β is the energy consumption coefficient per unit time (kgCO2 / s);

[0094] t is the molding time (s);

[0095] ξ represents the carbon emissions based on raw materials (kgCO2);

[0096] E min Minimum strength requirement (MPa);

[0097] Formula source: Based on the theoretical extension of Life Cycle Assessment (LCA), used for the synergistic optimization constraint calculation of carbon emissions and performance in step S7;

[0098] S72. Practical value of optimized closed loop: Under the premise of ensuring that the strength of the parts meets the standards, the system automatically adjusts parameters such as curing temperature curve to reduce energy consumption, and its output sustainable process benchmark directly guides subsequent batch production.

[0099] This invention does not simply superimpose existing technologies, but achieves breakthroughs through the following innovative combinations: forming a computing power closed loop by combining multi-scale simulation of quantum computing with real-time optimization of neuromorphic chips, thus resolving the contradiction between accuracy and efficiency in traditional computing; combining photothermal self-healing systems with digital twins to achieve digital real-time response for defect repair, overcoming the lag of traditional manual inspection and repair; and for the first time realizing full life-cycle management of composite material molding from performance compliance to sustainable development through the linkage of virtual verification platform verification and carbon footprint optimization.

[0100] The beneficial effects of this invention are as follows:

[0101] 1. This invention uses material property-driven digital modeling and multi-scale simulation to accurately capture the influence of carbon fiber anisotropy and weaving structure on molding, solving the quality fluctuation problem caused by the reliance on manual experience in traditional processes; combined with real-time closed-loop control and a defect self-healing system, it can dynamically correct molding deviations, avoid quality defects caused by improper fiber orientation, insufficient resin impregnation, etc., ensure the consistency of quality of parts in different batches, and meet the stringent requirements of high-end manufacturing fields for product performance stability.

[0102] 2. This invention utilizes parameter prediction and optimization of neuromorphic chips and verification technology through a virtual verification platform. This process breaks through the limitations of traditional trial-and-error methods, allowing for the rapid determination of the optimal solution without repeated parameter adjustments. The digital thread and closed-loop feedback mechanism enable real-time linkage between molding data and process optimization, avoiding efficiency losses caused by the disconnect between molding and testing processes, significantly shortening the process development cycle, and providing technical support for the rapid iteration of complex structural parts.

[0103] 3. This invention utilizes a dual-objective dynamic optimization model of carbon footprint and performance. While ensuring the mechanical properties of the parts, this process can selectively adjust parameters such as curing temperature and pressure to reduce energy consumption and material waste. The full life-cycle digital management effectively improves material utilization and reduces carbon emissions during the production process. While improving economic efficiency, it aligns with the development trend of green manufacturing and provides a new path for the sustainable development of the industry. Attached Figure Description

[0104] Figure 1 This is a flowchart of the digital molding process for carbon fiber composite parts according to the present invention. Detailed Implementation

[0105] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0106] like Figure 1 As shown in the figure, this invention provides a digital forming process for carbon fiber composite parts. The specific steps of this process are as follows:

[0107] S1. Bio-inspired dynamic digital modeling: Simulate the interaction between carbon fiber and resin through artificial neural networks, construct a model that dynamically adjusts with humidity and prepreg storage time, embed fiber growth trajectory algorithm, integrate geometric and material information, and provide a realistic digital foundation for step S2.

[0108] S2. Quantum computing-assisted multi-field coupling simulation: Based on the dynamic model in step S1, multi-scale multi-field coupling simulation is carried out using the parallel capabilities of quantum computing. Quantum characteristics are incorporated to analyze field changes, capture fiber stress patterns, and output key parameters.

[0109] S3. Neuromorphic chip driving parameter optimization: Receive quantum simulation parameters, construct the system with neuromorphic computing, convert historical data into synaptic weights, predict the difficulties in forming and generate strategies, and provide parameters for step S4.

[0110] S4. Magnetoelectric Fusion Adaptive Molding Control: The magnetoelectric device is activated according to the optimized parameters. Through the collaboration of magnetostrictive material and piezoelectric crystal, dual closed-loop control is achieved, and simulation data is exchanged in real time to ensure that molding proceeds according to the optimized path.

[0111] S5. Photothermal Response Type Defect Self-Healing Control: In magnetoelectric molding, the digital twin system monitors and locates the position and type of fiber defects, activates the photothermal system to trigger microcapsule repair, and links mechanical reinforcement to form a digital closed loop for defect repair.

[0112] S6, Brain-Computer Interactive Virtual Verification System: Maps the molded data and digital model after defect repair to the virtual verification platform. It collects neural electrical signals through brain-computer interface devices and converts them into quantitative parameters. Combined with force feedback interaction module devices, it realizes physical characteristic simulation. Researchers can transmit performance expectations through preset parameter commands, and the platform generates corresponding virtual molding schemes in real time based on input parameters.

[0113] S7. Carbon Footprint-Performance Dual-Objective Optimization: Through the verification scheme of the virtual verification platform, a carbon footprint and performance coupled model is constructed to optimize the parameter balance between performance and energy consumption, forming a synergistic closed loop and providing a sustainable benchmark for subsequent production.

[0114] The specific steps of bio-inspired dynamic digital modeling in step S1 are as follows:

[0115] S11. Technical basis of dynamic modeling: The interface mechanism between carbon fiber and resin is simulated by artificial neural network, focusing on capturing the mutual influence between the two under different environments. The model can respond to variables such as humidity fluctuation and prepreg storage time in real time. By embedding the carbon fiber bundle "growth trajectory" algorithm, the fiber weaving method and material distribution law are accurately mapped to construct a dynamic digital carrier that integrates geometric features and material properties.

[0116] The carbon fiber bundle growth trajectory algorithm is as follows: starting from the initial fiber arrangement of the prepreg, a trajectory equation is constructed based on the fiber weaving angle (θ), the prepreg storage time (t), and the humidity (H).

[0117] L(t,H)=L0·exp(-k1t+k2H)·sin(θ)

[0118] Where L0 is the initial fiber bundle length, k1 is the time decay coefficient (0.02~0.05 / day), and k2 is the humidity sensitivity coefficient (0.01~0.03 / %RH); the algorithm achieves dynamic trajectory correction by coupling the hidden layer nodes of the artificial neural network with the carbon fiber-resin interface interaction model (input is humidity and storage time, output is interface binding energy).

[0119] The artificial neural network uses a 3-layer feedforward network: the input layer contains 3 neurons (humidity, prepreg storage time, initial fiber orientation), the hidden layer contains 12 neurons (activation function is ReLU), and the output layer contains 2 neurons (interfacial binding energy, fiber bundle trajectory correction coefficient); the network is trained using the Adam optimizer with a learning rate of 0.001, 5000 iterations, and the loss function is mean squared error (MSE).

[0120] The training data for the artificial neural network came from 300 sets of carbon fiber-resin interface interaction experiments (covering the full range of working conditions with humidity of 20% to 80% RH and prepreg storage time of 1 to 30 days). Each set of experiments contained 50 sampling points. The data was used for model training after outliers were removed by the 3σ criterion. The ratio of training set to validation set was 7:3.

[0121] The anisotropic constitutive equation for carbon fiber:

[0122] σ ij =C ijkl ε kl

[0123] In the formula: σ ij Let i be the stress tensor (i,j = 1, 2, 3, representing the stress components in three-dimensional space);

[0124] C ijkl The stiffness tensor reflects the anisotropy of carbon fiber and is related to the fiber weaving angle and the number of layers.

[0125] ε kl For strain tensor (k,|=1,2,3, representing the three-dimensional spatial strain components);

[0126] Formula source: Based on the extension of the classic Hooke's law and combined with the characteristics of carbon fiber braided structures, it is used for the coupling modeling of material properties and digital model in step S1.

[0127] S12. The model's connecting value: This model integrates the structural parameters of the parts with the dynamic characteristics of the materials, providing high-precision initial data for subsequent quantum computing simulations. Its dynamic adjustment capability ensures that the digital model always fits the actual state of the materials.

[0128] The specific steps of quantum computing-assisted multi-field coupling simulation in step S2 are as follows:

[0129] S21. Technical Implementation of Multi-Scale Simulation: Based on a bio-inspired dynamic model, leveraging the parallel processing advantages of quantum computing, it breaks through the limitations of traditional computing power and realizes cross-scale analysis from the molecular scale to the macroscopic structure; it incorporates the quantum tunneling effect of carbon fiber and the quantum entangled state of resin molecules into the calculation, accurately reproducing the dynamic coupling process of multiple physical fields such as temperature field and pressure field;

[0130] Quantum computing employs the Variable Quantum Eigensolver (VQE) algorithm, implemented on the IBM Quantum Experience platform (16 qubits). In the cross-scale analysis, the molecular scale is defined as ≤10nm (simulating the movement of resin molecular chains), the macroscopic structure is defined as ≥100mm (simulating the overall stress distribution of the part), and the intermediate scale (10nm~100mm) is coupled through a multi-scale bridging algorithm (such as homogenization method).

[0131] The homogenization method uses a cubic representative volume element (RVE) with dimensions of 100 μm × 100 μm × 100 μm (containing 500 carbon fibers). Periodic boundary conditions are used to transform molecular-scale calculations (such as resin viscosity) into macroscopic input parameters. The transformation formula is as follows:

[0132] μ 宏观 =0.8μ 分子 +0.2σ 界面 (μ is viscosity, σ is interfacial stress);

[0133] The specific method for analyzing field changes using quantum properties is as follows: the quantum tunneling effect of carbon fiber is incorporated into the fiber interface stress calculation through a correction coefficient λ (see the formula for the correction coefficient of the quantum tunneling effect), where the quantum tunneling coefficient α ranges from 0.01 to 0.03 eV. -1 / 2 ·m 3 / 2 / kg 1 / 2(Based on experimental fitting of T300 carbon fiber); the bond energy changes of the molecular quantum entangled states of resin molecules were calculated using density functional theory (DFT), and correlated with the temperature field gradient. Establish entangled state energy (E ent )and Mapping relationship:

[0134]

[0135] Fiber orientation evolution equation:

[0136]

[0137] In the formula: θ is the instantaneous orientation angle of the fiber (°);

[0138] t represents time (s);

[0139] k is the orientation adjustment coefficient, which is related to fiber stiffness;

[0140] θ0 is the target orientation angle, in degrees, derived from the design value of the digital model;

[0141] μ is the pressure sensitivity coefficient;

[0142] The pressure gradient is expressed in Pa / m.

[0143] Formula source: Combining particle orientation theory in fluid mechanics with carbon fiber slip characteristics, used for dynamic prediction of fiber orientation in multi-scale simulation step S2;

[0144] Formula for correction factor of quantum tunneling effect:

[0145]

[0146] Where: λ is the corrected fiber interface stress transfer coefficient;

[0147] λ0 is the transfer coefficient without quantum effects;

[0148] α is the quantum tunneling coefficient (experimental fit value);

[0149] m is the fiber mass density (kg / m³) 3 );

[0150] ΔE is the energy barrier at the fiber-resin interface (eV);

[0151] Formula source: Based on the quantum mechanical tunneling effect theory, used to correct the deviation in the calculation of interface stress in the macroscopic simulation in step S2;

[0152] S22. Application and connection of simulation data: The simulation results not only capture the stress transmission law of the fiber interface, but also output key parameters such as fiber orientation and resin flow rate, providing data support for step S3. Through dynamic interaction with the preceding model, it is ensured that the simulation data is highly matched with the actual material state.

[0153] The specific steps for optimizing the neuromorphic chip driving parameters in step S3 are as follows:

[0154] S31. Construction of the parameter optimization system: Receive key parameters output by quantum simulation, construct an adaptive optimization system with neuromorphic computing as the core. The system transforms historical molding data into synaptic weights of a spiking neural network, and establishes the correlation between process parameters and molding quality by simulating the associative learning mechanism of the human brain.

[0155] The spiking neural network adopts a two-layer structure: an input layer (receiving quantum simulation parameters: fiber orientation, resin flow rate, etc., 5 dimensions) and an output layer (outputting optimization parameters: pressure, temperature, etc., 3 dimensions), connected by a fully connected method with a synaptic delay of 5ms; the pulse encoding adopts rate encoding (pulse frequency is positively correlated with parameter values, ranging from 10 to 100Hz); the historical data includes nearly 500 batches of molding pressure (0.1 to 10MPa), curing temperature (80 to 180℃), defect rate (0 to 5%), and 12 other dimensions;

[0156] The training of the spiking neural network uses the fluctuation amplitude of the reinforcement learning reward function R ≤ 5% as the convergence condition. The synaptic weight matrix is ​​updated every 100 iterations during the iteration process, and an early stopping strategy is adopted (training is terminated if R does not improve after 20 consecutive iterations).

[0157] The rule for converting historical data into synaptic weights is as follows: using nearly 500 batches of molding data (including pressure P, temperature T, and defect rate D) as samples, the data is standardized using the least squares method and then mapped to the synaptic weight matrix W of the spiking neural network, where:

[0158] W ij =0.8·P i +0.2·T j -0.5·D ij (i,j are neural element references)

[0159] The neuromorphic optimization system uses the Intel Loihi chip architecture, which includes an input layer (receiving quantum simulation parameters), a hidden layer (128 neurons), and an output layer (outputting optimized pressure and temperature parameters), and encodes data through pulse frequency;

[0160] Reinforcement learning reward function:

[0161] R = w1Δσ + w2Δt - w3ΔC

[0162] In the formula: R is the reward value (used for parameter optimization iteration);

[0163] w1, w2, and w3 are weighting coefficients (corresponding to performance, efficiency, and cost, respectively);

[0164] Δσ is the deviation (MPa) between the actual strength and the target strength;

[0165] Δt is the molding time deviation (s);

[0166] ΔC represents the production cost deviation (in yuan);

[0167] Formula source: Based on the reinforcement learning Q-Learning algorithm extension, used for constructing the objective function for process parameter optimization in step S3;

[0168] S32. Practical integration of prediction results: The system can identify potential problems such as fiber accumulation at the corners of parts in advance, and autonomously generate optimization strategies such as pressure gradient adjustment. Its output of precise parameter schemes directly provides operational basis for magneto-electric fusion molding control, realizing an intelligent transition from simulation analysis to actual production parameter setting.

[0169] The specific steps of the magnetoelectric fusion adaptive shaping control in step S4 are as follows:

[0170] S41. Synergistic mechanism of magnetoelectric control: The magnetoelectric composite drive device is activated according to the optimized parameters. Through the magnetic field response characteristics of magnetostrictive materials and the electric field vibration effect of piezoelectric crystals, a synergistic control mode of "magnetically controlled fiber orientation + electrically controlled resin flow" is formed to solve the molding problem of complex structural regions.

[0171] The magnetoelectric composite drive device includes: a magnetostrictive material (Terfenol-D, length 50–100 mm, diameter 10–20 mm), a piezoelectric crystal (PZT-5H, thickness 5–10 mm), and a control system; the magnetic field strength H is adjustable from 100–500 A / m, and the electric field frequency is adjustable from 1–10 kHz. Dual closed-loop control is achieved through a PID controller, where the proportional coefficient K of the magnetic field loop is... p =0.5, integral coefficient K i =0.1; Electric field ring proportionality coefficient K p =0.3, integral coefficient K i =0.05;

[0172] The dynamic response indicators of the PID controller are: magnetic field loop overshoot ≤5%, settling time ≤2s; electric field loop overshoot ≤3%, settling time ≤1s; the synchronous control accuracy of the magnetoelectric device is ±0.5ms (time error between magnetic field strength and electric field frequency), and real-time scheduling is achieved through STM32H743 microcontroller.

[0173] STM32H743 microcontroller configuration: 480MHz main frequency, 16-bit AD sampling accuracy, 1MHz sampling rate, parallel acquisition and processing of magnetic field / electric field signals through DMA direct memory access, and control command output delay ≤10μs;

[0174] Magnetically controlled fiber orientation equation:

[0175]

[0176] Where: M is the magnetization intensity (A / m, reflecting the magnetic response of the fiber);

[0177] χ represents magnetic susceptibility (related to the carbon content of carbon fiber);

[0178] H is the magnetic field strength (A / m);

[0179] κ is the orientation coupling coefficient;

[0180] The fiber orientation gradient (° / m);

[0181] Formula source: Combining magnetostriction effect and materials mechanics, used for dynamic control calculation of magnetic field strength and fiber orientation in step S4;

[0182] S42. Dynamic control connection logic: During the control process, the dynamic data of quantum computing simulation is called in real time. By comparing the actual forming state with the simulation expectation, the magnetic field strength and electric field frequency are dynamically adjusted. This real-time interaction with the simulation system ensures that the forming process strictly follows the optimized path.

[0183] The specific steps of photothermal response defect self-healing regulation in step S5 are as follows:

[0184] S51. Digital foundation for defect monitoring: During the magneto-electric fusion molding process, the digital twin system synchronously maps the molding state of the parts. By fusing data from multiple sensors, it accurately locates the position and type of defects such as fiber breakage and resin bubbles, providing real-time digital basis for repair decisions.

[0185] The multiple sensors include: a distributed fiber optic sensor (sampling frequency 1kHz, monitoring strain), an infrared thermal imager (resolution 640×512, monitoring temperature field), and an ultrasonic flaw detector (frequency 5MHz, locating bubbles); data fusion adopts the federated Kalman filter algorithm, with a fusion period of 100ms and a defect location error ≤0.5mm.

[0186] Sensor calibration uses a timestamp synchronization method: the fiber optic sensor and the infrared thermal imager are synchronized via GPS (synchronization error ≤1ms). Spatial calibration is based on the origin of the part coordinate system, and the position offset of each sensor is calibrated by a laser tracker (accuracy ±0.01mm) (x, y, z direction correction values ​​are 0.1mm, 0.1mm, and 0.05mm, respectively).

[0187] Defect identification criterion formula:

[0188]

[0189] In the formula: δ is the defect judgment coefficient (a defect is judged when ≥1);

[0190] ε 实测 The strain value (με) acquired by the sensor;

[0191] ε 模拟 The strain value (με) predicted by the digital twin;

[0192] ε 阈值 Maximum allowable strain deviation (μ, set based on material properties);

[0193] Formula source: Based on statistical process control theory, used for the quantitative identification of defects such as fiber breakage and resin bubbles in step S5;

[0194] S52. Closed-loop connection of self-healing repair: For different defects, the photothermal response system is activated: the fiber fracture area is triggered by a specific laser to release nanofibers from microcapsules for bridging, while the bubble area is heated by laser to promote the release of curing accelerator; the repair process is linked with mechanical devices for reinforcement, forming a digital closed loop from defect identification to repair.

[0195] The triggering condition for the photothermal system is as follows: when the digital twin system detects a defect judgment coefficient δ≥1, the laser control module is immediately activated, and parameters are automatically switched according to the defect type (fiber fracture: wavelength 1064nm, power 5-10W, spot diameter 0.5mm; resin bubble: wavelength 1064nm, power 2-5W, spot diameter 1mm). The laser action time is calculated from the defect area (action time = defect area / mm). 2 ×0.2s);

[0196] The microcapsules are urea-formaldehyde resin-coated carbon nanofibers (50-100 nm in diameter) with a wall thickness of 5-10 μm. The photothermal system uses a 1064 nm wavelength fiber laser, with an output power of 5-10 W (acting time 1-2 s) for the fiber fracture area and 2-5 W (acting time 3-5 s) for the resin bubble area. The mechanical device is a servo pressure head (pressure range 0.5-2 MPa), which automatically moves to the corresponding area according to the defect location to apply reinforcing pressure.

[0197] The specific steps of the brain-computer interactive virtual verification system in step S6 are as follows:

[0198] S61. Technical Architecture of Virtual Verification: The molding data after defect repair (including fiber orientation, resin distribution, and mechanical performance parameters) and dynamic digital model are mapped in real time to the virtual verification platform (built based on a 3D modeling engine), integrating an EEG signal acquisition module (such as an EEG device) and a force feedback interaction module (such as a tactile glove); R&D personnel output preset performance parameter instructions (such as target strength and stiffness thresholds) through the EEG signal acquisition module, and the platform converts the instructions into virtual molding parameters through a neural signal decoding algorithm, generating a matching virtual solution in real time;

[0199] The brain-computer interface uses EEG signal acquisition equipment (such as NeuroScanSynAmps2) to identify the performance expectations of R&D personnel (such as target values ​​for strength and stiffness) through P300 potential; the force feedback interaction module is a force feedback glove (such as GeomagicTouchX), with a feedback force range of 0-5N; the virtual verification platform is built on Unreal Engine 5, including a physics engine (simulating the mechanical properties of parts) and a rendering engine (generating virtual molding schemes in real time), with a data mapping frequency of 10Hz;

[0200] The total latency of the virtual verification platform is ≤50ms (brain-computer interface signal acquisition → virtual scheme generation → tactile feedback output); the simulation accuracy of virtual physical characteristics is: strength error ≤3%, stiffness error ≤2%, and real-time rendering is achieved through GPU acceleration (NVIDIA A100).

[0201] The virtual physical property simulation uses an 8-node hexahedral reduced integral element (C3D8R) with an element size of 5mm×5mm×5mm. Material properties are mapped in real time through the dynamic model in step S1, and the boundary conditions are consistent with the physical forming environment (such as constraint method and loading rate).

[0202] S62. Application and connection of verification results: In the virtual environment, the physical properties of the parts (such as compressive strength and surface hardness) are simulated through the force feedback interaction module, and the rationality of the process parameters is verified through the virtual-real parameter comparison algorithm. The verification results are output as a quantitative parameter table (including the deviation values ​​of virtual and measured parameters) as the basis for confirming the final process scheme and providing a data benchmark for carbon footprint-performance dual-objective optimization.

[0203] Formula for virtual-real mapping error:

[0204]

[0205] In the formula: e represents the relative error of the virtual-to-real mapping of the virtual verification platform;

[0206] n is the number of sampling points;

[0207] X 虚拟,i For the i-th parameter value of the virtual verification platform, such as strength or size;

[0208] X 物理,i Let be the measured value of the i-th parameter of the physical part;

[0209] Formula source: Based on the root mean square error (RMSE) improvement, used for the consistency assessment of virtual verification and physical entity in step S6.

[0210] The specific steps of the carbon footprint-performance dual-objective optimization in step S7 are as follows:

[0211] S71. Model basis for dual-objective optimization: Based on the process scheme verified by the virtual verification platform, a dynamic coupling model of the whole life cycle carbon footprint and part performance is constructed. Carbon emission data such as raw material production and equipment energy consumption are transformed into quantifiable performance constraints. The mechanical performance and environmental protection indicators are balanced through a multi-objective evolutionary algorithm.

[0212] The multi-objective evolutionary algorithm adopts the NSGA-II algorithm, with a population size of 100, a crossover probability of 0.8, a mutation probability of 0.05, and 50 iterations. The boundary for calculating the carbon footprint of the entire life cycle is 'raw material production (carbon fiber, resin) - molding and processing - waste recycling'. The carbon emission data of the raw material production stage comes from the Ecoinvent database (version 3.8), and the carbon emission of equipment energy consumption is calculated as CO2 emission = power consumption (kWh) × 0.62 (kgCO2 / kWh).

[0213] The NSGA-II algorithm uses simulated binary crossover (SBX) for the crossover operation with a crossover distribution exponent of 20; it uses polynomial mutation for the mutation operation with a mutation distribution exponent of 20; the carbon emission factor in the waste recycling stage is: carbon fiber recycling energy consumption corresponds to CO2 emissions of 0.15 kg / kg (based on pyrolysis recycling process).

[0214] Carbon footprint-performance coupling equation:

[0215] C = γE + βt + ξ, where E ≥ E min

[0216] In the formula: C represents the total life cycle carbon emissions (kgCO2);

[0217] γ is the carbon emission factor per unit intensity (kgCO2 / MPa);

[0218] E represents the mechanical strength of the part (MPa);

[0219] β is the energy consumption coefficient per unit time (kgCO2 / s);

[0220] t is the molding time (s);

[0221] ξ represents the carbon emissions based on raw materials (kgCO2);

[0222] E min Minimum strength requirement (MPa);

[0223] Formula source: Based on the theoretical extension of Life Cycle Assessment (LCA), used for the synergistic optimization constraint calculation of carbon emissions and performance in step S7;

[0224] S72. Practical value of optimized closed loop: Under the premise of ensuring that the strength of the parts meets the standards, the system automatically adjusts parameters such as curing temperature curve to reduce energy consumption, and its output sustainable process benchmark directly guides subsequent batch production.

[0225] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0226] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A digital forming process for carbon fiber composite parts, characterized in that: The specific steps of this process are as follows: S1. Bio-inspired dynamic digital modeling: Simulates the interaction mechanism between carbon fiber and resin through artificial neural networks, constructs a model that dynamically adjusts with humidity and prepreg storage time, and embeds a carbon fiber bundle growth trajectory algorithm. S2. Quantum computing-assisted multi-field coupling simulation: Based on the dynamic model in step S1, multi-scale multi-field coupling simulation is carried out using the parallel capabilities of quantum computing. Quantum characteristics are incorporated to analyze field changes, capture fiber stress patterns, and output key parameters. S3. Neuromorphic chip-driven parameter optimization: Upon receiving quantum simulation parameters, an optimization system is constructed using neuromorphic computing. Historical data is converted into synaptic weights, forming difficulties are predicted and strategies are generated, and optimized parameters are output. S4. Magnetoelectric Fusion Adaptive Molding Control: The magnetoelectric device is activated according to the optimized parameters. Through the magnetic field response characteristics of the magnetostrictive material and the electric field vibration effect of the piezoelectric crystal, dual closed-loop control is achieved, and simulation data is interactive in real time. S5. Photothermal Response Defect Self-Healing Control: In magnetoelectric fusion molding, the digital twin system monitors and locates the position and type of fiber defects, activates the photothermal system to trigger microcapsule repair, and links mechanical devices for reinforcement. S6. Brain-computer interactive virtual verification system: The molded data and digital model after defect repair are mapped to the virtual verification platform. The neural electrical signals are converted into quantitative parameters through the brain-computer interface, and the platform generates the corresponding virtual molding scheme in real time. S7. Carbon Footprint-Performance Dual-Objective Optimization: Through the verification scheme of the virtual verification platform, a dynamic coupling model of carbon footprint and part performance is constructed, and the curing temperature curve parameters are automatically adjusted to reduce energy consumption.

2. The digital forming process for carbon fiber composite parts according to claim 1, characterized in that: The specific steps of the bio-inspired dynamic digital modeling in step S1 are as follows: S11. Technical basis of dynamic modeling: The interface mechanism between carbon fiber and resin is simulated by artificial neural network to capture the mutual influence between the two under different environments. The model can respond in real time to humidity fluctuations and prepreg storage time variables. By embedding carbon fiber bundle growth trajectory algorithm, the fiber weaving method and material distribution law are mapped. S12. The value of the model in connecting different parts: This model integrates the structural parameters of the parts with the dynamic characteristics of the materials. Its dynamic adjustment capability ensures that the digital model always fits the actual state of the materials.

3. The digital forming process for carbon fiber composite parts according to claim 1, characterized in that: The specific steps of quantum computing-assisted multi-field coupling simulation in step S2 are as follows: S21. Technical Implementation of Multi-Scale Simulation: Based on a bio-inspired dynamic model, the parallel processing advantage of quantum computing is utilized to achieve cross-scale analysis from the molecular scale to the macroscopic structure; the quantum tunneling effect of carbon fiber and the quantum entangled state of resin molecules are incorporated into the calculation to restore the dynamic coupling process of the physical field. S22. Application and integration of simulation data: The simulation results can capture the stress transmission law of the fiber interface and also output key parameters such as fiber orientation and resin flow rate.

4. The digital forming process for carbon fiber composite parts according to claim 1, characterized in that: The specific steps for optimizing the neuromorphic chip driving parameters in step S3 are as follows: S31. Construction of the parameter optimization system: Receive key parameters output by quantum simulation, construct an adaptive optimization system with neuromorphic computing as the core. The system transforms historical molding data into synaptic weights of a spiking neural network, and establishes the correlation between process parameters and molding quality by simulating the associative learning mechanism of the human brain. S32. Practical application of prediction results: The system can identify potential problems such as fiber accumulation at the corners of parts in advance and generate pressure gradient adjustment and optimization strategies autonomously.

5. The digital forming process for carbon fiber composite parts according to claim 1, characterized in that: The specific steps of the magnetoelectric fusion adaptive shaping control in step S4 are as follows: S41. Synergistic mechanism of magnetoelectric control: The magnetoelectric composite drive device is activated according to the optimized parameters. Through the magnetic field response characteristics of magnetostrictive materials and the electric field vibration effect of piezoelectric crystals, a synergistic control mode of magnetic fiber orientation + electro-controlled resin flow is formed. S42. Dynamic control connection logic: During the control process, the dynamic data of quantum computing simulation is called in real time. By comparing the actual forming state with the simulation expectation, the magnetic field strength and electric field frequency are dynamically adjusted.

6. The digital forming process for carbon fiber composite parts according to claim 1, characterized in that: The specific steps for controlling the self-healing of photothermal responsive defects in step S5 are as follows: S51. Digital foundation for defect monitoring: During the magneto-electric fusion molding process, the digital twin system synchronously maps the molding state of the part and locates the location and type of fiber breakage and resin bubble defects by fusing data from multiple sensors. S52. Closed-loop connection of self-healing repair: For different defects, the photothermal response system is activated: the fiber fracture area is triggered by a specific laser to release nanofibers from microcapsules for bridging, while the bubble area is heated by laser to promote the release of curing accelerator; the repair process is linked to mechanical devices for reinforcement.

7. The digital forming process for carbon fiber composite parts according to claim 1, characterized in that: The specific steps of the brain-computer interactive virtual verification system in step S6 are as follows: S61. Technical architecture of virtual verification: The formed data after defect repair and dynamic digital model are mapped to the virtual verification platform in real time, and the EEG signal acquisition module and force feedback interaction module are integrated; the R&D personnel output preset performance parameter instructions through the EEG signal acquisition module, and the platform converts the instructions into virtual forming parameters through the neural signal decoding algorithm, and generates matching virtual solutions in real time. S62. Application and connection of verification results: In the virtual environment, the physical characteristics of the parts are simulated through the force feedback interaction module, and the rationality of the process parameters is verified through the virtual-real parameter comparison algorithm; the verification results are output as a quantitative parameter table.

8. The digital forming process for carbon fiber composite parts according to claim 1, characterized in that: The specific steps of the carbon footprint-performance dual-objective optimization in step S7 are as follows: S71. Model basis for dual-objective optimization: Based on the process scheme verified by the virtual verification platform, a dynamic coupling model of the whole life cycle carbon footprint and part performance is constructed. The data of raw material production, equipment energy consumption and carbon emission are transformed into quantifiable performance constraints. The mechanical performance and environmental protection indicators are balanced through a multi-objective evolutionary algorithm. S72. Practical value of optimized closed loop: The system automatically adjusts curing temperature curve parameters to reduce energy consumption while ensuring that the strength of the parts meets the standards.

Images