Thermo-mechanical coupling carbon fiber composite material intelligent forming system and equipment

Abstract

The application relates to the technical field of high-pressure resin transfer molding (HP-RTM), in particular to a thermal-mechanical coupling carbon fiber composite material intelligent forming system and equipment, which comprises the following steps: collecting real-time temperature field and strain field data by using a partition temperature control mold, obtaining thermal-mechanical coupling state parameters through signal filtering and abnormal value elimination processing, evaluating a forming quality index in combination with ultrasonic probe array detection of internal defect distribution, dynamically compensating process parameters according to the quality index, and constructing a deep learning model to optimize the forming process. The application can realize real-time monitoring and closed-loop control of multi-physical field coupling, significantly reduce the defect rate, shorten the debugging cycle, improve the production efficiency, and provide an intelligent solution for carbon fiber composite material forming.

Description

Technical Field

[0001] This invention belongs to the field of high pressure resin transfer molding (HP-RTM) technology, specifically a smart molding system and equipment based on thermo-mechanical coupling carbon fiber composite materials. Background Technology

[0002] Carbon fiber composites have been widely used in aerospace, new energy vehicles, and wind power due to their high specific strength, high specific modulus, and excellent corrosion resistance. However, their molding process is complex, especially the thermo-mechanical coupling involved in high-pressure resin transfer molding (HP-RTM), which has a decisive impact on molding quality. Traditional molding processes have significant shortcomings in resin flow control, temperature field uniformity during curing, and multi-physics field coupling optimization, leading to problems such as high defect rates, high trial-and-error costs, and low production efficiency. For example, under high pressure (10-20 MPa), resin is prone to fiber erosion or uneven wetting, resulting in defects such as dry spots and pores; warping and residual stress caused by uneven temperature field during curing further exacerbate the molding difficulty. In addition, traditional processes rely on experience, have long debugging cycles (dozens of trial moldings), and scrap rates can reach 20%, making it difficult to meet the high precision and high consistency requirements of lightweight manufacturing for high-end equipment.

[0003] In existing technologies, a carbon fiber material molding autoclave, disclosed in CN118721797B, improves workpiece quality by using an autoclave design incorporating both an oil bath heating mechanism and an air heating mechanism, combined with a soft silicone encapsulation of the workpiece, to maintain a constant temperature during molding. However, this technical solution primarily relies on static temperature control strategies, lacking the ability to dynamically monitor and control the resin flow field and the thermochemical field of the curing reaction in real time. Furthermore, it fails to consider the impact of mold thermoelastic deformation on cavity geometry, potentially leading to uneven resin flow or excessive curing gradients, resulting in defects such as dry spots and porosity. Another carbon fiber product molding fixture and molding process, disclosed in CN114750435B, achieves shape adaptation and position adjustment during the carbon fiber product molding process through the design of contour-following pressure blocks and positioning components, reducing molding limitations. However, this technology remains experience-driven, lacking system modeling and intelligent decision support for thermo-mechanical coupling effects. It does not incorporate real-time sensor data fusion and digital twin technology, making it unable to dynamically predict the resin flow front and curing distribution, and difficult to cope with sudden anomalies (such as sudden increases in injection pressure or temperature fluctuations). Furthermore, this fixture has limited adaptability to complex structural components, making it difficult to meet the needs of lightweight manufacturing for high-end equipment.

[0004] The aforementioned problems indicate that existing carbon fiber composite molding technologies still have significant shortcomings in areas such as fully coupled modeling of thermo-mechanical coupling effects, real-time monitoring and dynamic control, and intelligent process optimization. Specifically, traditional process optimization relies on finite element simulation and experimental verification, but offline simulation cannot be controlled in real time, and simulation results are disconnected from real production data, making it difficult to dynamically correct process parameters; multi-physics coupling calculations are complex, requiring separate modeling of resin flow (fluid dynamics), curing reaction (thermochemistry), and mold deformation (solid mechanics), making collaborative optimization difficult; manual parameter adjustments are slow to respond, lack intelligent decision-making capabilities, and are unable to cope with sudden anomalies. Therefore, there is an urgent need for an intelligent molding system and control method based on thermo-mechanical coupling, which can achieve real-time simulation and closed-loop control of multi-physics coupling through digital twin technology, reducing defect rates, shortening process debugging cycles, and improving production efficiency, thereby meeting the needs of lightweight and green manufacturing in high-end equipment.

[0005] This invention focuses on the thermo-mechanical coupling effect in the HP-RTM process, addressing the following core issues: (1) Resin flow control is difficult, and fiber preforms are easily washed away or unevenly wetted under high pressure; (2) The curing process is irreversible, and uneven temperature field leads to component warping and residual stress; (3) The process relies on experience, resulting in high trial-and-error costs and low efficiency. By introducing digital twin technology and integrating thermo-mechanical coupling simulation with real-time sensor data, the resin flow front and degree of curing distribution are accurately predicted, and the process parameters are dynamically adjusted based on model prediction control to achieve "one-time molding qualification". At the same time, by accumulating process data to build an AI model, the reliance on human experience is reduced, promoting the transformation of composite material manufacturing from "experience-driven" to "data-driven", which aligns with the development trend of lightweight and green manufacturing of high-end equipment. Summary of the Invention

[0006] This invention provides an intelligent molding system and equipment for carbon fiber composite materials based on thermo-mechanical coupling. Its main purpose is to solve the problems of high defect rate, long debugging cycle and low production efficiency caused by thermo-mechanical coupling in the high pressure resin transfer molding (HP-RTM) process. By using digital twin technology, it realizes real-time optimization and closed-loop control of multi-physics coupling, thereby improving molding quality and reducing production costs.

[0007] To achieve the above objectives, the present invention provides an intelligent molding system for carbon fiber composite materials based on thermo-mechanical coupling, comprising:

[0008] The zoned temperature control mold with built-in fiber optic grating sensor collects real-time temperature and strain field data of the target component. The real-time temperature field data includes: temperature distribution in the resin flow area, temperature distribution on the mold surface, and peak value of curing exothermics.

[0009] The real-time temperature field and strain field data are processed by signal filtering and outlier removal to obtain real-time thermo-mechanical coupling state parameters, which include: resin viscosity change rate, mold thermal expansion and curing gradient.

[0010] The internal defect distribution of the target component is detected by using a pre-constructed ultrasonic probe array. The dynamic deformation of the mold cavity is calculated based on the real-time thermo-mechanical coupling state parameters. The molding quality index of the target component is evaluated by combining the internal defect distribution and the dynamic deformation of the mold cavity.

[0011] Determine whether the molding quality index is lower than a preset quality threshold;

[0012] If the molding quality index is lower than the quality threshold, the injection pressure and mold zone temperature are dynamically compensated based on the molding quality index to obtain optimized process parameters.

[0013] If the molding quality index is not lower than the quality threshold, then the current process parameters will be used as the optimized process parameters.

[0014] Obtain the historical molding dataset of the target component, and set the initial process window of the target component based on the historical molding dataset. The initial process window includes: injection pressure range, mold temperature range, and curing time range.

[0015] Based on the initial process window and the pre-built deep learning model, an initial process optimization model is constructed, and the initial process optimization model is trained using the historical molding dataset to obtain the target process optimization model.

[0016] Based on the optimized process parameters, the molding process of the target component is dynamically adjusted using the target process optimization model.

[0017] Optionally, the step of performing signal filtering and outlier removal processing on the real-time temperature field and strain field data to obtain real-time thermo-mechanical coupling state parameters includes:

[0018] The real-time temperature field and strain field data are subjected to low-pass filtering to obtain preliminary state parameters;

[0019] The signal-to-noise ratio threshold of the fiber Bragg grating sensor is obtained, and outlier removal processing is performed on the preliminary state parameters based on the signal-to-noise ratio threshold to obtain real-time thermo-mechanical coupling state parameters, wherein the real-time thermo-mechanical coupling state parameters are represented as: resin viscosity change rate, mold thermal expansion amount and curing degree gradient.

[0020] Optionally, the step of calculating the dynamic deformation of the mold cavity based on the real-time thermo-mechanical coupling state parameters, and evaluating the molding quality index of the target component by combining the internal defect distribution with the dynamic deformation of the mold cavity, includes:

[0021] Based on the real-time thermo-mechanical coupling state parameters, the dynamic deformation of the mold cavity is calculated using the mold thermoelastic deformation model.

[0022] Combining the internal defect distribution with the dynamic deformation of the mold cavity, the molding quality index of the target component is evaluated using a weighted scoring method, wherein the weight allocation is dynamically adjusted according to the defect type and the magnitude of the deformation.

[0023] Optionally, the step of dynamically compensating for injection pressure and mold zone temperature based on the molding quality index to obtain optimized process parameters includes:

[0024] A dynamic compensation strategy matrix is ​​constructed based on the molding quality index, wherein the dynamic compensation strategy matrix includes the injection pressure increment and the mold zone temperature adjustment value.

[0025] The current process parameters are compensated and calculated based on the dynamic compensation strategy matrix to obtain optimized process parameters, which include: target injection pressure, target zone temperature, and target curing time.

[0026] Optionally, setting the initial process window for the target component based on the historical molding dataset includes:

[0027] Obtain the process boundary conditions of the historical molding dataset, wherein the historical molding dataset is a combination of multiple sets of process parameters with molding quality labels, and the process boundary conditions include: maximum injection pressure, minimum mold temperature and longest curing time.

[0028] The maximum injection pressure, minimum mold temperature, and longest curing time are respectively used as the upper and lower limits of the initial process window to obtain the range of the initial process window.

[0029] Based on the preset safety margin, the optimal combination of process parameters is selected within the initial process window range to obtain the initial process window.

[0030] Optionally, training the initial process optimization model using the historical molding dataset to obtain the target process optimization model includes:

[0031] The historical dataset is randomly divided into a training dataset and a validation dataset.

[0032] Training samples are extracted sequentially from the training dataset;

[0033] The training samples are input into the initial process optimization model to obtain the model prediction results;

[0034] Identify the molding quality label of the training sample and determine whether the model prediction result is consistent with the molding quality label of the training sample;

[0035] If the model prediction result is consistent with the molding quality label of the training sample, then the initial process window is used as the target process window.

[0036] If the model prediction result is inconsistent with the molding quality label of the training sample, the initial process window is adjusted according to the training sample to obtain the target process window.

[0037] Based on the deep learning model and the target process window, a verification process optimization model is constructed.

[0038] Validation samples are extracted sequentially from the validation dataset, and the validation samples are input into the validation process optimization model to obtain the model validation results.

[0039] Identify the molding quality label of the verification sample and determine whether the model verification result is consistent with the molding quality label of the verification sample:

[0040] If the model verification result is inconsistent with the molding quality label of the verification sample, then the inconsistent data in the verification sample is obtained, the inconsistent data is used as the training sample, the verification process optimization model is used as the initial process optimization model, and the step of inputting the training sample into the initial process optimization model is returned.

[0041] If the model verification result is consistent with the molding quality label of the verification sample, then the verification process optimization model will be used as the target process optimization model.

[0042] Optionally, adjusting the initial process window based on training samples to obtain the target process window includes:

[0043] The training samples are represented as follows:

[0044] Wherein, represents the training sample, represents the injection pressure, represents the mold temperature, and represents the curing time;

[0045] The initial process window is represented as follows:

[0046] Wherein, represents the initial injection pressure range, represents the minimum injection pressure, represents the maximum injection pressure, represents the initial mold temperature range, represents the minimum mold temperature, represents the maximum mold temperature, represents the initial curing time range, represents the shortest curing time, and represents the longest curing time;

[0047] Determine the type of difference between the model prediction results and the molding quality label, wherein the type of difference includes: premature curing and insufficient curing;

[0048] If the difference type is premature curing, the initial process window is delayed based on the training samples to obtain the target process window;

[0049] If the difference type is insufficient solidification, the initial process window is accelerated based on the training samples to obtain the target process window.

[0050] Optionally, the difference type between the judgment model prediction result and the molding quality label includes:

[0051] Obtain the target degree of curing and the predicted degree of curing from the model prediction results for the molded quality label;

[0052] Determine whether the target degree of cure is within the predicted degree of cure range;

[0053] If the target degree of cure is not within the predicted degree of cure range, the difference type is set to premature cure;

[0054] If the target degree of cure is within the predicted degree of cure range, the difference type is set to insufficient cure.

[0055] Optionally, the step of dynamically adjusting the molding process of the target component using the target process optimization model based on the optimized process parameters includes:

[0056] The optimized process parameters are input into the target process optimization model to obtain the predicted forming state;

[0057] The predicted molding state is matched with the target process window to obtain molding process adjustment schemes, which include: injection path optimization, zoned temperature distribution optimization, and extension or shortening of curing time.

[0058] To achieve the above objectives, the present invention also provides an intelligent molding system for carbon fiber composite materials based on thermo-mechanical coupling, comprising:

[0059] The data acquisition module is used to acquire real-time temperature field and strain field data of the target component using a partitioned temperature control mold with built-in fiber optic grating sensors. The real-time temperature field data includes: temperature distribution in the resin flow area, temperature distribution on the mold surface, and peak value of curing exothermic reaction. The real-time temperature field and strain field data are processed by signal filtering and outlier removal to obtain real-time thermo-mechanical coupling state parameters, which include: resin viscosity change rate, mold thermal expansion, and curing degree gradient.

[0060] The quality assessment module is used to detect the internal defect distribution of the target component using a pre-built ultrasonic probe array, calculate the dynamic deformation of the mold cavity based on the real-time thermo-mechanical coupling state parameters, and evaluate the molding quality index of the target component by combining the internal defect distribution and the dynamic deformation of the mold cavity.

[0061] The process optimization module is used to determine whether the molding quality index is lower than a preset quality threshold. If the molding quality index is lower than the quality threshold, the injection pressure and mold zone temperature are dynamically compensated according to the molding quality index to obtain optimized process parameters. If the molding quality index is not lower than the quality threshold, the current process parameters are used as optimized process parameters, the historical molding dataset of the target component is obtained, and the initial process window of the target component is set according to the historical molding dataset. The initial process window includes: injection pressure range, mold temperature range, and curing time range.

[0062] The dynamic adjustment module is used to construct an initial process optimization model based on the initial process window and the pre-built deep learning model, train the initial process optimization model using the historical molding dataset to obtain a target process optimization model, and dynamically adjust the molding process of the target component using the target process optimization model based on the optimized process parameters.

[0063] To address the above problems, the present invention also provides an electronic device, the electronic device comprising:

[0064] Memory, storing at least one instruction;

[0065] The processor executes the instructions stored in the memory to implement the aforementioned intelligent molding system and equipment based on thermo-mechanical coupling carbon fiber composite materials.

[0066] To address the aforementioned problems, the present invention also provides a computer-readable storage medium storing at least one instruction, which is executed by a processor in an electronic device to implement the aforementioned intelligent molding system and device based on thermo-mechanical coupling carbon fiber composite materials.

[0067] To address the problems described in the background art, this invention utilizes a zoned temperature-controlled mold with an embedded fiber optic grating sensor to collect real-time temperature and strain field data of the target component, achieving intelligent real-time monitoring of the thermo-mechanical coupling state during the molding process. Signal filtering and outlier removal are performed on the real-time temperature and strain field data to obtain real-time thermo-mechanical coupling state parameters, transforming complex multi-physics data into intuitive state indicators. An ultrasonic probe array is used to detect the internal defect distribution of the target component, completing defect localization and quantitative analysis under non-destructive testing, providing a basis for subsequent molding quality assessment. Based on the real-time thermo-mechanical coupling state parameters, the dynamic deformation of the mold cavity is calculated, obtaining a prediction of the mold cavity's geometric changes based on a thermoelastic deformation model, providing conditions for subsequent process optimization. By combining the internal defect distribution and the dynamic deformation of the mold cavity, the molding quality index of the target component is evaluated, achieving a numerical assessment of molding quality. Since the molding quality index has an acceptable range, i.e., it is not necessary to compensate and adjust all molding quality indices, it is possible to determine whether the molding quality index is below a quality threshold, thus enabling the screening of molding states within the quality threshold. If the molding quality index is lower than the quality threshold, it indicates that the molding state is not within the acceptable range. Therefore, dynamic compensation of injection pressure and mold zone temperature is required based on the molding quality index to obtain optimized process parameters and complete the molding quality optimization of the target component. If the molding quality index is not lower than the quality threshold, it indicates that the molding state is within the acceptable range. By using the current process parameters as the optimized process parameters, unnecessary compensation adjustments can be skipped, accelerating the entire molding process optimization process. By obtaining the historical molding dataset of the target component and setting the initial process window based on the historical molding dataset, the basis for molding process optimization is obtained. Furthermore, based on the initial process window and the deep learning model, an initial process optimization model is constructed, and the initial process optimization model is trained using the historical molding dataset to obtain the target process optimization model. This step utilizes the data-driven characteristics of the deep learning model to construct a model that can intelligently optimize the molding process. This target process optimization model is not affected by human subjective factors and has high accuracy. Finally, by inputting the optimized process parameters into the target process optimization model, the dynamic adjustment of the molding process of the target component is completed. Since this process does not require extensive manual intervention, the consumption of human resources is reduced. Therefore, the present invention can reduce the human resource consumption in the molding process optimization process and improve the accuracy and efficiency of the molding process. Attached Figure Description

[0068] Figure 1 This is a flowchart illustrating an embodiment of the intelligent molding system and equipment for carbon fiber composite materials based on thermo-mechanical coupling provided by the present invention.

[0069] Figure 2This is a functional block diagram of an intelligent molding system for carbon fiber composite materials based on thermo-mechanical coupling, provided in an embodiment of the present invention.

[0070] Figure 3 This is a schematic diagram of the electronic device for implementing the intelligent molding system and equipment based on thermo-mechanical coupling carbon fiber composite materials according to an embodiment of the present invention. Detailed Implementation

[0071] This invention provides an intelligent molding system and equipment for carbon fiber composite materials based on thermo-mechanical coupling. Its core lies in achieving real-time optimization and closed-loop control of multi-physics coupling through digital twin technology, thereby solving the problems of high defect rate, long debugging cycle, and low production efficiency caused by thermo-mechanical coupling in high-pressure resin transfer molding (HP-RTM) processes. The technical solution of this invention will be described in detail below with reference to the accompanying drawings and their reference numerals, as well as specific embodiments.

[0072] like Figure 1 As shown in the flowchart, the intelligent molding system and equipment for carbon fiber composite materials based on thermo-mechanical coupling provided by this invention illustrates the operation logic of the entire system. First, the data acquisition module uses a zoned temperature-controlled mold with built-in fiber Bragg grating sensors to collect real-time temperature and strain field data of the target component. Specifically, the zoned temperature-controlled mold consists of multiple independently controllable heating zones, each embedded with a fiber Bragg grating sensor to monitor the temperature distribution in the resin flow zone, the temperature distribution on the mold surface, and the peak value of curing exothermics in real time. These data are then converted into real-time thermo-mechanical coupling state parameters after signal filtering and outlier removal, including the resin viscosity change rate, mold thermal expansion, and curing degree gradient. The signal filtering uses a low-pass filtering algorithm, as shown in the following formula:

[0073] ;

[0074] in, This represents the filtered output at the current moment. This represents the input signal at the current moment. This represents the filtered output from the previous time step. The filter coefficient typically ranges from 0.1 to 0.3. Low-pass filtering of the initial state parameters effectively removes high-frequency noise interference. Furthermore, by combining the signal-to-noise ratio threshold of the fiber Bragg grating sensor with outlier removal of the initial state parameters, the reliability of the data is further improved.

[0075] Subsequently, the quality assessment module utilizes a pre-built ultrasonic probe array to detect the internal defect distribution of the target component. The ultrasonic probe array is positioned around the mold cavity, emitting high-frequency ultrasonic waves and receiving reflected signals to generate an internal defect distribution map of the target component. Based on real-time thermo-mechanical coupling parameters, the dynamic deformation of the mold cavity is calculated using a mold thermoelastic deformation model. The mathematical expression for the mold thermoelastic deformation model is as follows:

[0076] ;

[0077] in, This indicates the change in the length of the mold cavity. This represents the coefficient of linear expansion of the material. Indicates the initial length of the mold cavity. Let F represent the temperature change, E represent the applied force, E represent the elastic modulus of the material, and A represent the area of ​​force application. This model comprehensively considers the effects of thermal expansion and mechanical stress on the geometry of the mold cavity, thus accurately predicting the dynamic deformation. Based on this, and combining the distribution of internal defects with the dynamic deformation of the mold cavity, a weighted scoring method is used to evaluate the molding quality index of the target component. The calculation formula for the weighted scoring method is as follows:

[0078] ;

[0079] Where Q represents the molding quality index, and D represents the quantified value of the internal defect distribution. This indicates the amount of dynamic deformation of the mold cavity. and These are the corresponding weight coefficients, and they satisfy... The weighting is dynamically adjusted based on the defect type and deformation magnitude. For example, for crack-type defects, It can be set to 0.7, but for porosity defects, It can be set to 0.5.

[0080] Next, the process optimization module determines whether the molding quality index is lower than a preset quality threshold. If the molding quality index is lower than the threshold, it indicates that the current molding state is not within the acceptable range, and dynamic compensation of the injection pressure and mold zone temperature is required based on the molding quality index to obtain optimized process parameters. The formula for constructing the dynamic compensation strategy matrix is ​​as follows:

[0081] ;

[0082] ;

[0083] in, Indicates the injection pressure increment. This indicates the temperature adjustment value for each mold zone. and These are the corresponding proportionality coefficients. Here, Q represents the current molding quality index, and the current process parameters are calculated using a dynamic compensation strategy matrix to obtain optimized process parameters, including target injection pressure, target zone temperature, and target curing time. If the molding quality index is not lower than the quality threshold, the current process parameters are directly used as optimized process parameters, thus avoiding unnecessary compensation adjustments and accelerating the molding process optimization process.

[0084] To further optimize the molding process, the system acquires historical molding datasets of the target component and sets the initial process window based on these datasets. The historical molding datasets contain multiple combinations of process parameters labeled with molding quality. Each set records key parameters such as injection pressure, mold temperature, and curing time, along with their corresponding molding quality results. The initial process window setting process is as follows: First, the process boundary conditions from the historical molding datasets are extracted, including the maximum injection pressure, minimum mold temperature, and longest curing time, and these are used as the upper and lower limits of the initial process window, respectively. Then, based on a preset safety margin, the optimal combination of process parameters is selected within the initial process window range to obtain the initial process window. For example, if the maximum injection pressure in the historical molding dataset is 10 MPa, the minimum mold temperature is 80°C, and the longest curing time is 60 minutes, then the initial process window can be set to an injection pressure range of 8-10 MPa, a mold temperature range of 80-120°C, and a curing time range of 40-60 minutes.

[0085] Based on the initial process window, the system uses a deep learning model to construct an initial process optimization model, and trains it using historical molding datasets to obtain the target process optimization model. For example... Figure 2 As shown, the functional block diagram of the intelligent molding system for carbon fiber composites based on thermo-mechanical coupling clearly illustrates the collaborative relationships between the modules. During training, the historical molding dataset is randomly divided into a training dataset and a validation dataset. The training dataset is used to extract training samples and input them into the initial process optimization model to obtain the model's prediction results; the validation dataset is used to extract validation samples and input them into the validation process optimization model to evaluate the model's generalization ability. If the model's prediction results are consistent with the molding quality labels of the training samples, the initial process window is used as the target process window; if they are inconsistent, the initial process window is adjusted according to the training samples. The adjustment process includes determining the type of difference between the model's prediction results and the molding quality labels, such as premature curing or insufficient curing. If the difference type is premature curing, the initial process window is delayed; if the difference type is insufficient curing, it is accelerated. Through continuous iterative optimization, the target process optimization model is finally obtained.

[0086] Finally, the dynamic adjustment module dynamically adjusts the molding process of the target component based on the optimized process parameters and the target process optimization model. Specific steps include inputting the optimized process parameters into the target process optimization model to obtain the predicted molding state; matching the predicted molding state with the target process window to obtain a molding process adjustment scheme. The molding process adjustment scheme may involve injection path optimization, zoned temperature distribution optimization, and extending or shortening the curing time. For example, if the predicted molding state shows insufficient curing in a certain area, the molding quality can be improved by increasing the mold temperature in that area or extending the curing time.

[0087] like Figure 3 As shown, the electronic device implementing the above-described thermo-mechanically coupled carbon fiber composite intelligent molding system and equipment includes a memory and a processor. The memory stores at least one instruction, while the processor executes the instructions stored in the memory to implement the above-described method. Furthermore, the present invention also provides a computer-readable storage medium storing at least one instruction, which, when executed by the processor in the electronic device, can implement the above-described method.

[0088] In summary, this invention achieves intelligent real-time monitoring of the thermo-mechanical coupling state during the molding process through a partitioned temperature-controlled mold with built-in fiber optic grating sensors. Furthermore, it completes closed-loop control of the entire process, from data acquisition to process optimization, by combining an ultrasonic probe array and a deep learning model. Through key technologies such as dynamic compensation strategy matrices and weighted scoring methods, the molding quality is significantly improved and production costs are reduced, providing reliable technical support for the intelligent manufacturing of carbon fiber composite materials.

Claims

1. A smart molding system for carbon fiber composite materials based on thermo-mechanical coupling, characterized in that, The system includes: The data acquisition module is used to acquire real-time temperature field and strain field data of the target component using a partitioned temperature control mold with built-in fiber optic grating sensors. The real-time temperature field data includes the temperature distribution of the resin flow area, the temperature distribution of the mold surface, and the peak value of curing exothermics. The real-time temperature field and strain field data are processed by signal filtering and outlier removal to obtain real-time thermo-mechanical coupling state parameters, including the resin viscosity change rate, the mold thermal expansion, and the degree of curing gradient. The quality assessment module is used to detect the internal defect distribution of the target component using a pre-built ultrasonic probe array, calculate the dynamic deformation of the mold cavity based on the real-time thermo-mechanical coupling state parameters, and evaluate the molding quality index of the target component by combining the internal defect distribution and the dynamic deformation of the mold cavity. The process optimization module is used to determine whether the molding quality index is lower than a preset quality threshold. If the molding quality index is lower than the quality threshold, the injection pressure and mold zone temperature are dynamically compensated according to the molding quality index to obtain optimized process parameters. If the molding quality index is not lower than the quality threshold, the current process parameters are used as optimized process parameters, the historical molding dataset of the target component is obtained, and the initial process window of the target component is set according to the historical molding dataset. The initial process window includes the injection pressure range, the mold temperature range, and the curing time range. The dynamic adjustment module is used to construct an initial process optimization model based on the initial process window and the pre-built deep learning model, train the initial process optimization model using the historical molding dataset to obtain a target process optimization model, and dynamically adjust the molding process of the target component using the target process optimization model based on the optimized process parameters.

2. The intelligent molding system for carbon fiber composite materials based on thermo-mechanical coupling as described in claim 1, characterized in that, The process of performing signal filtering and outlier removal on the real-time temperature field and strain field data to obtain real-time thermo-mechanical coupling state parameters includes: The real-time temperature field and strain field data are subjected to low-pass filtering to obtain preliminary state parameters; The signal-to-noise ratio (SNR) threshold of the fiber Bragg grating sensor is obtained, and outlier removal processing is performed on the preliminary state parameters based on the SNR threshold to obtain real-time thermo-mechanical coupling state parameters.

3. The intelligent molding system for carbon fiber composite materials based on thermo-mechanical coupling as described in claim 1, characterized in that, The step of calculating the dynamic deformation of the mold cavity based on the real-time thermo-mechanical coupling state parameters, and evaluating the molding quality index of the target component by combining the internal defect distribution with the dynamic deformation of the mold cavity, includes: Based on the real-time thermo-mechanical coupling state parameters, the dynamic deformation of the mold cavity is calculated using the mold thermoelastic deformation model. By combining the internal defect distribution with the dynamic deformation of the mold cavity, the molding quality index of the target component is evaluated using a weighted scoring method.

4. The intelligent molding system for carbon fiber composite materials based on thermo-mechanical coupling as described in claim 1, characterized in that, The step of dynamically compensating for injection pressure and mold zone temperature based on the molding quality index to obtain optimized process parameters includes: A dynamic compensation strategy matrix is ​​constructed based on the molding quality index, wherein the dynamic compensation strategy matrix includes the injection pressure increment and the mold zone temperature adjustment value. The current process parameters are compensated and calculated based on the dynamic compensation strategy matrix to obtain the optimized process parameters.

5. The intelligent molding system for carbon fiber composite materials based on thermo-mechanical coupling as described in claim 1, characterized in that, The step of setting the initial process window for the target component based on historical molding datasets includes: Obtain the process boundary conditions of the historical molding dataset, wherein the historical molding dataset is a combination of multiple process parameters with molding quality labels, and the process boundary conditions include maximum injection pressure, minimum mold temperature and longest curing time. The maximum injection pressure, minimum mold temperature, and longest curing time are respectively used as the upper and lower limits of the initial process window to obtain the range of the initial process window. Based on the preset safety margin, the optimal combination of process parameters is selected within the initial process window range to obtain the initial process window.

6. The intelligent molding system for carbon fiber composite materials based on thermo-mechanical coupling as described in claim 1, characterized in that, The step of training the initial process optimization model using the historical molding dataset to obtain the target process optimization model includes: The historical dataset is randomly divided into a training dataset and a validation dataset. Training samples are extracted sequentially from the training dataset; The training samples are input into the initial process optimization model to obtain the model prediction results; Identify the molding quality label of the training sample and determine whether the model prediction result is consistent with the molding quality label of the training sample; If the model prediction result is inconsistent with the molding quality label of the training sample, the initial process window is adjusted according to the training sample to obtain the target process window.

7. The intelligent molding system for carbon fiber composite materials based on thermo-mechanical coupling as described in claim 6, characterized in that, The step of adjusting the initial process window based on training samples to obtain the target process window includes: Determine the type of difference between the model prediction results and the molding quality label, wherein the type of difference includes premature curing and insufficient curing; If the difference type is premature curing, the initial process window is delayed based on the training samples to obtain the target process window; If the difference type is insufficient solidification, the initial process window is accelerated based on the training samples to obtain the target process window.

8. The intelligent molding system for carbon fiber composite materials based on thermo-mechanical coupling as described in claim 7, characterized in that, The types of differences between the prediction results of the judgment model and the molding quality label include: Obtain the target degree of curing and the predicted degree of curing from the model prediction results for the molded quality label; Determine whether the target degree of cure is within the predicted degree of cure range; If the target degree of cure is not within the predicted degree of cure range, the difference type is set to premature cure; If the target degree of cure is within the predicted degree of cure range, the difference type is set to insufficient cure.

9. The intelligent molding system for carbon fiber composite materials based on thermo-mechanical coupling as described in claim 1, characterized in that, The step of dynamically adjusting the molding process of the target component using the target process optimization model based on the optimized process parameters includes: The optimized process parameters are input into the target process optimization model to obtain the predicted forming state; The predicted molding state is matched with the target process window to obtain molding process adjustment schemes, which include injection path optimization, zoned temperature distribution optimization, and extension or shortening of curing time.

10. An electronic device, characterized in that, The electronic device includes: Memory, storing at least one instruction; The processor executes instructions stored in the memory to implement the intelligent molding system based on thermo-mechanical coupling carbon fiber composite materials as described in any one of claims 1 to 9.

Images