A method for monitoring the curing temperature field of composite material components

By setting temperature sensors on the outer surface of composite material components and establishing a temperature field mapping model, and combining contact and non-contact sensors, the real-time and accuracy problems of temperature field monitoring during the curing process of composite material components were solved, achieving high-precision temperature field monitoring and improving the quality control of the manufacturing process.

CN118810076BActive Publication Date: 2026-11-10NANJING UNIV OF AERONAUTICS & ASTRONAUTICS
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202410996245.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-07-24
Publication Date
2026-11-10
Estimated Expiration
2044-07-24

AI Technical Summary

Technical Problem

Existing technologies make it difficult to achieve real-time and accurate monitoring of the temperature field during the curing process of composite material components. Direct monitoring methods affect the component structure, while indirect monitoring methods have limited accuracy.

Method used

By setting temperature sensors on the outer surface of the vacuum bag or the non-mold side of the mold, a mapping model between the outer surface temperature distribution and the internal temperature field is established. By combining contact and non-contact sensors, the model is corrected using a small amount of sensor data, thus achieving high-precision real-time monitoring of the temperature field.

Benefits of technology

It enables high-precision, real-time temperature field monitoring during the curing process of composite material components, improves the quality control level of the manufacturing process, and avoids the influence of sensors on material properties.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118810076B_ABST
    Figure CN118810076B_ABST
Patent Text Reader

Abstract

A method for monitoring the curing temperature field of composite material components is characterized by the following steps during the curing process: Temperature sensors are installed on the outer surface of a vacuum bag or the non-mold side of a mold to measure the temperature distribution on the outer surface. A mapping model is established between the outer surface temperature distribution and the temperature field distribution on the surface or inside of the composite material component. Then, during the curing process, the surface or internal temperature field of the composite material component is inferred in real time based on the outer surface temperature field. By arranging a small number of temperature sensors on or inside the component, the measured values ​​from these sensors are used to correct the model mapping, resulting in a high-precision measurement of the surface or internal temperature field of the composite material component. This invention establishes a mapping relationship between the external temperature field and the internal temperature field of the component, and accurately measures the curing temperature field on the surface or inside the component using data from a small number of temperature sensors.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of composite material manufacturing and molding, and in particular to a method for monitoring the curing state of composite material components, specifically a method for monitoring the curing temperature field of composite material components. Background Technology

[0002] Composite materials, due to their excellent properties such as light weight, high strength, and corrosion resistance, are widely used in high-tech fields such as aerospace, automotive, and shipbuilding. However, the manufacturing process of composite materials is complex, especially during the curing process, where the distribution of the temperature field has a significant impact on the quality of the final product. The curing process of composite material components involves complex chemical reactions and heat transfer processes, requiring strict control of the temperature field distribution to ensure the uniformity and stability of material properties.

[0003] Controlling the curing temperature field requires real-time monitoring. Existing temperature monitoring methods are divided into two categories: direct monitoring and indirect monitoring. Direct monitoring methods involve monitoring the temperature of components using temperature sensors. One approach is to embed multiple distributed temperature sensors on or inside the material surface. However, this method cannot provide global temperature information and may affect the structural integrity of composite material components, making it difficult to meet the industrial demands for high-efficiency and high-quality composite material production. Another approach is to use infrared thermal imagers to obtain temperature field information on the surface of vacuum bags. However, this method is greatly affected by environmental heat sources and cannot accurately reflect the actual temperature field of the component. Indirect monitoring methods refer to obtaining a small amount of easily monitored temperature data and then using numerical simulation combined with actual data to interpolate and predict the global temperature. However, these methods rely on a large amount of data input. For example, CN116341388A discloses a method for predicting the peak curing temperature based on a radial basis function neural network. Although it has high accuracy, it can only predict at a specific time point. Another example is the reference (DOI:10.1007 / s40436-023-00455-z) which proposes a mapping model between the temperature field of the vacuum bag surface and the temperature field of the component using a convolutional neural network. Although it can predict in real time, it can only obtain the surface temperature of the component, and the accuracy of the mapping model is limited due to parameter and modeling errors. Therefore, existing direct measurement methods are in principle unable to directly obtain the temperature field of the component surface or interior, while indirect measurement methods, although capable of measuring the temperature field, are still limited by the accuracy of the model and cannot obtain accurate temperature field results. Summary of the Invention

[0004] The purpose of this invention is to address the problem that the actual temperature field of composite components is difficult to monitor in real time and accurately during the current process of composite material curing. The invention proposes a method for monitoring the curing temperature field of composite material components that integrates models and actual data. This method overcomes the limitations of traditional methods that introduce a large number of sensors into the interior or surface of composite material components, thereby affecting the quality of the components. By establishing a mapping relationship between the external temperature field and the internal temperature field of the component, the curing temperature field on the surface or inside of the component can be accurately measured with only a small amount of temperature sensor data correction. This provides a new and feasible technical solution for online monitoring of the temperature field of composite material components.

[0005] The technical solution of this invention is:

[0006] A method for monitoring the curing temperature field of composite material components is characterized by the following steps during the curing process: Temperature sensors are installed on the outer surface of a vacuum bag or the outer surface of a mold (on the non-mold side) to measure the temperature distribution on the outer surface. A mapping model is established between the outer surface temperature distribution and the temperature field distribution on the surface or inside of the composite material component. Then, the temperature field on the surface or inside of the composite material component is inferred in real time based on the outer surface temperature field during the curing process. By arranging several (or a small number, less than 10) temperature sensors on or inside the component, the model mapping is corrected in real time using the measured values ​​from these few temperature sensors to obtain a high-precision temperature field on the surface or inside of the composite material component.

[0007] The temperature sensors used in this invention include both contact and non-contact sensors. In convective heat transfer heating processes, contact sensors such as distributed thermocouples are used, as their contact nature provides highly accurate local temperature data. In contact heat transfer heating processes, non-contact infrared sensors can be used, which can rapidly capture changes in the surface temperature field without interfering with the internal structure of the component. Furthermore, the two sensing methods can be used in combination to adapt to different monitoring needs and environmental conditions, enhancing the flexibility and accuracy of the monitoring method.

[0008] The mapping model of this invention is established through a reverse heat transfer process, involving a complex heat transfer system composed of heat sources, components, molds, and auxiliary materials. The description of heat transfer characteristics relies on the transient heat transfer equation, which integrates factors such as heat conduction in various parts of the system, system dissipation, and temperature difference with the environment. The steps for establishing the mapping model of this invention are as follows: First, the data processing and storage process requires noise filtering and data alignment of the data collected by temperature sensors. Data from different sensors are integrated to form a complete temperature field dataset. The integration process must consider the spatial location and measurement accuracy of the sensors to ensure data consistency and accuracy. The integrated data is then stored in a database for subsequent training. Data storage can be based on data capacity and access speed, choosing relational databases, time-series databases, etc. Next is the model building and training process. Since the mapping model in this invention is not only based on the traditional heat transfer equation but also integrates machine learning technology to handle complex heating scenarios, the model building process includes two parts: first, heat transfer system modeling, using the transient heat transfer equation to describe the heat transfer characteristics of the heat source, components, molds, and auxiliary materials. The heat transfer characteristics of each part of the heat transfer system are described by the transient heat transfer equation.

[0009]

[0010] in, Characterizing the rate of temperature change of each part of the system, Characterizing the real-time temperature of each part of the system, h·A·(TT) ′ It represents the temperature difference between the system and the environment.

[0011] Secondly, the invention integrates physical priors with machine learning algorithms. It employs a multi-subgraph architecture convolutional neural network to establish a mapping model between surface infrared temperature images and the actual temperature field of the component, addressing the problem of monitoring the curing temperature field in contact heat transfer scenarios. The heat transfer characteristics are broken down into three parts: temperature change rate, real-time temperature, and temperature difference. Subgraphs for each of these three parts are then established to build and train the multi-subgraph architecture convolutional neural network. A physical information neural network integrating heat transfer physics laws and data-driven approaches is also established to create a mapping model between distributed temperature data and the actual temperature field of the component, addressing the problem of monitoring the curing temperature field in convective heat transfer scenarios. Finally, the invention includes a real-time monitoring and correction process. Real-time collected external surface temperature data is input into the trained mapping model. The model infers the temperature field distribution on or inside the surface of the composite material component based on the input data, outputs the inference results, and compares them with the actual temperature points measured by a small number of thermocouples placed at the target location. The measured data is used to correct the model, improving monitoring accuracy.

[0012] The correction method first obtains the deviation between a small amount of real temperature data on the surface or inside of the component and the temperature data inferred in real time using a mapping relationship. This deviation is then used to globally adjust the distribution of the temperature data inferred in real time using the mapping relationship, thereby obtaining accurate temperature field data for the surface or inside of the component. When the deviation is applied to a numerical model, it is used to correct the output value of the numerical model. The temperature values ​​at each point can be proportionally summed based on their magnitude relative to the temperature values ​​at the measured points. When the deviation is applied to a machine learning model, it can be achieved through online training with a small amount of data, locally freezing the network parameters of the trained model, and fine-tuning the network parameters of local layers. This results in a new model that retains the original model information while adapting to real data, thus achieving deviation correction.

[0013] The beneficial effects of this invention are:

[0014] The method of this invention enables high-precision, real-time monitoring of the temperature field during the curing process of composite material components, thereby improving the quality control level of the manufacturing process. This method effectively avoids the impact of embedding numerous sensors on material properties, as is common in traditional methods, and has high practical value and broad application prospects.

[0015] This invention addresses the shortcomings of existing temperature monitoring technologies in terms of accuracy, real-time performance, and global coverage.

[0016] This invention, during the curing process of composite material components, measures the temperature distribution on the outer surface of a vacuum bag or the non-mold side of a mold by placing temperature sensors on the outer surface of the vacuum bag. A mapping model is established between this outer surface temperature distribution and the temperature field distribution on the surface or inside the composite material component. Then, during the curing process, the surface or internal temperature field of the composite material component is inferred in real time based on the outer surface temperature field. By deploying a small number of temperature sensors on or inside the component, the measured values ​​from these sensors are used to correct the model mapping during actual curing, resulting in a high-precision measurement of the surface or internal temperature field of the composite material component. This method, utilizing the fusion of model and measured data, achieves high-precision real-time monitoring of the temperature field of the composite material component, thereby significantly improving the quality control accuracy and efficiency in the composite material manufacturing process. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the implementation of the present invention.

[0018] Figure 2 This is a schematic diagram of the temperature field mapping model construction in an embodiment of the present invention.

[0019] Figure 3 This is a schematic diagram of model block correction in an embodiment of the present invention.

[0020] Figure 4This is a diagram illustrating the monitoring effect of the model in an embodiment of the present invention. Detailed Implementation

[0021] The following description, in conjunction with the accompanying drawings and specific embodiments, provides further details.

[0022] A method for monitoring the curing temperature field of composite material components is disclosed. This method, given a known heat source distribution location, selects different intelligent algorithms and combines them with prior physical knowledge related to heat transfer to indirectly monitor the actual temperature field of the component. The method can be used in contact heat transfer and convective heat transfer heating scenarios, and can be applied in various process scenarios such as autoclave curing and in-situ curing. Specifically, it includes the following steps:

[0023] Step 1: Select different types of temperature sensors to acquire data based on the location of the heat source in the scenario, for use in subsequent model training.

[0024] The selection and distribution of temperature sensors vary depending on the heat source scenario. This embodiment considers a system where mold heating is used as the heat source, thus it is a contact heat transfer scenario. An infrared thermal imager is placed on top of the vacuum bag to acquire its surface temperature field information. Temperature field data is collected every 10 seconds and stored in time-series image format. Figure 1 As shown, since infrared thermal imagers mostly use wide-angle lenses, the data they collect needs to be distorted and processed using algorithms such as bilateral filtering or wavelet transform denoising to facilitate subsequent model training.

[0025] Step 2: Construct a temperature field mapping model that incorporates physical priors.

[0026] Based on the heat source scenario addressed in Step 1 and the obtained training dataset, combined with the features of the model to be trained and the dataset features, a reasonable physical prior is selected to establish the temperature field mapping model. Traditional convolutional neural networks primarily focus on spatial features in images, neglecting the temporal continuity or dynamic changes between images. In this embodiment, while the outer surface of the vacuum bag exhibits the same temperature field distribution during the heating and cooling stages, the temperature field distribution of the corresponding components differs due to the directionality of heat transfer. Using a traditional convolutional neural network would result in a one-input-many-output situation, failing to achieve accurate mapping. In this embodiment, we attempt to employ a transient heat transfer equation.

[0027]

[0028] As prior physical knowledge, it is used to characterize the rate of temperature change, the measured temperature, and the temperature difference between the system environment and the measured temperature. Based on this, the training image set obtained in step one is processed to obtain sub-image sets for the rate of temperature change, the measured temperature, and the temperature difference between the system environment. These sub-image sets are then used to train a convolutional neural network. This extends the traditional convolutional neural network, establishing a three-sub-image architecture to achieve a one-to-one temperature mapping with single input and single output. The mapping model construction process is as follows: Figure 2 .

[0029] Step 3: Use discrete temperature data of the components to correct the model and verify the accuracy of the established model.

[0030] In this embodiment, the component temperature field obtained by mapping the three-subgraph architecture convolutional neural network model incorporating transient heat transfer in step two is image information, which belongs to high-dimensional data. Since in actual process, the local point temperature of the component can only be obtained through distributed thermocouples, the mapping data needs to be discretized during model correction, and the model correction is performed using block-based thermocouple temperature measurement data, such as... Figure 3 Temperature field monitoring was conducted using the modified model at a curing time of t = 500 s, and the results are as follows. Figure 4 The average absolute error of the model-mapped temperature field is 1.43℃, and the relative error is less than 2%. According to the method of the present invention, the curing temperature field of composite material components can be monitored in real time with high monitoring accuracy.

[0031] The parts not covered in this invention are the same as those in the prior art and are implemented using existing techniques.

Claims

1. A method for monitoring the curing temperature field of composite material components, characterized in that, During the curing process of composite material components, temperature sensors are installed on the outer surface of the vacuum bag or the non-mold side of the mold to measure the temperature distribution on the outer surface. A mapping model is established between the temperature distribution on the outer surface and the temperature field distribution on the surface or inside of the composite material component. Then, during the curing process, the temperature field on the surface or inside of the composite material component is inferred in real time based on the temperature field on the outer surface. By arranging several temperature sensors on the surface or inside the component, the model mapping is corrected in real time using the sensor measurements, resulting in a high-precision temperature field on the surface or inside of the composite material component. The method for establishing the mapping model is an inverse heat transfer fitting process, using transient heat transfer equations. Describe the heat transfer process of the outer and inner surfaces, wherein, Characterizing the rate of temperature change of each part of the system, Characterizes the real-time temperature of each part of the system. To characterize the system's heat dissipation to the external environment, a mapping relationship between the temperature fields of the outer and inner surfaces is established based on online numerical prediction or machine learning models. The steps for establishing the mapping model are as follows: The first step is data processing and storage. Noise filtering and data alignment are required for the data collected by the temperature sensors. Data from different sensors are then integrated to form a complete temperature field dataset. The integration process must consider the spatial location and measurement accuracy of the sensors to ensure data consistency and accuracy. The integrated data is then stored in a database for subsequent training. Data storage is based on data capacity and access speed, with relational databases and time-series databases being selected accordingly. Next is the model building and training process. Since the mapping model integrates machine learning technology to handle complex heating scenarios based on the traditional heat transfer equation, the model building process includes two parts: first, heat transfer system modeling, using transient heat transfer equations to describe the heat transfer characteristics of heat sources, components, molds, and auxiliary materials; second, the fusion of physical priors and machine learning algorithms. A multi-subgraph architecture convolutional neural network is used to establish a mapping model between surface infrared temperature images and the actual temperature field of components to solve the problem of monitoring the curing temperature field in contact heat transfer scenarios. The heat transfer characteristics are divided into three parts: temperature change rate, real-time temperature, and temperature difference. Subgraphs of the three parts are established to build and train the multi-subgraph architecture convolutional neural network. A physical information neural network that integrates heat transfer physics laws and data-driven approaches is established to establish a mapping model between distributed temperature data and the actual temperature field of components to solve the problem of monitoring the curing temperature field in convective heat transfer scenarios. Finally, there is the real-time monitoring and correction process. The real-time collected external surface temperature data is input into the trained mapping model. The mapping model infers the temperature field distribution on or inside the surface of the composite material component in real time based on the input data. The inference results are output and compared with the actual temperature points measured by a small number of thermocouples set at the target location. The measured data is used to correct the model to improve the monitoring accuracy.

2. The method according to claim 1, characterized in that, The temperature sensor includes contact sensors and non-contact sensors. Distributed contact sensors or non-contact infrared sensors are used on the outer surface of the vacuum bag or the outer surface of the mold on the non-mold side, or a measurement method combining the two sensing methods is adopted.

3. The method according to claim 1, characterized in that, The correction method is as follows: First, obtain the deviation between a small amount of real temperature data of the component surface or interior and the temperature data inferred in real time using a mapping relationship. Then, use this deviation to adjust the global distribution of the temperature data inferred in real time using a mapping relationship, thereby obtaining accurate temperature field data of the component surface or interior.

Citation Information

Patent Citations

  • Epoxy resin curing peak temperature prediction method based on RBF neural network

    CN116341388A

  • Simulation method and system for temperature field of oil-immersed power transformer

    CN117436334A

  • Temperature field testing device in composite material shaping process

    CN2583649Y