Prediction method of epoxy resin curing peak temperature based on RBF neural network

Abstract

The present invention discloses a method for predicting epoxy resin curing peak temperature based on a RBF neural network. The method includes establishing an epoxy resin composite material curing simulation model, collecting epoxy resin composite material curing temperatures, including oven temperature and epoxy resin composite material curing peak temperature, establishing a three-layer RBF neural network, dividing the collected temperature data into a training set and a test set, inputting the training set data into the three-layer RBF neural network for training to obtain a trained RBF neural network, and then inputting the test set data into the RBF neural network for prediction to obtain predicted data. The RBF neural network-based epoxy resin curing temperature prediction method of the present invention can accurately and efficiently predict the peak temperature during the curing process of epoxy resin composite materials.

Description

Technical Field

[0001] The invention belongs to the technical field of insulating materials and relates to an epoxy resin curing peak temperature prediction method based on RBF neural network. Background Art

[0002] Due to their excellent performance, epoxy resin castings are commonly used in various power equipment, such as dry bushings, GIS, and GIL. The refined manufacturing and design optimization of such power equipment are key to achieving advanced manufacturing of power equipment. When improving and optimizing the performance of epoxy resin castings, it is necessary to consider many factors, such as the epoxy resin matrix formulation system, fillers, curing process, structural dimensions, etc., resulting in high-dimensional parameters in the optimization input. At the same time, the curing process is also a complex process involving chemical processes, heat transfer processes, flow processes, and mechanical deformation processes. The high-dimensional parameters make the use of traditional methods for refined design require huge time and economic costs. Similarly, during the optimization of the curing process, fast calculations are required to meet the requirements of various optimization methods. Therefore, how to use experiments and simulation calculations to obtain a large amount of accurate training data requires further research.

[0003] Epoxy resin castings in power equipment not only serve as electrical insulation but also as mechanical support, placing higher demands on the electrical, thermal, and mechanical performance of epoxy resin castings. Epoxy resin castings are generally composed of epoxy resin composite materials. Epoxy resin prepolymer is poured into a mold in an oven and then cured through different temperature processes. During the curing process, the resin generates a large amount of heat, which causes a temperature peak. If this peak temperature is too high, it will affect the internal stress of the composite material and the overall mechanical performance of the equipment. To improve the reliability of the equipment, it is necessary to predict the temperature peak during the curing process in advance. If the predicted peak temperature is too high, appropriate measures can be taken during the curing process of the epoxy resin casting to reduce the accumulation of curing heat. This can reduce the peak temperature and the internal stress generated during the curing process, resulting in better performance of the epoxy resin device and reducing the time of the epoxy resin curing process.

[0004] Currently, most predictions are made by simulating the temperature process through a combination of finite element simulation and experiments. Finite element simulation requires a lot of time to simulate a temperature process. For large-sized epoxy resin castings, the number of grids is large, and the simulation time increases a lot. In addition, finite element calculations cannot meet the requirements of rapid monitoring of the internal structure of the material during the production process of digital equipment. Experiments also require a lot of time, have large errors, and lack practicality. Summary of the Invention

[0005] The purpose of the present invention is to provide an epoxy resin curing temperature prediction method based on RBF neural network, which can accurately and efficiently predict the peak temperature during the curing process of epoxy resin composite materials.

[0006] The technical solution adopted by the present invention is a method for predicting the peak curing temperature of epoxy resin based on RBF neural network, comprising the following steps:

[0007] Step 1: Establishing a simulation model for curing of epoxy resin composite materials, and collecting curing temperatures of epoxy resin composite materials, including oven temperature and curing peak temperature of epoxy resin composite materials;

[0008] Step 2: Establish a three-layer RBF neural network, divide the collected temperature data into a training set and a test set, input the data in the training set into the three-layer RBF neural network for training to obtain a trained RBF neural network, and then input the data in the test set into the RBF neural network for prediction to obtain predicted data.

[0009] Wherein, step 1 specifically includes the following steps:

[0010] Step 1.1, establish a simulation model for epoxy resin composite curing;

[0011] Step 1.2: Create an interpolation function in the model to simulate the temperature change of the oven during the curing of epoxy resin composite materials:

[0012]

[0013] Wherein, t represents time, f(t) represents the temperature of the incubator, h1 represents the temperature of the first insulation stage of the incubator, h2 represents the temperature of the second insulation stage of the incubator, k1 represents the heating rate of the incubator in the initial stage, k2 represents the heating rate of the incubator after the first insulation stage, t1 represents the insulation time of the first insulation stage of the incubator, t2 represents the insulation time of the second insulation stage of the incubator, k3 represents the cooling rate of the incubator in the cooling stage, k3=1℃ / min;

[0014] Step 1.3: During the heating process of the incubator, the temperature inside the epoxy resin and at the interface between the metal and the epoxy resin is detected using a domain probe. During the first insulation stage of the incubator, the peak temperature inside the epoxy resin is M 11 , the peak temperature at the metal and epoxy interface is M 21 In the second insulation stage of the incubator, the peak temperature inside the epoxy resin is M 12 , the peak temperature at the metal and epoxy interface is M 22 .

[0015] In step 1.1, the epoxy resin composite material curing simulation model has epoxy resin in the middle and metal plates on the bottom and sides.

[0016] The heating rate k1 of the incubator in the initial stage ranges from 1 to 5°C / min.

[0017] The temperature h1 of the first insulation stage of the incubator ranges from 80 to 100°C, and the temperature h2 of the second insulation stage of the incubator ranges from 180 to 240°C.

[0018] In step 2, the data in the training set is input into the three-layer RBF neural network for training. When the root mean square error of the training result is less than 2, the trained RBF neural network is obtained.

[0019] In step 2, before inputting the data in the training set into the three-layer RBF neural network for training, the data is normalized.

[0020] Each set of collected temperature data includes 8 data, namely h1, h2, k1, t1, M 11 、 M 21 、 M 12 、 M 22 .

[0021] The beneficial effects of the present invention are that the RBF neural network is used to predict the peak curing temperature of the epoxy resin composite material, which saves time, has high predicted temperature accuracy and is highly practical. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a structural schematic diagram of the epoxy resin composite material curing simulation model in the present invention;

[0023] Figure 2 is the temperature change curve of the incubator simulated in the present invention;

[0024] Figure 3 is the temperature curve during the curing process of the epoxy resin composite material of the present invention;

[0025] Figure 4 This is the first heat preservation stage of the incubator in the embodiment of the present invention. M 11 Point temperature peak prediction result diagram;

[0026] Figure 5 This is the second insulation stage of the incubator in the embodiment of the present invention. M 21 Point temperature peak prediction result diagram;

[0027] Figure 6This is the first heat preservation stage of the incubator in the embodiment of the present invention. M 12 Point temperature peak prediction result diagram;

[0028] Figure 7 This is the second insulation stage of the incubator in the embodiment of the present invention. M 22 Point temperature peak prediction result diagram. DETAILED DESCRIPTION

[0029] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0030] The present invention provides an epoxy resin curing peak temperature prediction method based on RBF neural network, comprising the following steps:

[0031] Step 1: Establishing a simulation model for curing of epoxy resin composite materials, and collecting curing temperatures of epoxy resin composite materials, including oven temperature and curing peak temperature of epoxy resin composite materials;

[0032] Step 1 specifically includes the following steps:

[0033] Step 1.1, establish the epoxy resin composite curing simulation model, refer to Figure 1 The middle part of the model is epoxy resin, the bottom is aluminum profile 6063-T83, one side is structural steel, and the other side is aluminum profile 6063-T83;

[0034] Step 1.2: Create an interpolation function in the model to simulate the temperature change of the oven during the curing of epoxy resin composite materials:

[0035]

[0036] Where t represents time, f(t) represents the temperature of the incubator, h1 represents the temperature of the first insulation stage of the incubator, ranging from 80 to 100°C, h2 represents the temperature of the second insulation stage of the incubator, ranging from 180 to 240°C, k1 represents the heating rate of the incubator in the initial stage, ranging from 1 to 5°C / min, k2 represents the heating rate of the incubator after the first insulation stage, t1 represents the insulation time of the first insulation stage of the incubator, ranging from 180 min to 240 min, t2 represents the insulation time of the second insulation stage of the incubator, k3 represents the cooling rate of the incubator in the cooling stage, k3 = 1°C / min. For the interpolation function diagram, see Figure 2 ;

[0037] Step 1.3: During the heating process of the incubator, the temperature inside the epoxy resin and at the interface between the metal and the epoxy resin is detected using a domain probe. During the first insulation stage of the incubator, the peak temperature inside the epoxy resin is M 11, the peak temperature at the metal and epoxy interface is M 21 In the second insulation stage of the incubator, the peak temperature inside the epoxy resin is M 12 , the peak temperature at the metal and epoxy interface is M 22 , see Figure 3 .

[0038] By changing h1, h2, k1, t1, different M 11 、 M 21 、 M 12 、 M 22 ， This embodiment collects 135 sets of temperature data.

[0039] Step 2: Establish a three-layer RBF neural network in Matlab and divide the collected temperature data into training set and test set. The training set includes 100 groups of temperature data and the remaining 35 groups are used as test sets. Each group of collected temperature data includes 8 data, namely h1, h2, k1, t1, M 11 , M 21 , M 12 , M 22 , the data in the training set are normalized to eliminate the influence of singular data and amplitude on the prediction results.

[0040] In MATLAB, call the newrbe toolbox and set the radial basis function expansion rate. Input the normalized data into a three-layer RBF neural network for training. When the root mean square error (RMSE) of the training results is less than 2, the trained RBF neural network is obtained. Then, input the test data into the RBF neural network for prediction. The predicted data is then denormalized. This denormalization facilitates comparison with the original data and measures the performance of the neural network.

[0041] Figure 4 It is the first insulation stage of the incubator M 11 Point temperature peak prediction result diagram, Figure 5 It is the second insulation stage of the incubator M 21 Point temperature peak prediction result diagram, Figure 6 It is the first insulation stage of the incubator M 12Point temperature peak prediction result diagram, Figure 7 It is the second insulation stage of the incubator M 22 Point temperature peak prediction result diagram, from Figure 4-7 It can be seen from the figure that the temperature peak value predicted by the epoxy resin curing peak temperature prediction method based on RBF neural network of the present invention is highly accurate and practical.

Claims

1. A method for predicting the peak temperature of epoxy resin curing based on RBF neural network, characterized in that: The following steps are involved: Step 1: Establishing a simulation model for curing of epoxy resin composite materials, and collecting curing temperatures of epoxy resin composite materials, including oven temperature and curing peak temperature of epoxy resin composite materials; The step 1 specifically includes the following steps: Step 1.1, establish a simulation model for epoxy resin composite curing; Step 1.2: Create an interpolation function in the model to simulate the temperature change of the oven during the curing of epoxy resin composite materials: Wherein, t represents time, f(t) represents the temperature of the incubator, h1 represents the temperature of the first insulation stage of the incubator, h2 represents the temperature of the second insulation stage of the incubator, k1 represents the heating rate of the incubator in the initial stage, k2 represents the heating rate of the incubator after the first insulation stage, t1 represents the insulation time of the first insulation stage of the incubator, t2 represents the insulation time of the second insulation stage of the incubator, k3 represents the cooling rate of the incubator in the cooling stage, k3=1℃ / min; Step 1.3: During the heating process of the incubator, the temperature inside the epoxy resin and at the interface between the metal and the epoxy resin is detected using a domain probe. During the first insulation stage of the incubator, the peak temperature inside the epoxy resin is M 11 , the peak temperature at the metal and epoxy interface is M 21 In the second insulation stage of the incubator, the peak temperature inside the epoxy resin is M 12 , the peak temperature at the metal and epoxy interface is M 22 ; Step 2: Establish a three-layer RBF neural network, divide the collected temperature data into a training set and a test set, input the data in the training set into the three-layer RBF neural network for training to obtain a trained RBF neural network, and then input the data in the test set into the RBF neural network for prediction to obtain predicted data.

2. The method for predicting epoxy resin curing peak temperature based on RBF neural network according to claim 1, wherein: In step 1.1, the epoxy resin composite material curing simulation model has epoxy resin in the middle and metal plates on the bottom and sides.

3. The method for predicting epoxy resin curing peak temperature based on RBF neural network according to claim 1, wherein: The heating rate k1 of the incubator in the initial stage is in the range of 1 to 5°C / min.

4. The method for predicting epoxy resin curing peak temperature based on RBF neural network according to claim 1, wherein: The temperature h1 of the first insulation stage of the incubator ranges from 80 to 100°C, and the temperature h2 of the second insulation stage of the incubator ranges from 180 to 240°C.

5. The method for predicting epoxy resin curing peak temperature based on RBF neural network according to claim 1, wherein: In step 2, the data in the training set is input into the three-layer RBF neural network for training. When the root mean square error of the training result is less than 2, a trained RBF neural network is obtained.

6. The method for predicting epoxy resin curing peak temperature based on RBF neural network according to claim 5, characterized in that: In step 2, before inputting the data in the training set into the three-layer RBF neural network for training, the data is first normalized.

7. The method for predicting epoxy resin curing peak temperature based on RBF neural network according to claim 1, wherein: Each set of collected temperature data includes 8 data, namely h1, h2, k1, t1, M 11 、 M 21 、 M 12 、 M 22 .

Images