Electromagnetic coupling response waveform prediction method based on one-dimensional CNN

Abstract

The invention discloses an electromagnetic coupling response waveform prediction method based on a one-dimensional CNN, and the method comprises the steps: analyzing, sampling and constructing a cable electromagnetic radiation response prediction data set, and dividing the data set into a training set, a verification set and a test set; a CNN-LSTM model is constructed, a one-dimensional convolutional neural network CNN 1d encodes sequence information of signals and extracts local features of the signals, then a long-short term memory network LSTM learns a sequential relationship of the signals, and finally a full connection layer performs prediction output; training the model by using the training set, and verifying the trained model by using the verification set; comparing the waveform predicted by the model with the waveform obtained by full-wave simulation to obtain a cable electromagnetic radiation response prediction model; and inputting the test set into the trained cable electromagnetic radiation response prediction model for prediction, and further adjusting model parameters. According to the method, rapid prediction of the wire harness response waveform under electromagnetic irradiation can be reflected more practically, the calculation time is saved, and engineering application is facilitated.

Description

Technical Field

[0001] The invention relates to an electromagnetic coupling response waveform prediction method based on one-dimensional CNN, and belongs to the field of electromagnetic radiation response waveform prediction. Background Art

[0002] With the continuous development of electronic technology, the threat of electromagnetic interference faced by electronic equipment has become increasingly severe. Electromagnetic radiation can cause varying degrees of interference or even damage to electronic equipment, seriously affecting the reliability of electronic equipment. Studies have shown that the damage caused to electronic equipment by electromagnetic interference is related to the coupling path of electromagnetic interference, the sensitivity of electronic equipment and the source of electromagnetic interference. Electromagnetic radiation generates interference signals (induced voltage and induced current) on the interconnected cables inside the electronic system through field line coupling. These interference signals will affect the electronic equipment at the cable terminal.

[0003] How to obtain interference signals on the cables of electronic equipment has become a prerequisite for analyzing and evaluating the electromagnetic radiation effects of electronic systems. Since the electromagnetic radiation effect test is a lossy test and the test price is expensive, it is easier to obtain the interference signal waveform of the electronic system by simulation. Since there are many factors that affect the wiring harness coupling voltage waveform, the current methods mostly use the electromagnetic radiation modulation signal given in the standard as the excitation source for simulation, which can only be refined to a point and cannot reflect the actual situation under microwave radiation. Summary of the invention

[0004] The technical problem solved by the present invention is: to overcome the shortcomings of the prior art, to provide an electromagnetic coupling response waveform prediction method based on a one-dimensional CNN, to achieve rapid prediction of the wire harness response waveform under electromagnetic irradiation that can better reflect the actual situation, to save calculation time, and to facilitate engineering applications.

[0005] The technical solution of the present invention is: a method for predicting electromagnetic coupling response waveform based on one-dimensional CNN, comprising:

[0006] Analyze the factors that affect the electromagnetic radiation response waveform of cables, sample and construct a cable electromagnetic radiation response prediction data set within the transformation range of each factor, and divide the data set into training set, validation set and test set;

[0007] Construct a CNN-LSTM model, including a one-dimensional convolutional neural network CNN 1d, a long short-term memory network LSTM and a fully connected layer. The one-dimensional convolutional neural network is used to encode the sequence information of the signal and extract the local features of the signal. Then the long short-term memory network LSTM learns the temporal relationship of the signal, and finally the fully connected layer is used for prediction output.

[0008] The CNN-LSTM model is trained using the training set, and the CNN-LSTM model after each round of training is verified using the verification set. The waveform predicted by the CNN-LSTM model during verification is compared with the electromagnetic radiation response voltage waveform of the cable obtained by full-wave simulation, and the parameters of the CNN-LSTM model are adjusted accordingly to obtain a preliminary cable electromagnetic radiation response prediction model.

[0009] The test set is input into the preliminary cable electromagnetic radiation response prediction model for prediction. The prediction results are compared with the cable electromagnetic radiation response voltage waveform obtained by full-wave simulation to evaluate the prediction performance of the CNN-LSTM model. The parameters of the CNN-LSTM model are further adjusted to obtain the final cable electromagnetic radiation response prediction model.

[0010] The physical factors that affect the electromagnetic response waveform of the cable are input into the final cable electromagnetic radiation response prediction model to predict the electromagnetic response waveform of the corresponding cable.

[0011] Preferably, when constructing a cable electromagnetic radiation response prediction data set, physical factors that affect the cable electromagnetic response waveform are selected as input variable parameters, and the cable electromagnetic response waveform is selected as the output variable.

[0012] Preferably, the factors affecting the electromagnetic radiation response waveform of the cable include: transmission line parameters, excitation source parameters and excitation source waveform; wherein:

[0013] Transmission line parameters include: cable length, cable height from the ground plane;

[0014] The excitation source parameters include: transient electric field peak, center frequency, incident angle, polarization angle and azimuth angle.

[0015] Preferably, the variation range of each factor is specifically as follows:

[0016] The polarization angle ranges from 0 to π; the incident angle ranges from 0 to π / 2; the azimuth angle ranges from 0 to π; the transient electric field peak value of the excitation source ranges from 20 kV / m to 200 kV / m; the center frequency ranges from 0.5 GHz to 2.5 GHz,

[0017] The cable length ranges from 50 cm to 100 cm, and the cable height from the ground plane ranges from 2 cm to 10 cm.

[0018] Preferably, the CNN-LSTM model is specifically:

[0019] The convolution kernel size of the one-dimensional convolutional neural network CNN 1d is 20 and the number of convolution kernels is 64;

[0020] Two parallel long short-term memory networks (LSTMs) are used to learn the temporal relationship of signals. The number of neurons in the two parallel long short-term memory networks (LSTMs) is set to 20 and 10 respectively.

[0021] The feature maps output by the long short-term memory network LSTM are connected through the channel splicing operation to obtain the spliced ​​fusion features;

[0022] The concatenated fusion features are first input into a fully connected layer with 10 neurons to perform linear transformation, weight the fusion information, and then a fully connected layer with 1 neuron is used to predict the signal waveform.

[0023] Preferably, when sampling within the transformation range of each factor to construct the cable electromagnetic radiation response prediction data set, sampling is performed uniformly within the transformation range of each factor.

[0024] Preferably, the input data needs to be normalized to the interval [-1, 1] before training, and the loss function used in the training process is the mean square error loss, the optimization algorithm is the Adam algorithm, and the initial learning rate is 0.01.

[0025] In a second aspect, a terminal device is provided, including:

[0026] a memory for storing instructions executed by at least one processor;

[0027] The processor is used to execute the instructions stored in the memory to implement the method described above.

[0028] According to a third aspect, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions. When the computer instructions are executed on a computer, the computer executes the method described above.

[0029] Compared with the prior art, the present invention has the following advantages:

[0030] The electromagnetic coupling response waveform prediction method based on the CNN-LSTM model provided by the present invention performs a large number of samplings on physical factors, obtains the output waveform of the sampling point through full-wave numerical simulation, divides the simulation data into a training set, a verification set and a test set, constructs a network model based on CNN-LSTM, and uses the simulation data to train, test and verify it, thereby improving the calculation efficiency and breaking through the previous calculation of the electromagnetic radiation response of the cable under the fixed electromagnetic radiation threat that could only be refined to a point, so that the prediction model can better reflect the electromagnetic radiation threat suffered by the cable under actual conditions, and realize the rapid prediction of the electromagnetic radiation response of the cable. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 A cable model under electromagnetic radiation in the present invention;

[0032] Figure 2 is the prediction network structure of the present invention;

[0033] Figure 3 The internal structure of the LSTM of the present invention;

[0034] Figure 4 This is the prediction result of the verification sample of the present invention. DETAILED DESCRIPTION

[0035] The present invention provides an electromagnetic coupling response waveform prediction method based on one-dimensional CNN, comprising the following steps:

[0036] Step 1: Construct a cable electromagnetic radiation response prediction dataset

[0037] Step 1.1 Select the physical factors that affect the electromagnetic response waveform of the cable as input variables, and the electromagnetic response waveform of the cable as the output variable.

[0038] Step 1.2: Select sample points from the input variable set.

[0039] Step 1.3 uses full-wave simulation software to calculate the coupled voltage and current waveforms of the cable at each sample point.

[0040] Step 1.4 combines each sample point with the coupling voltage waveform corresponding to the sample point to construct a data set; and divides the data set into a training set, a validation set, and a test set.

[0041] Step 2 constructs a CNN-LSTM model, including a one-dimensional convolutional layer (CNN 1d), two long short-term memory network layers (LSTM, Long Short-Term Memory), a splicing layer and two fully connected layers. The connection method is as follows: the signal first passes through the one-dimensional convolutional layer to encode the sequence information. Subsequently, the encoded information is input into the LSTM network with 20 and 10 neurons respectively to extract the temporal relationship of the sequence. Finally, the two different temporal features are spliced ​​on the channel to obtain the fusion feature, and two fully connected layers with 10 and 1 neurons are used to predict the output electromagnetic signal waveform.

[0042] Step 3: Use the training set to train the CNN-LSTM model, and use the validation set to validate the network model after each round of training; compare the waveform predicted by the network with the waveform obtained by full-wave simulation to obtain the cable electromagnetic radiation response prediction model.

[0043] Step 4: Input the sample points of the test set into the trained cable electromagnetic radiation response prediction model for prediction, and perform denormalization on the prediction results to obtain the final prediction results.

[0044] Furthermore, in step 1.1, the factors that affect the electromagnetic radiation response waveform of the cable mainly include transmission line parameters, excitation source parameters and excitation source waveform. The transmission line parameters mainly include cable length and cable height from the ground plane. The excitation source parameters mainly include transient electric field peak value, center frequency, incident angle, polarization angle and azimuth angle.

[0045] Furthermore, in step 1.2, the polarization angle varies in the range of 0 to π; the incident angle varies in the range of 0 to π / 2; the azimuth angle varies in the range of 0 to π; the transient electric field peak of the excitation source varies in the range of 20 kV / m to 200 kV / m; the center frequency varies in the range of 0.5 GHz to 2.5 GHz, and each variable is uniformly sampled within the corresponding variation range.

[0046] Furthermore, in step 1.2, the cable length varies in a range of 50 cm to 100 cm, and the height of the cable from the ground plane varies in a range of 2 cm to 10 cm, and both are sampled evenly within the variation range.

[0047] Furthermore, the full-wave simulation algorithms used in step 1.3 include transmission line matrix method, finite integral method and time-domain finite difference method, etc. The response voltage and current waveforms of the cable are obtained through full-wave simulation.

[0048] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0049] Step 1: Build a cable electromagnetic radiation response prediction dataset

[0050] Step 1.1: Select physical factors that affect the electromagnetic radiation response waveform of the cable as input variables, and the electromagnetic radiation response waveform of the cable as output variables; select sample points from the input variable set.

[0051] The specific method is:

[0052] Step 1.1 Select the physical factors that affect the cable electromagnetic radiation response waveform

[0053] The factors that affect the electromagnetic radiation response waveform of the cable are mainly transmission line parameters, excitation source parameters and excitation source waveform. The transmission line parameters mainly include cable length L and cable height from the ground plane h. The excitation source parameters mainly include transient electric field peak E0, center frequency f0, incident angle ψ, polarization angle α and azimuth angle φ, as well as the electromagnetic excitation source waveform Ax.

[0054] Figure 1 is a cable model under electromagnetic radiation. In this embodiment, the cable line type is selected as a single core line, and the electromagnetic excitation source is selected as a high-power microwave specified in the IEC standard. The expression is E(t) = E o e -αt sin(ωo t), where E0 is the peak value of the transient electric field, ω o =2πf o , f o is the center frequency, α=10 8 rad / s is the damping constant. The cable terminal impedance, cable radius, conductivity of the cable material, relative dielectric constant, etc. are all non-random parameters, and their influence on the electromagnetic radiation response waveform of the cable can be determined. Therefore, they can be determined as fixed parameters and are not used as inputs to the prediction model.

[0055] Step 1.2 Select sample points from the input variable set

[0056] For excitation source parameters, such as Figure 1 As shown, the polarization angle α represents the angle between the electric field and the incident plane, and its range is 0 to π; the incident angle ψ represents the angle between the wave vector direction and the XOY plane, and its range is 0 to π / 2; the azimuth angle φ represents the angle between the projection of the wave vector on the XOY plane and the X-axis, and its range is 0 to π, where the wave vector direction is the propagation direction of the incident plane electromagnetic wave, that is, the normal direction of the incident plane. The transient electric field peak value of the excitation source varies from 20k / V to 100k / V, and the center frequency varies from 0.5GHz to 2.5GHz. The incident angle ψ, azimuth angle φ and polarization angle α of the electromagnetic radiation are adjusted by changing the incident plane wave propagation direction vector V and the electric field vector E. That is:

[0057] V=(-sin D(ψ)×cos D(φ),-sin D(ψ)×sin D(φ),-cos D(ψ))

[0058] E=(cos D(φ)×cos D(ψ)×cos D(α)-sin D(φ)×sin D(α),sin D(φ)×cosD(ψ)×cos D(α)+cos D(φ)×sin D(α),-sin D(ψ)×cos D(α))

[0059] The cable length ranges from 50cm to 100cm, the cable height from the ground plane ranges from 2cm to 10cm, and the cable radius ranges from 1mm to 5mm, and both are sampled evenly within the range.

[0060] In this embodiment, the transient electric field peak value E0, the center frequency f o , polarization angle α, incident angle ψ and azimuth angle φ, as well as the cable length L, cable radius r and cable height H from the ground plane are:

[0061] E0=20h1,h1=1,2,...,5

[0062] f0=0.5i1,i1=1,2,...,5

[0063] α=18°×j1, j1=-5,-4,...,0,...,4,5

[0064] ψ=18°×k1,k1=0,1,...,5

[0065] φ=18°×l1,l1=-5,-4,...,0,...,4,5

[0066] L=50+10u1,u1=0,1,...,5

[0067] H=2v1,v1=1,...,5

[0068] r=1+1w1,w1=1,...,5

[0069] It should be noted that after the excitation source parameters are determined, the electromagnetic radiation waveform is obtained through the excitation source expression, and it is used as one of the physical factors affecting the electromagnetic radiation response waveform of the cable and is used as an input variable.

[0070] A total of 544,500 training sample data were collected.

[0071] E0=20h2,h2=1,3,5

[0072] f0=0.5i2,i2=1,2,...,5

[0073] α=2°+36°×j2,j2=-2,-1,...,2

[0074] ψ=2°+36°×k2,k2=0,1,2

[0075] φ=2°+36°×l2,l2=-2,-1,...,2

[0076] L=52+20u2,u2=0,1,2

[0077] H=0.2+2v2,v2=1,3,5

[0078] r=1+1w2,w2=1,3,5

[0079] A total of 6075 validation sample data were collected

[0080] E0=20h3,h3=1,3,5

[0081] f0=0.5i3,i3=1,2,...,5

[0082] α=4°+36°×j3,j3=-2,-1,...,2

[0083] ψ=4°+36°×k3,k3=0,1,2

[0084] φ=4°+36°×l3,l3=-2,-1,...,2

[0085] L=54+20u3,u3=0,1,2

[0086] H=0.4+2v3,v3=1,3,5

[0087] r=1+1w3,w3=1,3,5

[0088] A total of 6075 test sample data were collected.

[0089] The impedance of both ends of the cable is 50Ω and the conductivity is σ g =100S / m, dielectric constant is ε rg =2.3.

[0090] Step 1.3 uses full-wave simulation software to calculate the response voltage and current waveforms on the cable at each sample point. In this embodiment, the transmission line matrix method (Transmission-Line Modeling, TLM) is used in the full-wave simulation software to calculate the output waveform at each sample point.

[0091] Step 1.4 combines each sample point with the coupling voltage waveform corresponding to the sample point to construct a data set; and divides the data set into a training set, a validation set, and a test set.

[0092] Step 2. The collected sample data contains multiple time series variables, such as excitation source, transient electric field peak, center frequency, polarization angle, incident angle, etc. There is a significant nonlinear relationship between these variables and has obvious spatial characteristics. Therefore, it is necessary to design a special network structure to learn the characteristics of multivariate data and realize accurate prediction of electromagnetic radiation response. The overall network structure is as follows: Figure 2As shown in the figure, the CNN-LSTM model is used to fully learn the signal context information: first, a one-dimensional convolutional layer is used to encode the sequence information of the signal and extract the local features of the signal, then a long short-term memory network (LSTM) is used to learn the temporal relationship of the signal, and finally a fully connected layer is used to predict the output. The collected data is input into the network. The first layer of the one-dimensional convolutional neural network (CNN 1d), whose convolution kernel size is 20 and the number of convolution kernels is 64, is used to learn the nonlinear relationship between multivariate variables. Subsequently, two parallel LSTMs are used to learn the temporal correlation of features. The neurons of the two parallel LSTMs are set to 20 and 10 respectively, which is helpful to learn the features of different time series and enhance the feature representation ability of the model, so as to better capture the complex relationship in the sequence data. The feature maps output by the LSTM are connected through the channel concatenation operation to facilitate the capture of semantic information in the text data. Finally, two fully connected layers with 10 and 1 neurons are used, and the fused features are first input into the fully connected layer with 10 neurons to perform linear transformation and weight the fused information. Subsequently, a fully connected layer with a warp element of 1 is used to predict the signal waveform. , realizing the prediction of electromagnetic radiation response.

[0093] Step 3: Use the training set to train the CNN-LSTM model, and use the verification set to verify the network model after each round of training; compare the waveform predicted by the network during verification with the waveform obtained by full-wave simulation to obtain the cable electromagnetic radiation response prediction model.

[0094] It should be noted that the input data needs to be normalized before training, and the data is normalized to the interval [-1,1]. The loss function used in the training process is the mean square error loss, the optimization algorithm is the Adam algorithm, and the initial learning rate is 0.01.

[0095] Step 4: Input the sample points of the test set into the trained cable electromagnetic radiation response prediction model for prediction, and perform denormalization on the prediction results to obtain the final prediction results. The prediction results are compared with the electromagnetic radiation response voltage waveform of the cable obtained by the full-wave simulation software to evaluate the prediction performance of the CNN-LSTM model, and further adjust the parameters of the CNN-LSTM model to achieve verification and obtain the final cable electromagnetic radiation response prediction model.

[0096] Subsequently, input variables (physical factors that affect the electromagnetic response waveform of the cable) are input into the final cable electromagnetic radiation response prediction model, and the model can predict the electromagnetic response waveform of the corresponding cable.

[0097] In this embodiment, 544,500 training samples are used to train the model, 6075 groups of validation samples are used to validate the model, adjust the hyperparameters in the model, and perform a preliminary assessment of the model capability.

[0098] Figure 4 It is the prediction result of the cable electromagnetic radiation response prediction model of the CNN-LSTM model at the test sample point. The input variables of the test sample point are: the polarization angle of the incident plane wave is -40°, the azimuth angle is -68°, the incident angle is 4°, the peak value of the excitation source field strength is 60kV / m, and the frequency is 0.5GHz; the cable length is 54cm, and the height of the cable from the ground plane is 2.4cm; and the excitation source waveform corresponding to the test sample point.

[0099] On a 6-core 16G personal commercial desktop computer, the full-wave numerical calculation time is about 30s, while the cable electromagnetic radiation response prediction model using a one-dimensional CNN takes about 5ms to make a prediction.

[0100] The electromagnetic coupling response waveform prediction method based on one-dimensional CNN provided by the present invention improves the calculation efficiency and breaks through the previous calculation of the electromagnetic radiation response of cables under the fixed electromagnetic radiation threat that can only be refined to a point, so that the prediction model can better reflect the electromagnetic radiation threat suffered by the cable in actual situations, realize the rapid prediction of the electromagnetic radiation response of the cable, and provide a new solution for the prediction of electromagnetic radiation response. The parameters of the CNN-LSTM model are shown in Table 1.

[0101] Table 1 Model parameters

[0102]

[0103] The present invention proposes a method for predicting electromagnetic coupling response waveform based on one-dimensional CNN, the purpose of which is to determine the physical factors that affect the electromagnetic radiation response waveform of the wiring harness, and obtain the output waveform of the sampling point by sampling a large number of physical factors through full-wave numerical simulation, and divide the simulation data into training set, verification set and test set, and build a CNN-LSTM model, which is trained, tested and verified with simulation data. The method can achieve a rapid prediction of the wiring harness response waveform under electromagnetic radiation that can better reflect the actual situation, and it is easier to obtain the interference signal waveform of the electronic system, saving calculation time and facilitating engineering applications.

[0104] In a second aspect, the present invention provides a terminal device, comprising:

[0105] a memory for storing instructions executed by at least one processor;

[0106] The processor is used to execute the instructions stored in the memory to implement the method described above.

[0107] In a third aspect, the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, and when the computer instructions are executed on a computer, the computer executes the method described above.

[0108] The contents not described in detail in the specification of the present invention belong to the prior art known to the professional and technical personnel in this field.

Claims

1. A method for predicting electromagnetic coupling response waveform based on one-dimensional CNN, characterized in that include: Analyze the factors that affect the electromagnetic radiation response waveform of cables, sample and construct a cable electromagnetic radiation response prediction data set within the transformation range of each factor, and divide the data set into training set, validation set and test set; Construct a CNN-LSTM model, including a one-dimensional convolutional neural network CNN 1d, a long short-term memory network LSTM and a fully connected layer. The one-dimensional convolutional neural network is used to encode the sequence information of the signal and extract the local features of the signal. Then the long short-term memory network LSTM learns the temporal relationship of the signal, and finally the fully connected layer is used for prediction output. The CNN-LSTM model is trained using the training set, and the CNN-LSTM model after each round of training is verified using the verification set. The waveform predicted by the CNN-LSTM model during verification is compared with the electromagnetic radiation response voltage waveform of the cable obtained by full-wave simulation, and the parameters of the CNN-LSTM model are adjusted accordingly to obtain a preliminary cable electromagnetic radiation response prediction model. The test set is input into the preliminary cable electromagnetic radiation response prediction model for prediction. The prediction results are compared with the cable electromagnetic radiation response voltage waveform obtained by full-wave simulation to evaluate the prediction performance of the CNN-LSTM model. The parameters of the CNN-LSTM model are further adjusted to obtain the final cable electromagnetic radiation response prediction model. The physical factors that affect the electromagnetic response waveform of the cable are input into the final cable electromagnetic radiation response prediction model to predict the electromagnetic response waveform of the corresponding cable.

2. The electromagnetic coupling response waveform prediction method based on one-dimensional CNN according to claim 1 is characterized in that: When constructing the cable electromagnetic radiation response prediction dataset, the physical factors that affect the cable electromagnetic response waveform are selected as input variable parameters, and the cable electromagnetic response waveform is selected as the output variable.

3. The electromagnetic coupling response waveform prediction method based on one-dimensional CNN according to claim 1 is characterized in that: Factors that affect the electromagnetic radiation response waveform of cables include: transmission line parameters, excitation source parameters, and excitation source waveform; among them: Transmission line parameters include: cable length, cable height from the ground plane; The excitation source parameters include: transient electric field peak, center frequency, incident angle, polarization angle and azimuth angle.

4. The electromagnetic coupling response waveform prediction method based on one-dimensional CNN according to claim 3 is characterized in that: The specific range of variation of each factor is as follows: The polarization angle ranges from 0 to π; the incident angle ranges from 0 to π / 2; the azimuth angle ranges from 0 to π; the transient electric field peak value of the excitation source ranges from 20 kV / m to 200 kV / m; the center frequency ranges from 0.5 GHz to 2.5 GHz, The cable length ranges from 50 cm to 100 cm, and the cable height from the ground plane ranges from 2 cm to 10 cm.

5. The electromagnetic coupling response waveform prediction method based on one-dimensional CNN according to claim 1 is characterized in that: The specific CNN-LSTM model is: The convolution kernel size of the one-dimensional convolutional neural network CNN 1d is 20 and the number of convolution kernels is 64; Two parallel long short-term memory networks (LSTMs) are used to learn the temporal relationship of signals. The number of neurons in the two parallel long short-term memory networks (LSTMs) is set to 20 and 10 respectively. The feature maps output by the long short-term memory network LSTM are connected through the channel splicing operation to obtain the spliced ​​fusion features; The concatenated fusion features are first input into a fully connected layer with 10 neurons to perform linear transformation, weight the fusion information, and then a fully connected layer with 1 neuron is used to predict the signal waveform.

6. The electromagnetic coupling response waveform prediction method based on one-dimensional CNN according to claim 1, characterized in that: When sampling within the transformation range of each factor to construct the cable electromagnetic radiation response prediction data set, the sampling is uniformly performed within the transformation range of each factor.

7. The electromagnetic coupling response waveform prediction method based on one-dimensional CNN according to claim 1 is characterized in that: Before training, the input data needs to be normalized to the interval [-1, 1]. The loss function used in the training process is the mean square error loss, the optimization algorithm is the Adam algorithm, and the initial learning rate is 0.

01.

8. A terminal device, characterized in that: include: a memory for storing instructions executed by at least one processor; A processor, configured to execute instructions stored in a memory to implement a method according to any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and when the computer instructions are executed on a computer, the computer is enabled to execute the method according to any one of claims 1 to 7.

Images