Array antenna assembly electromagnetic transmission performance prediction method based on BP neural network

Abstract

The application discloses a kind of array antenna assembly electromagnetic transmission performance prediction method based on BP neural network, first in host computer software, the process parameters of electromagnetic transmission performance of array antenna assembly process can be set, by changing KK connector insertion angle, KK connector insertion process and the insertion depth of PCB board H 1, KK connector insertion process and the insertion depth of antenna board H 2 parameter, the change result of KK connector transmission performance S parameter, VSWR parameter under different process parameters in array antenna assembly process can be realized, finally assess the electromagnetic transmission performance of array antenna under the assembly condition of this process parameter;The beneficial effects of the application are as follows: by inputting different assembly process parameters, the electromagnetic transmission performance of array antenna assembly process can be effectively predicted, and by using back propagation, the connection weight between neurons in each layer can be adjusted layer by layer, so that the error is continuously reduced until the set error threshold is reached.

Description

Technical Field

[0001] This invention relates to the field of quality control in the assembly process of array antennas, specifically a method for predicting the electromagnetic transmission performance of array antenna assembly based on a BP neural network. Background Technology

[0002] An array antenna is a converter based on the principle of electromagnetic radiation, transforming guided waves propagating on a transmission line into electromagnetic waves propagating in free space, or vice versa. Currently, antennas are widely used in various fields such as radar detection, navigation, and communication.

[0003] In the assembly process of array antennas, the mechanical assembly precision has a complex correlation with electromagnetic transmission performance. Traditional array antenna assembly methods do not reflect the real-time assembly performance during the assembly process. Traditional simulation methods are time-consuming and cannot meet the requirements for online monitoring of electromagnetic transmission performance during the assembly process of complex products. Furthermore, they cannot dynamically predict the impact of assembly parameters on electromagnetic transmission performance. They can only be repeatedly assembled and adjusted through anechoic chamber power-on testing, which still makes it difficult to achieve optimal performance indicators. The assembly electromagnetic transmission performance of array antennas can only be known after the product is installed. Once critical components are deformed due to improper assembly operations, irreversible damage will occur, ultimately seriously affecting the product's assembly electromagnetic transmission performance. Summary of the Invention

[0004] The purpose of this invention is to provide a method for predicting the electromagnetic transmission performance of array antenna assembly based on a BP neural network, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for predicting the electromagnetic transmission performance of array antenna assembly based on a BP neural network, comprising the following steps:

[0006] Step 1: Construct the array antenna model and set the corresponding process parameters for assembly;

[0007] Step 2: Conduct simulation experiments on the transmission performance of the array antenna and obtain simulation data samples;

[0008] Step 3: Process the simulation sample data;

[0009] Step 4: Establish a BP neural network prediction model for the electromagnetic transmission performance during the array antenna assembly process;

[0010] Step 5: Validate the BP neural network prediction model.

[0011] Furthermore, in step 1, corresponding process parameters are set in the host computer software, including the KK connector insertion angle. KK connector insertion process and PCB board insertion depthH 1. KK connector insertion process and antenna board insertion depth H 2. Simultaneously, the host computer opens a RESTful API interface and receives real-time data from the slave computer's control program regarding the array antenna assembly process and the insertion angle of the KK connector. KK connector insertion process and PCB board insertion depth H 1. KK connector insertion process and antenna board insertion depth H 2.

[0012] Furthermore, in step 2, the array antenna model is imported into simulation software to calculate its electromagnetic transmission performance and obtain simulation data samples. Then, data cleaning is used to remove redundant parameters from the simulation results, extracting parameters that affect the array antenna's transmission performance. Next, a Python script is used to batch read the array antenna assembly simulation data samples. Finally, the input process parameters of the simulation data samples are compared with the dB(S) values ​​of the KK connector in the simulation results. 12 The parameter is combined with the parameters mag(VSWR1) and mag(VSWR2) to generate a sample set.

[0013] Furthermore, the simulation sample data includes dB(S) 11 ) parameters, dB(S 12 ) parameters, dB(S 21 ) parameters, dB(S 22 The parameters include dB(S), mag(VSWR1), and mag(VSWR2), among which the parameters affecting the transmission performance of the array antenna include dB(S). 12 ) parameter, mag (VSWR1) parameter, mag (VSWR2) parameter.

[0014] Furthermore, a normalization function is used to perform global normalization on the array antenna simulation data samples. The normalized data ranges between [0,1]. The normalization function is:

[0015] ,

[0016] in, X ( t ) indicates the first t One sample, X norm ( t ) indicates the first t The result after normalization of each sample X max Represents the maximum value in the sample set. X min This represents the minimum value in the sample set.

[0017] Furthermore, in step 4, the BP neural network prediction model includes an input layer, an output layer, and a hidden layer, and its specific configuration is as follows: Adam is set as the optimizer with a default learning rate of 1e-3, a Dense fully connected layer is set, and Dropout regularization is added to the fully connected layer with a dropout rate of 0.2.

[0018] Furthermore, in step 4, the BP neural network prediction model is initialized, the activation function of the neurons in the neural network is selected, and then the assembly process parameters are input into the neural network prediction model. After forward propagation, the dB(S12) parameter and the predicted values ​​of the mag(VSWR1) and mag(VSWR2) parameters of the KK connector are output. If the error between the predicted value and the true value does not meet the requirements, the error is backpropagated in the neural network. In this process, the connection weights and biases between neurons in each layer are adjusted layer by layer, so that the error is continuously reduced, that is, the predicted value is continuously closer to the true value. This process is repeated until the training reaches the set error threshold.

[0019] Furthermore, in step 5, the BP neural network trained in step 4 is used to predict the electromagnetic transmission performance of the array antenna assembly under different process parameter conditions. The KK connector insertion angle during the array antenna assembly process is received from the lower-level control program via a RESTful API interface. KK connector insertion process and PCB board insertion depth H 1. KK connector insertion process and antenna board insertion depth H 2. Obtain the dB(S) of the KK connector. 12 The output values ​​of the BP neural network prediction model are compared with the simulation values ​​under the same process parameter conditions to verify the prediction accuracy of the BP neural network prediction model.

[0020] Furthermore, the sample set is divided into a training set, a validation set, and a test set. 80% of the data samples are selected as the training set, 10% as the validation set, and 10% as the test set. In step 4, the ratio of the test set to the training set can be adjusted. At the same time, the assembly process parameters are adjusted and multiple training sessions are performed to make the network have a stronger generalization ability.

[0021] Compared with the prior art, the beneficial effects of the present invention are: by inputting different assembly process parameters, the electromagnetic transmission performance of the array antenna assembly process can be effectively predicted, and by using back propagation, the connection weights between neurons in each layer can be adjusted layer by layer, thereby continuously reducing the error until the set error threshold is reached. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart illustrating the electromagnetic transmission performance prediction process of the BP neural network array antenna assembly process according to the present invention.

[0024] Figure 2 This is a visualization of the simulation results of the electromagnetic transmission performance of the array antenna of this invention;

[0025] Figure 3 This is a diagram of the model structure of the BP neural network of the present invention;

[0026] Figure 4 This is an error curvature diagram of the BP neural network training process of this invention;

[0027] Figure 5 This is a comparison chart between the predicted values ​​and simulated values ​​of the BP neural network model of this invention. Detailed Implementation

[0028] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] Please see Figure 1-5 In this embodiment of the invention, the method for predicting the electromagnetic transmission performance of an array antenna assembly based on a BP neural network includes the following steps:

[0030] Step 1: Construct the array antenna model and set the corresponding process parameters for assembly;

[0031] Step 2: Conduct simulation experiments on the transmission performance of the array antenna and obtain simulation data samples;

[0032] Step 3: Process the simulation sample data;

[0033] Step 4: Establish a BP neural network prediction model for the electromagnetic transmission performance during the array antenna assembly process;

[0034] Step 5: Validate the BP neural network prediction model.

[0035] In step 1, the process parameters for the array antenna assembly process are set, including the following steps:

[0036] S1: The process parameters for electromagnetic transmission performance during array antenna assembly can be set in the host computer software: KK connector insertion angle. KK connector insertion process and PCB board insertion depth H 1. KK connector insertion process and antenna board insertion depth H 2; whereby, by changing the insertion angle of the KK connector KK connector insertion process and PCB board insertion depth H 1. KK connector insertion process and antenna board insertion depth H The two parameters allow for the measurement of variations in the S-parameters and VSWR parameters of the KK connector during the array antenna assembly process under different process parameters. This ultimately enables the evaluation of the electromagnetic transmission performance of the array antenna under the assembly conditions specified by the process parameters. It is worth noting that changing the insertion angle of the KK connector... KK connector insertion process and PCB board insertion depth H 1. KK connector insertion process and antenna board insertion depth H The parameters 2 can be used to obtain the results of the changes in the S-parameter and VSWR parameters of the KK connector transmission performance during the assembly process of the array antenna under different process parameters, and finally evaluate the electromagnetic transmission performance of the array antenna under the assembly conditions of the process parameters.

[0037] S2: The host computer system receives data on the array antenna assembly process (KK connector insertion angle) collected by the slave computer control program in real time through an open RESTful API interface. KK connector insertion process and PCB board insertion depth H 1. KK connector insertion process and antenna board insertion depth H 2, its receiving frequency is f s The receiving time interval is .

[0038] In step 2, the array antenna model is imported into simulation software to calculate its electromagnetic transmission performance and obtain simulation data samples. Then, data cleaning is used to remove redundant parameters from the simulation results, extracting parameters that affect the array antenna's transmission performance. Next, a Python script is used to batch read the array antenna assembly simulation data samples. Finally, the input process parameters of the simulation data samples are compared with the dB(S) values ​​of the KK connector in the simulation results. 12 The simulation sample data is generated by merging the ) parameter with the mag(VSWR1) and mag(VSWR2) parameters, and the simulation sample data includes dB(S) 11 ) parameters, dB(S12 ) parameters, dB(S 21 ) parameters, dB(S 22 The parameters include dB(S), mag(VSWR1), and mag(VSWR2), among which the parameters affecting the transmission performance of the array antenna include dB(S). 12 ) parameter, mag (VSWR1) parameter, mag (VSWR2) parameter;

[0039] The steps for cleaning and removing redundant parameters from the simulation results are as follows: First, the open function is used to read the text data, which contains unnecessary data values ​​such as frequency points and dB(S11) parameters. Then, the read_table function is used to filter the dB(S12), mag(VSWR1), and mag(VSWR2) parameters required for training from the text data and save them into an Excel file. This process is repeated until the next text file is reached.

[0040] Next, a normalization function is used to perform global normalization on the array antenna simulation data samples. The normalized data ranges between [0,1]. The normalization function is:

[0041] ,

[0042] in, X ( t ) indicates the first t One sample, X norm ( t ) indicates the first t The result after normalization of each sample X max Represents the maximum value in the sample set. X min This represents the minimum value in the sample set. The sample set is divided into a training set, a validation set, and a test set. 80% of the data samples are selected as the training set, 10% of the data samples are selected as the validation set, and 10% of the data samples are selected as the test set. In step 4, the ratio of the test set to the training set can be adjusted. At the same time, the assembly process parameters are adjusted and multiple training sessions are performed to make the network have a stronger generalization ability.

[0043] In step 4, the BP neural network prediction model is initialized, and the activation functions of the neurons in the neural network are selected. Then, the assembly process parameters are input into the BP neural network prediction model. After forward propagation, the predicted values ​​of the dB(S12) parameter and the mag(VSWR1) and mag(VSWR2) parameters of the KK connector are output. If the error between the predicted value and the true value does not meet the requirements, the error is backpropagated in the neural network. During this process, the connection weights and biases between neurons in each layer are adjusted layer by layer, thereby continuously reducing the error, i.e., the predicted value continuously approaches the true value. This process is repeated until the training reaches the set error threshold. Here, the assembly process parameters are the input variables of the BP neural network prediction model, and the dB(S12) parameter of the KK connector is... 12 The parameters ) and mag(VSWR1) and mag(VSWR2) are the output data of the BP neural network prediction model.

[0044] The BP neural network prediction model includes parameters such as the number of input layer neurons, the number of output layer neurons, the number of hidden layers, and the number of neurons in each hidden layer. The number of input layer neurons is the number of process parameters that affect the electromagnetic transmission performance of the array antenna assembly, and the number of output layer neurons is 1.

[0045] In step 5, the BP neural network trained in step 4 is used to predict the electromagnetic transmission performance of the array antenna assembly under different process parameters. The KK connector insertion angle during the array antenna assembly process is received from the lower-level control program via a RESTful API interface. KK connector insertion process and PCB board insertion depth H 1. KK connector insertion process and antenna board insertion depth H 2. Obtain the dB(S) of the KK connector. 12 The output values ​​of the BP neural network prediction model are compared with the simulation values ​​under the same process parameter conditions to verify the prediction accuracy of the BP neural network prediction model.

[0046] Part of the sample set is shown in Table 1:

[0047]

[0048] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0049] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for predicting the electromagnetic transmission performance of an array antenna assembly based on a BP neural network, characterized in that, Includes the following steps: Step 1: Construct the array antenna model and set the corresponding process parameters for assembly; Step 2: Conduct simulation experiments on the transmission performance of the array antenna and obtain simulation data samples; In step 1, the corresponding process parameters are set in the host computer software, including the KK connector insertion angle. KK connector insertion process and PCB board insertion depth H 1. KK connector insertion process and antenna board insertion depth H 2. Simultaneously, the host computer opens a RESTful API interface and receives real-time data from the slave computer's control program regarding the array antenna assembly process and the insertion angle of the KK connector. KK connector insertion process and PCB board insertion depth H 1. KK connector insertion process and antenna board insertion depth H 2; In step 2, the array antenna model is imported into simulation software to calculate its electromagnetic transmission performance and obtain simulation data samples. Then, data cleaning is used to remove redundant parameters from the simulation results, extracting parameters that affect the array antenna's transmission performance. Next, a Python script is used to batch read the array antenna assembly simulation data samples. Finally, the input process parameters of the simulation data samples are compared with the dB(S) values ​​of the KK connector in the simulation results. 12 The parameter ) is combined with the parameters mag(VSWR1) and mag(VSWR2) to generate a sample set; Step 3: Process the simulation sample data; Step 4: Establish a BP neural network prediction model for the electromagnetic transmission performance during the array antenna assembly process; Step 5: Validate the BP neural network prediction model.

2. The method for predicting the electromagnetic transmission performance of an array antenna assembly based on a BP neural network according to claim 1, characterized in that: The simulation sample data includes dB(S) 11 ) parameters, dB(S 12 ) parameters, dB(S 21 ) parameters, dB(S 22 The parameters include dB(S), mag(VSWR1), and mag(VSWR2), among which the parameters affecting the transmission performance of the array antenna include dB(S). 12 ) parameter, mag (VSWR1) parameter, mag (VSWR2) parameter.

3. The method for predicting the electromagnetic transmission performance of an array antenna assembly based on a BP neural network according to claim 1, characterized in that: The array antenna simulation data samples are globally normalized using a normalization function. The normalized data ranges from [0,1]. The normalization function is as follows: ， in, X ( t ) indicates the first t One sample, X norm ( t ) indicates the first t The result after normalization of each sample X max Represents the maximum value in the sample set. X min This represents the minimum value in the sample set.

4. The method for predicting the electromagnetic transmission performance of an array antenna assembly based on a BP neural network according to claim 1, characterized in that, Step 4 includes the following steps: Step 4.1: Establish a BP neural network prediction model. The BP neural network prediction model includes an input layer, an output layer, and a hidden layer. The number of neurons in the input layer is the number of process parameters that affect the electromagnetic transmission performance of the array antenna assembly. The number of neurons in the output layer is 1. Its specific configuration is as follows: set Adam as the optimizer, with a default learning rate of 1e-3, set a Dense fully connected layer, and add Dropout regularization to the fully connected layer with a dropout rate of 0.

2. Step 4.2: Initialize the BP neural network prediction model, select the activation function of the neurons in the neural network, and then input the assembly process parameters into the neural network prediction model. After forward propagation, output the predicted values ​​of the dB(S12) parameter and the mag(VSWR1) and mag(VSWR2) parameters of the KK connector. If the error between the predicted value and the true value does not meet the requirements, the error is backpropagated in the neural network. In this process, the connection weights and biases between neurons in each layer are adjusted layer by layer, so that the error is continuously reduced, that is, the predicted value is continuously closer to the true value. Repeat this process until the training reaches the set error threshold.

5. The method for predicting the electromagnetic transmission performance of an array antenna assembly based on a BP neural network according to claim 1, characterized in that: In step 5, the BP neural network trained in step 4 is used to predict the electromagnetic transmission performance of the array antenna assembly under different process parameters. The KK connector insertion angle during the array antenna assembly process is received from the lower-level control program via a RESTful API interface. KK connector insertion process and PCB board insertion depth H 1. KK connector insertion process and antenna board insertion depth H 2. Obtain the dB(S) of the KK connector. 12 The output values ​​of the BP neural network prediction model are compared with the simulation values ​​under the same process parameter conditions to verify the prediction accuracy of the BP neural network prediction model.

6. The method for predicting the electromagnetic transmission performance of an array antenna assembly based on a BP neural network according to claim 3, characterized in that: The sample set is divided into a training set, a validation set, and a test set. 80% of the data samples are selected as the training set, 10% as the validation set, and 10% as the test set. In step 4, the ratio of the test set to the training set can be adjusted. At the same time, the assembly process parameters are adjusted and multiple training sessions are performed to make the network have a stronger generalization ability.

Images