AI-Based Phased Array Antenna Calibration Method, System, Device, and Storage Medium

Abstract

The present invention discloses an AI-based phased array antenna calibration method, system, device and storage medium. The method includes: collecting original data of the radiation power corresponding to each radiation unit at different phases; processing the original data to obtain a data set for training a model; constructing an AI-based machine learning model; training the constructed machine learning model with the data set to obtain a trained model; measuring a given phased array antenna to obtain data for inputting into the trained model; inputting the measurement data into the trained model to output the predicted feeding phase of the radiation unit; calculating the difference between the predicted feeding phase and the set phase, and finding the average value of the error, and regarding this average value as the initial phase to which calibration should be performed. Thus, the calibration of the phased array antenna is completed. The present invention uses an AI-based machine learning model to reduce the influence of manufacturing errors and noise on beamforming and achieve relatively high-precision calibration of the phased array antenna.

Description

Technical Field

[0001] The present invention belongs to the technical fields of machine learning and antenna measurement, and particularly relates to an AI-based phased array antenna calibration method, system, and storage medium. Background Art

[0002] A phased array antenna is composed of multiple fixed radiation units, which are interconnected, and phase shifters can be used for each unit to shape the beam to a given angle in space. With the continuous breakthrough of phased array technology, it has been widely applied in fields such as 5G communication, medical imaging, and autonomous driving, greatly enriching our daily lives. Due to errors caused by design, environment, aging, noise, etc. in the phased array antenna, the performance of the phased array antenna is reduced. Therefore, how to calibrate the phased array antenna has also become a research hotspot at present.

[0003] Traditional calibration methods are often classified according to the distance between the measurement position and the antenna position. When measuring in the far field or mid-field, the most commonly used method is the rotation vector method, which calculates the amplitude-phase state of the radiation unit by solving the change in the far-field / mid-field electric field amplitude measured. When measuring in the near field, the single-channel detection method and the aperture field inversion method are often used. In the single-channel detection method, the probe moves in front of each radiation unit, turns on the current radiation unit and turns off other radiation units, and tests the amplitude-phase data of the current radiation unit at different frequencies, and obtains it by comparing the same frequency responses of different radiation units; the aperture field inversion uses the normal antenna measurement process for measurement, and after completion, the amplitude-phase data of each radiation unit on the phased array antenna is obtained by using the back-projection technique and compensating for the probe effect.

[0004] In the rotation vector method, when the number of radiation units is large, the amplitude change caused by a single unit is not obvious, and the calibration time required is also long; the near-field measurement method can only be carried out in a microwave anechoic chamber, with a large amount of data, low efficiency, and high requirements for the synchronization of measurement instruments. Summary of the Invention

[0005] The main purpose of the present invention is to overcome the disadvantages and deficiencies of the prior art, and provide an AI-based phased array antenna calibration method, system, and storage medium, which input the measurement data of fewer times into a trained AI model, and output the feeding phase to which each radiation unit should be calibrated in milliseconds, with the advantages of accurate measurement results, fast measurement speed, and low calculation cost.

[0006] To achieve the above purpose, the present invention adopts the following technical solutions:

[0007] In the first aspect, the present invention provides an AI-based phased array antenna calibration method, including the following steps:

[0008] For a phased array antenna with M*N radiating elements, numerical simulation is performed using simulation software to collect the original data of the radiated power corresponding to each radiating element at different phases;

[0009] Processing the raw data to obtain a data set for training a model;

[0010] Construct AI-based machine learning models;

[0011] Use the data set to train the constructed machine learning model to obtain a trained model;

[0012] Measure a given phased array antenna to obtain data for input into the trained model;

[0013] The measured data is input into the trained model, and the predicted feeding phase of the radiating element is output;

[0014] The difference between the predicted feeding phase and the set phase is calculated, and the average value of the error is calculated. The average value is regarded as the initial phase to which the phased array antenna should be calibrated. At this point, the phased array antenna calibration is completed.

[0015] As a preferred technical solution, the radiation power data refers to the radiation power data measured when the feeding phase of a certain radiation unit is gradually adjusted from 0° to 360° at fixed intervals through a phase modulator, and the feeding phases of other radiation units are set to random values, and noise of different powers is applied through simulation software.

[0016] As a preferred technical solution, the AI-based machine learning model is specifically:

[0017] Establishing a phased array antenna calibration XGBoost model using random initial hyperparameters;

[0018] The XGBoost model consists of a set of base learners and a hyperparameter tuner;

[0019] The base learner uses a regression tree. For each input data, the base learner will output a weight value. The weight values of all base learners are processed by the function to obtain the predicted data.

[0020] The hyperparameter tuner uses the Bayesian optimization method to tune the hyperparameters. During the training process, the hyperparameter tuner adjusts the hyperparameters of the model according to the training situation to reduce the error of the model.

[0021] As a preferred technical solution, the random range of the random hyperparameter is as follows:

[0022] The feature weight reduction coefficient ranges from 0.01 to 0.3;

[0023] The maximum tree depth ranges from 2 to 16;

[0024] The sum of the weights of the leaf nodes ranges from 0.1 to 10;

[0025] The ratio range of sample random sampling is from 0.5 to 1.0;

[0026] The range of the L1 regularization term of the weights is from 0 to 10;

[0027] The range of the L2 regularization term of the weights is from 0 to 10;

[0028] The range of the minimum loss function decrease value is from 0 to 20;

[0029] The range of the feature random sampling ratio is from 0.5 to 1.0;

[0030] As a preferred technical solution, training the constructed machine learning model specifically includes:

[0031] Processing the original data, adjusting all phases of each piece of data in the original data to be between 0° and 360°, to obtain a dataset for training;

[0032] Using randomly generated initial hyperparameters to establish an XGBoost model, inputting the dataset into the model, and using the approximate greedy algorithm to sequentially establish each base learner, and optimizing the output weights of the base learner during this process;

[0033] Using Bayesian optimization to optimize the hyperparameters of the model, and calibrating the model with the new hyperparameters;

[0034] If the error of the calibrated model does not meet the requirements, continue to use Bayesian optimization for hyperparameter tuning until an error that meets the accuracy requirements is obtained.

[0035] As a preferred technical solution, during the training process, Bayesian optimization is used for hyperparameter tuning, specifically:

[0036] Inputting the randomly generated initial hyperparameters into the Gaussian model, and inputting the dataset into the calibrated model, and correcting the model output to make the model closer to the true distribution of the function;

[0037] Selecting the hyperparameter combination points that need to be evaluated next from the corrected Gaussian model, so that the Gaussian model can approach the true distribution of the target function faster;

[0038] If the error of the selected hyperparameter combination meets the set requirements, terminate the algorithm and exit, and output the current hyperparameter combination and the error of the calibrated model;

[0039] If the error of the calibration model does not meet the requirements, the hyperparameter combination point and the error are input into the Gaussian model to correct the Gaussian model, and the step of selecting the next hyperparameter combination point to be evaluated from the corrected Gaussian model is re-executed until an error that meets the accuracy requirements is obtained.

[0040] As a preferred technical solution, the average value of the error is one of the square mean, arithmetic mean, geometric mean, and harmonic mean.

[0041] In a second aspect, the present invention further provides an AI-based phased array antenna calibration system, which is applied to the AI-based phased array antenna calibration method, including a radiation power data acquisition module, a data preprocessing module, a model building module, a model training module, an input data acquisition module, a predicted feed phase acquisition module, and an antenna calibration module;

[0042] The radiation power data acquisition module is used to collect the original data of the radiation power corresponding to each radiation element at different phases by using simulation software to perform numerical simulation for a phased array antenna having M*N radiation elements;

[0043] The data preprocessing module is used to process the raw data to obtain a data set for training the model;

[0044] The model building module is used to construct an AI-based machine learning model;

[0045] The model training module is used to train the constructed machine learning model using the data set to obtain a trained model;

[0046] The input data acquisition module is used to measure a given phased array antenna to obtain data for input into a trained model;

[0047] The predicted feeding phase acquisition module is used to input the measurement data into the trained model and output the predicted feeding phase of the radiating unit;

[0048] The antenna calibration module is used to calculate the difference between the predicted feeding phase and the set phase, and to find the average value of the error, and to regard the average value as the initial phase to be calibrated to, so that the phased array antenna calibration is completed.

[0049] In a third aspect, the present invention further provides an electronic device, the electronic device comprising:

[0050] at least one processor; and,

[0051] a memory communicatively connected to the at least one processor; wherein,

[0052] The memory stores computer program instructions executable by the at least one processor. The computer program instructions are executed by the at least one processor to enable the at least one processor to execute the AI-based phased array antenna calibration method described above.

[0053] In a fourth aspect, the present invention also provides a computer-readable storage medium storing a program, characterized in that when the program is executed by a processor, the AI-based phased array antenna calibration method described above is implemented.

[0054] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0055] (1) The performance of phased array antennas often degrades due to various noise interferences and component manufacturing errors. The present invention uses machine learning to learn the characteristics of these errors from various data and incorporates the impacts caused by these errors into the model, without the need to specifically concern about what impacts these errors individually produce.

[0056] (2) The calibration speed of the present invention is fast. Except for the time required for initial training of the machine learning model, only a small number of measurements are needed for subsequent calibration. After inputting the measured data into the model, the calibration result can be obtained within milliseconds.

[0057] (3) With the method proposed in the present invention, the machine learning model can be freely replaced as long as the model meets the condition of single output, which has strong flexibility. Therefore, the most suitable machine learning model can be selected for training in the face of different phased array antenna usage scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] To more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0059] Figure 1 It is a flowchart of the AI-based phased array antenna calibration method according to an embodiment of the present invention;

[0060] Figure 2 It is a flowchart of training an XGboost model based on Bayesian optimization according to an embodiment of the present invention;

[0061] Figure 3 It is a block diagram of the AI-based phased array antenna calibration system according to an embodiment of the present invention.

[0062] Figure 4This is the structural diagram of the electronic device according to the embodiment of the present invention. Detailed implementation manners

[0063] In order to enable those skilled in the art of this technology to better understand the solution of this application, the following will clearly and completely describe the technical solutions in the embodiments of this application with reference to the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of this application.

[0064] Referring to "embodiments" in this application means that the specific features, structures or characteristics described in connection with the embodiments can be included in at least one embodiment of this application. The phrase appears in various positions in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art understand explicitly and implicitly that the embodiments described in this application can be combined with other embodiments.

[0065] As Figure 1 described, a phased array antenna calibration method based on AI in this embodiment includes the following steps:

[0066] S1. For a phased array antenna with 8 radiation units, use simulation software to perform numerical simulation and collect the radiation power data corresponding to each radiation unit at different phases respectively;

[0067] The simulation software used is System Vue, and the interval of feed phase adjustment is set to 22.5°. Therefore, the set of feed phases for a single radiation unit is 0°, 22.5°, 45°,..., 360°, a total of 17 phases, and the random phases set for other radiation units follow a mathematical expectation of 0 and a variance of. For the simulation of noise, noises with a signal-to-noise ratio of 10 dB, 14 dB, 18 dB, 22 dB, 26 dB, 30 dB are applied respectively, a total of 6 different power noises. Each combination is simulated 1000 times. Therefore, the number of generated data samples is 8 * 17 * 6 * 2000 = 1,632,000. An example of the original data samples generated in this step is shown in Table 1 (taking 10 as an example)

[0068] Table 1

[0069]

[0070]

[0071] S2. Process the described original data to obtain a data set for training the model;

[0072] To ensure that all phases in the dataset are between 0° and 360°, for each radiation unit in Table 1, if the phase value is less than 0°, 360° is added to adjust it to the specified range. An example of the adjusted dataset is shown in Table 2 (taking 10 entries as an example).

[0073] Table 2

[0074]

[0075] S3. Construct an XGBoost machine learning model;

[0076] Establish a phased array antenna calibration XGBoost model using random initial hyperparameters. The XGBoost model consists of a set of base learners and a hyperparameter tuner;

[0077] The base learners use regression trees. For each input data, the base learners will output a weight value, and the weight values of all base learners are processed by a function to obtain the predicted data;

[0078] The hyperparameter tuner uses the Bayesian optimization method to tune the hyperparameters. During the training process, the hyperparameter tuner adjusts the hyperparameters of the model according to the training situation to reduce the error of the model.

[0079] Furthermore, train the constructed machine learning model. Specifically:

[0080] Process the original data by adjusting all phases of each piece of data in the original data to be between 0° and 360° to obtain a dataset for training;

[0081] Establish an XGBoost model using randomly generated initial hyperparameters. Input the dataset into the model and use the approximate greedy algorithm to sequentially establish each base learner. During this process, optimize the output weights of the base learners;

[0082] Use Bayesian optimization to optimize the hyperparameters of the model and calibrate the model with the new hyperparameters;

[0083] If the error of the calibrated model does not meet the requirements, continue to use Bayesian optimization to tune the hyperparameters until an error that meets the accuracy requirements is obtained.

[0084] Even further, the random ranges of the random hyperparameters are shown in Table 3:

[0085] Table 3

[0086]

[0087]

[0088] S4. Use the data set to train the constructed machine learning model to obtain a trained model;

[0089] During the training process, 90% of the data set is used for training, and the remaining 10% of the data set is used to test the model. That is, the data volume of the training set is 1,468,800 groups, and the data volume of the test set is 163,200 groups. During the training process, Bayesian optimization is used to tune the hyperparameters. The steps are as follows:

[0090] (1) Input the randomly generated initial hyperparameters into the Gaussian model, and input the data set into the calibration model to correct the model output to make the model closer to the true distribution of the function.

[0091] (2) Select the hyperparameter combination points that need to be evaluated next from the corrected Gaussian model to enable the Gaussian model to approach the true distribution of the target function faster.

[0092] (3) If the error of the selected hyperparameter combination meets the set requirements, terminate the algorithm and exit, and output the current hyperparameter combination and the error of the calibration model.

[0093] (4) If the error of the calibration model does not meet the requirements, input the hyperparameter combination point and the error into the Gaussian model to correct the Gaussian model, and re-execute step (2) until an error that meets the accuracy requirements is obtained.

[0094] S5. Measure the given phased array antenna to obtain the data for inputting into the model;

[0095] For the given 4*2 phased array, perform the following measurement steps: Set the feeding phase of the radiation unit to 0°, 22.5°, 45°,..., 360°, and at the same time set the phases of other radiation units to 0, and measure the radiation power data at this time. Execute the above steps for each of the 8 radiation units once, so a total of 17*8, that is, 136 measurements are performed.

[0096] S6. Input the measurement data into the model, and the model outputs the predicted feeding phase of the radiation unit;

[0097] Input all the measurement data in step S5 into the model to obtain the corresponding output. The measurement data input into the model is the phase data of 7 radiation units all set to 0° and the measured radiation power, and let the model output the predicted feeding phase of the remaining one radiation unit. Taking the feeding phases of each radiation unit as 22.5°, 0°, 0°, 0°, 0°, 0°, 0°, 0° and the measured power as -29.6 dB as an example, the data input into the model is 0°, 0°, 0°, 0°, 0°, 0°, 0°, -29.6 dB, and the feeding phase of the remaining one radiation unit, 30.8°, is output.

[0098] S7. Calculate the difference between the predicted feeding phase and the set phase, and find the average value of the error. Regard this average value as the initial phase to which calibration should be made;

[0099] Continuing with the example mentioned in step S6, calculate the difference of 8.3° between the predicted feeding phase of 30.8° and 22.5°, denoted as Calculate the root mean square, denoted as which is the initial phase to which the first radiation unit should be calibrated. Similarly, calculate Complete the calibration of the remaining radiation units.

[0100] The present invention utilizes machine learning to learn the characteristics of these errors from various data and incorporates the impacts caused by these errors into the model, without caring about what specific impacts these errors individually produce. Input the measurement data with a small number of times into the trained AI model, and output the feeding phases to which each radiation unit should be calibrated in milliseconds. It has the advantages of accurate measurement results, fast measurement speed, and low calculation cost.

[0101] It should be noted that for the foregoing method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations. However, those skilled in the art should know that the present invention is not limited by the described action sequence, because according to the present invention, certain steps can be performed in other sequences or simultaneously.

[0102] Based on the same idea as the AI-based phased array antenna calibration method in the above embodiments, the present invention also provides an AI-based phased array antenna calibration system, which can be used to execute the above AI-based phased array antenna calibration method. For the sake of convenience of description, in the structural schematic diagram of the AI-based phased array antenna calibration system embodiment, only the parts related to the embodiments of the present invention are shown. Those skilled in the art can understand that the illustrated structure does not constitute a limitation on the device, and it may include more or fewer components than those shown, or combine some components, or have different component arrangements.

[0103] Please refer to Figure 3 In another embodiment of the present application, an AI-based phased array antenna calibration system 100 is provided. The system includes a radiation power data acquisition module 101, a data preprocessing module 102, a model construction module 103, a model training module 104, an input data acquisition module 105, a predicted feeding phase acquisition module 106, and an antenna calibration module 107.

[0104] The radiation power data acquisition module 101 is used to, for a phased array antenna having M*N radiation units, use simulation software to perform numerical simulations to respectively collect the original data of the radiation power corresponding to each radiation unit at different phases;

[0105] The data preprocessing module 102 is used to process the raw data to obtain a data set for training a model;

[0106] The model building module 103 is used to construct an AI-based machine learning model;

[0107] The model training module 104 is used to train the constructed machine learning model using the data set to obtain a trained model;

[0108] The input data acquisition module 105 is used to measure a given phased array antenna to obtain data for inputting into a trained model;

[0109] The predicted feeding phase acquisition module 106 is used to input the measurement data into the trained model and output the predicted feeding phase of the radiating unit;

[0110] The antenna calibration module 107 is used to calculate the difference between the predicted feeding phase and the set phase, and to find the average value of the error, and to regard the average value as the initial phase to be calibrated to, and the phased array antenna calibration is now completed.

[0111] It should be noted that the AI-based phased array antenna calibration system of the present invention corresponds one-to-one to the AI-based phased array antenna calibration method of the present invention. The technical features and beneficial effects described in the above-mentioned embodiment of the AI-based phased array antenna calibration method are applicable to the embodiment of the AI-based phased array antenna calibration method. For specific contents, please refer to the description in the embodiment of the method of the present invention. It will not be repeated here. This is hereby declared.

[0112] In addition, in the implementation of the AI-based phased array antenna calibration system in the above-mentioned embodiment, the logical division of each program module is only an example. In actual applications, the above-mentioned functions can be assigned to different program modules as needed, for example, for the convenience of corresponding hardware configuration requirements or software implementation. That is, the internal structure of the AI-based phased array antenna calibration system is divided into different program modules to complete all or part of the functions described above.

[0113] See also Figure 4 In one embodiment, an electronic device for implementing an AI-based phased array antenna calibration method is provided. The electronic device 200 may include a first processor 201, a first memory 202 and a bus, and may also include a computer program stored in the first memory 202 and executable on the first processor 201, such as an AI-based phased array antenna calibration program 203.

[0114] Among them, the first memory 202 at least includes one type of readable storage medium, and the readable storage medium includes flash memory, mobile hard disk, multimedia card, card-type memory (such as SD or DX memory, etc.), magnetic memory, magnetic disk, optical disc, etc. In some embodiments, the first memory 202 may be an internal storage unit of the electronic device 200, such as the mobile hard disk of the electronic device 200. In some other embodiments, the first memory 202 may also be an external storage device of the electronic device 200, such as a plug-in mobile hard disk, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, etc. equipped on the electronic device 200. Further, the first memory 202 may also include both the internal storage unit and the external storage device of the electronic device 200. The first memory 202 can be used not only to store application software installed on the electronic device 200 and various types of data, such as the code of the phased array antenna calibration program 203 based on AI, but also to temporarily store the data that has been output or will be output.

[0115] In some embodiments, the first processor 201 may be composed of integrated circuits. For example, it may be composed of a single packaged integrated circuit, or may be composed of multiple integrated circuits with the same or different functions packaged, including a combination of one or more Central Processing Units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips, etc. The first processor 201 is the control core (Control Unit) of the electronic device, connecting various components of the entire electronic device through various interfaces and circuits, and by running or executing the programs or modules stored in the first memory 202, and calling the data stored in the first memory 202, to execute various functions of the electronic device 200 and process data.

[0116] Figure 4 Only the electronic device with components is shown. Those skilled in the art can understand that Figure 4 The shown structure does not constitute a limitation on the electronic device 200, and it may include fewer or more components than shown, or combine certain components, or have a different component layout.

[0117] The phased array antenna calibration program 203 based on AI stored in the first memory 202 in the electronic device 200 is a combination of multiple instructions. When running in the first processor 201, it can achieve:

[0118] For a phased array antenna with M*N radiating elements, numerical simulation is performed using simulation software to collect the original data of the radiated power corresponding to each radiating element at different phases;

[0119] Processing the raw data to obtain a data set for training a model;

[0120] Construct AI-based machine learning models;

[0121] Use the data set to train the constructed machine learning model to obtain a trained model;

[0122] Measure a given phased array antenna to obtain data for input into the trained model;

[0123] The measured data is input into the trained model, and the predicted feeding phase of the radiating element is output;

[0124] The difference between the predicted feeding phase and the set phase is calculated, and the average value of the error is calculated. The average value is regarded as the initial phase to which the phased array antenna should be calibrated. At this point, the phased array antenna calibration is completed.

[0125] Furthermore, if the module / unit integrated in the electronic device 200 is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a non-volatile computer-readable storage medium. The computer-readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a mobile hard disk, a magnetic disk, an optical disk, a computer memory, and a read-only memory (ROM).

[0126] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium used in the embodiments provided in the present application can include non-volatile and / or volatile memories. Non-volatile memories can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memories can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM), etc.

[0127] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this specification.

[0128] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications made without departing from the spirit and principle of the present invention shall be equivalent replacement methods and are all included in the protection scope of the present invention.

Claims

1. An AI-based phased array antenna calibration method, characterized in that, The steps include: For a phased array antenna with M*N radiating elements, numerical simulation is performed using simulation software to collect the original data of the radiated power corresponding to each radiating element at different phases; Processing the raw data to obtain a data set for training a model; Construct AI-based machine learning models; Use the data set to train the constructed machine learning model to obtain a trained model; Measure a given phased array antenna to obtain data for input into the trained model; The measured data is input into the trained model, and the predicted feeding phase of the radiating element is output; The difference between the predicted feeding phase and the set phase is calculated, and the average value of the error is calculated. The average value is regarded as the initial phase to which the phased array antenna should be calibrated. At this point, the phased array antenna calibration is completed.

2. The AI-based phased array antenna calibration method according to claim 1, wherein The raw data of the radiation power refers to the raw data of the radiation power measured when the feeding phase of a certain radiation unit is gradually adjusted from 0° to 360° at fixed intervals through a phase modulator, and the feeding phases of other radiation units are set to random values, and noise of different powers is applied through simulation software.

3. The AI-based phased array antenna calibration method according to claim 1, wherein The AI-based machine learning model is specifically: Establishing a phased array antenna calibration XGBoost model using random initial hyperparameters; The XGBoost model consists of a set of base learners and a hyperparameter tuner; The base learner uses a regression tree. For each input data, the base learner outputs a weight value. The weight values of all base learners are processed by the function to obtain the predicted data. The hyperparameter tuner uses the Bayesian optimization method to tune the hyperparameters. During the training process, the hyperparameter tuner adjusts the hyperparameters of the model according to the training situation to reduce the error of the model.

4. The AI-based phased array antenna calibration method according to claim 3, wherein The random range of the random initial hyperparameters is as follows: The feature weight reduction coefficient ranges from 0.01 to 0.3; The maximum tree depth ranges from 2 to 16; The sum of leaf node weights ranges from 0.1 to 10; The proportion of random sampling ranges from 0.5 to 1.0; The L1 regularization term of the weight ranges from 0 to 10; The L2 regularization term of the weight ranges from 0 to 10; The minimum loss function decrease value range is 0~20; The feature random sampling ratio ranges from 0.5 to 1.

0.

5. The AI-based phased array antenna calibration method according to claim 3, wherein The training of the constructed machine learning model is specifically as follows: The original data is processed, and all phases of each data in the original data are adjusted to between 0° and 360° to obtain a data set for training; Use randomly generated initial hyperparameters to build the XGBoost model, input the data set into the model, use the approximate greedy algorithm to build each base learner in turn, and optimize the output weights of the base learners in the process; Use Bayesian optimization to optimize the model's hyperparameters and calibrate the model using the new hyperparameters; If the error of the calibration model does not meet the requirements, continue to use Bayesian optimization to tune the hyperparameters until an error that meets the accuracy requirements is obtained.

6. The AI-based phased array antenna calibration method according to claim 5, wherein Bayesian optimization is used to tune hyperparameters during training, specifically: The randomly generated initial hyperparameters are input into the Gaussian model, and the data set is input into the calibration model to correct the model output so that the model is closer to the true distribution of the function. Select the hyperparameter combination points to be evaluated in the next step from the modified Gaussian model so that the Gaussian model can be close to the true distribution of the objective function; If the error of the selected hyperparameter combination meets the set requirements, the algorithm is terminated and exited, and the hyperparameter combination and the error of the calibration model are output; If the error of the calibration model does not meet the requirements, the hyperparameter combination point and the error are input into the Gaussian model to correct the Gaussian model, and the step of selecting the next hyperparameter combination point to be evaluated from the corrected Gaussian model is re-executed until an error that meets the accuracy requirements is obtained.

7. The AI-based phased array antenna calibration method according to claim 1, characterized in that The average value of the error is one of a square mean, an arithmetic mean, a geometric mean, and a harmonic mean.

8. The AI-based phased array antenna calibration system is characterized in that An AI-based phased array antenna calibration method applied to any one of claims 1-7, comprising a radiation power data acquisition module, a data preprocessing module, a model building module, a model training module, an input data acquisition module, a predicted feed phase acquisition module, and an antenna calibration module; The radiation power data acquisition module is used to collect the original data of the radiation power corresponding to each radiation element at different phases by using simulation software to perform numerical simulation for a phased array antenna having M*N radiation elements; The data preprocessing module is used to process the raw data to obtain a data set for training the model; The model building module is used to construct an AI-based machine learning model; The model training module is used to train the constructed machine learning model using the data set to obtain a trained model; The input data acquisition module is used to measure a given phased array antenna to obtain data for input into a trained model; The predicted feeding phase acquisition module is used to input the measurement data into the trained model and output the predicted feeding phase of the radiating unit; The antenna calibration module is used to calculate the difference between the predicted feeding phase and the set phase, and to find the average value of the error, and to regard the average value as the initial phase to be calibrated to, so that the phased array antenna calibration is completed.

9. An electronic device, characterized in that, The electronic device comprises: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores computer program instructions that can be executed by the at least one processor, and the computer program instructions are executed by the at least one processor so that the at least one processor can execute the AI-based phased array antenna calibration method as described in any one of claims 1-7.

10. A computer-readable storage medium stores a program, characterized in that, When the program is executed by a processor, the AI-based phased array antenna calibration method described in any one of claims 1 to 7 is implemented.

Images