Phased-array antenna calibration method and calibration system

Abstract

The invention relates to the related technical field of antenna calibration, in particular to a phased-array antenna calibration method and system, and the method comprises the steps: building a three-dimensional near-field calibration reference, and simulating an initial error matrix; wave beam pointing deviation is determined through grouped excitation, and space angle error components are matched; generating a dynamic calibration weight matrix; and deploying distributed calibration nodes, and carrying out mutual coupling effect bias measurement and calibration decision. The technical problems that a static compensation coefficient is difficult to adapt to a dynamic working condition, thermal drift, mechanical deformation and electromagnetic interference cannot be effectively separated, and the stability of a communication link is influenced are solved, an initial error matrix is prepared by establishing a three-dimensional near-field calibration reference and combining phase difference and amplitude errors, collaborative optimization of grouped excitation and angle errors is carried out, and the stability of the communication link is improved. And in combination with the least square method and the deep separable convolutional neural network, environment disturbance is effectively separated and dynamically compensated, distributed calibration nodes are deployed, the mutual coupling calibration efficiency between array elements is improved, and the technical effect of influencing the stability of a communication link is further guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field related to antenna calibration, and in particular to a phased array antenna calibration method and a calibration system. Background Art

[0002] As the communication frequency band extends to millimeter waves and terahertz, the number of phased array antenna elements has increased dramatically. The expansion of the array scale has led to a significant increase in the mutual coupling effect between elements and the influence of environmental disturbances. Conventional static calibration methods cannot effectively separate the coupling effects of thermal drift, mechanical deformation and electromagnetic interference, resulting in the failure of calibration results with changes in working conditions, seriously affecting the quality of the communication link. In addition, the signal interference between adjacent sub-arrays is not effectively suppressed. In the context of multi-base station collaboration and complex electromagnetic environments, it will further aggravate signal aliasing and distortion, greatly reducing the beamforming effect, causing the bit error rate of the communication system to rise sharply, and even causing the communication link to be interrupted, which greatly limits the application efficiency of phased array antennas in complex scenarios.

[0003] In summary, the existing technology has technical problems such as the difficulty of adapting the static compensation coefficient to dynamic working conditions, the inability to effectively separate thermal drift, mechanical deformation and electromagnetic interference, and the impact on the stability of the communication link. Summary of the Invention

[0004] This application provides a phased array antenna calibration method and calibration system, aiming to solve the technical problems in the existing technology that the static compensation coefficient is difficult to adapt to dynamic working conditions, cannot effectively separate thermal drift, mechanical deformation and electromagnetic interference, and affects the stability of the communication link.

[0005] In view of the above problems, the technical solution to implement this application is: In a first aspect, the present application provides a phased array antenna calibration method, wherein the method includes: establishing a three-dimensional near-field calibration benchmark based on the phased array antenna configuration information, receiving the phase difference and amplitude error of the antenna unit, and formulating an initial error matrix; grouping the phased antenna elements, determining the beam pointing deviation of each group of phased antenna elements through the far-field direction vector, and configuring the spatial angle error component; based on the initial error matrix and the spatial angle error component, using the least squares method to determine the compensation coefficient, extracting the environmental disturbance compensation factor through a deep separable convolutional neural network, separating the coupling effects of thermal drift, mechanical deformation and electromagnetic interference, and generating a dynamic calibration weight matrix; according to the dynamic calibration weight matrix, in combination with the beam pointing accuracy requirements, deploying distributed calibration nodes, iteratively optimizing the calibration parameters in a multi-base station collaborative scenario, and at the same time, performing mutual coupling effect bias measurement and calibration decision-making.

[0006] Preferably, a first adaptive calibration index is configured according to the amplitude mutual coupling effect between the phased array antenna elements; a second adaptive calibration index is configured according to the phase mutual coupling effect between the phased array antenna elements; a test period is determined based on the first adaptive calibration index and the second adaptive calibration index; within the test period, the element phase difference is determined according to the amplitude of the received signal of the signal transmitted independently by each phased array antenna element, and the mutual coupling effect bias measurement and calibration decision are performed: the coupling coefficient between adjacent array elements is calculated by port S parameter measurement, the bias amount of each phased array antenna element is determined, and fuzzy decision making is used to perform adaptive calibration.

[0007] Preferably, the phased antenna array plane is divided into N independent sub-arrays, each independent sub-array contains P phased antenna array elements, and the isolation between the independent sub-arrays meets the isolation limit conditions; according to the timing control logic, the N independent sub-arrays are polled and excited by high-speed radio frequency switches.

[0008] Preferably, the adjacent sub-array excitation intervals corresponding to the N independent sub-arrays meet a time interval limitation condition; at the same time, a preset initial phase weight is applied to each independent sub-array according to the deviation between the independent sub-array beam pointing direction and the target direction.

[0009] Preferably, a first time factor is determined based on the operating frequency of the phased array antenna; a free space propagation model is used to determine the time required for the first adjacent subarray signal to be transmitted, decay in energy, and affect the acquisition of the second adjacent subarray signal in accordance with the set interference threshold, thereby obtaining a second time factor; a third time factor corresponding to the switching time of the high-speed RF switch and a fourth time factor corresponding to the signal processing delay are determined, and the time interval limiting condition is set in combination with the first time factor and the second time factor.

[0010] Preferably, the pointing deviation value between the independent sub-array beam pointing and the target direction is decomposed into an azimuth error component and a pitch angle error component; and the space angle error component is configured through the azimuth error component and the pitch angle error component.

[0011] Preferably, according to the number and arrangement of phased array antenna elements, a spherical scanning frame with a radius of U times the central wavelength of the working frequency band is built in a microwave darkroom; based on the spherical scanning frame, a vector network analyzer and a double-ridged horn probe are configured to collect full-airspace field strength data through a preset scanning path; with the center of the array surface as the coordinate origin, the physical position parameters of each phased array antenna element are associated with the full-airspace field strength data to obtain a calibration reference database.

[0012] Preferably, the objective function is configured as the mean square error between the beam pattern and the target beam pattern, and the constraint condition of the objective function is the adjustment range of the compensation coefficient; based on the objective function, the weighted least squares method is used to determine the overdetermined equation group, and the weight coefficient is dynamically allocated according to the far and near field of view attenuation characteristics of each phased antenna array element from the center of the array surface.

[0013] Preferably, the mapping relationship between environmental parameters and amplitude and phase errors in historical calibration data is used as a training sample; a separation thermal drift branch, a mechanical deformation branch, and an electromagnetic interference branch are set in the output layer of a deep separable convolutional neural network to decouple the coupling effect and formulate the dynamic calibration weight matrix.

[0014] In a second aspect, the present application provides a phased array antenna calibration system, wherein the system includes: an initial error matrix formulation module: based on the phased array antenna configuration information, a three-dimensional near-field calibration benchmark is established, the phase difference and amplitude error of the receiving antenna unit are received, and the initial error matrix is ​​formulated; a group excitation module: the phased antenna array elements are grouped to excite, and the beam pointing deviation of each group of phased antenna array elements is determined through the far-field direction vector, and the spatial angle error component is configured; a coupling analysis module: based on the initial error matrix and the spatial angle error component, the compensation coefficient is determined using the least squares method, and the environmental disturbance compensation factor is extracted through a deep separable convolutional neural network, the coupling effects of thermal drift, mechanical deformation and electromagnetic interference are separated, and a dynamic calibration weight matrix is ​​generated; a calibration decision module: according to the dynamic calibration weight matrix, distributed calibration nodes are deployed in combination with the beam pointing accuracy requirements, and calibration parameters are iteratively optimized in a multi-base station collaborative scenario. At the same time, mutual coupling effect bias measurement and calibration decision are performed.

[0015] In summary, one or more technical solutions provided in this application establish a three-dimensional near-field calibration benchmark, combine the phase difference and amplitude error to formulate the initial error matrix, perform collaborative optimization of group excitation and angle error, combine the least squares method with the deep separable convolutional neural network, effectively separate environmental disturbances and dynamically compensate, deploy distributed calibration nodes, improve the mutual coupling calibration efficiency between array elements, and thus ensure the technical effect that affects the stability of the communication link. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A flow chart of a phased array antenna calibration method is provided for this application.

[0017] Figure 2 A schematic structural diagram of a phased array antenna calibration system is provided for this application.

[0018] Description of the reference numerals: initial error matrix formulation module M100, group excitation module M200, coupling analysis module M300, calibration decision module M400. DETAILED DESCRIPTION

[0019] Example 1: The present application will be described in detail below with reference to the accompanying drawings. Figure 1 As shown, the present application provides a phased array antenna calibration method, wherein the method includes: S1: Based on the phased array antenna configuration information, a three-dimensional near-field calibration benchmark is established to receive the phase difference and amplitude error of the antenna unit and formulate the initial error matrix. S2: The phased array elements are grouped and excited. The beam pointing deviation of each group of phased array elements is determined through the far-field direction vector to configure the spatial angle error component.

[0020] Specifically, the three-dimensional near-field calibration benchmark refers to a three-dimensional spatial calibration reference system constructed in the near-field area with the phased array antenna face as the center, which is used to accurately measure and characterize the radiation characteristics of the antenna unit; the initial error matrix is ​​formulated based on the collected antenna unit phase difference and amplitude error, and these error information are integrated into a matrix form through spherical wave expansion for subsequent processing; group excitation refers to dividing the phased antenna array elements into multiple groups according to certain rules, and applying excitation signals to each group in turn to generate a radiation field; the far-field direction vector is a unit vector describing the antenna radiation direction, which is used to determine the beam pointing in far-field measurements; the beam pointing deviation refers to the deviation angle between the actual beam pointing and the ideal beam pointing; the spatial angle error component is to decompose the beam pointing deviation into error components in directions such as azimuth and pitch angle according to the spatial coordinate system, so as to facilitate more detailed analysis and calibration.

[0021] Execution steps: In the process of establishing a three-dimensional near-field calibration benchmark and formulating the initial error matrix, according to the phased array antenna configuration information, such as the number of array elements and arrangement, a spherical scanning frame is built in the microwave darkroom, and a vector network analyzer and a double-ridged horn probe are configured to collect full-space field strength data. With the center of the array surface as the coordinate origin, the physical position parameters of each phased antenna array element are associated with the collected field strength data to construct a calibration benchmark database, providing a reliable basis for subsequent error analysis; through the spherical wave expansion method, the phase difference and amplitude error of the antenna unit are integrated into the initial error matrix, and the element value represents the amplitude and phase error of the corresponding array element. The initial error matrix can fully reflect the initial error state of the antenna array element and provide basic data for subsequent calibration compensation.

[0022] The phased antenna array elements are excited in groups. Furthermore, the phased antenna array plane is divided into N independent sub-arrays, each containing P phased antenna elements. The isolation between independent sub-arrays must meet the isolation limit conditions (such as isolation ≥ 30dB) to reduce mutual interference between sub-arrays. According to the timing control logic, the N independent sub-arrays are polled and excited in sequence using high-speed RF switches. The excitation duration of each sub-array is T seconds. For example, if the phased array antenna has 100 elements, divided into 4 sub-arrays, each containing 25 elements, at an operating frequency of 10GHz, each sub-array is excited for 1 second through the switching of the high-speed RF switch, and the excitation of all sub-arrays is completed in sequence.

[0023] During the excitation process, the far-field radiation pattern of each sub-array is obtained through far-field measurement equipment to determine the beam pointing deviation of each group of phased antenna elements. The deviation is decomposed into azimuth error and elevation error components through coordinate transformation, and then the spatial angle error component is configured to accurately locate the beam pointing of each group of elements. This provides key spatial error information for the subsequent dynamic calibration weight matrix generation, enabling the calibration process to compensate for specific spatial angle errors, effectively improving calibration accuracy and antenna performance.

[0024] S3: Based on the initial error matrix and the spatial angle error component, the least squares method is used to determine the compensation coefficient, and the environmental disturbance compensation factor is extracted through a deep separable convolutional neural network. The coupling effects of thermal drift, mechanical deformation and electromagnetic interference are separated to generate a dynamic calibration weight matrix; S4: According to the dynamic calibration weight matrix, distributed calibration nodes are deployed in combination with the beam pointing accuracy requirements, and the calibration parameters are iteratively optimized in a multi-base station collaborative scenario. At the same time, mutual coupling effect bias measurement and calibration decision-making are performed.

[0025] Specifically, the initial error matrix is ​​a matrix integrated based on the collected antenna unit phase difference and amplitude error through mathematical methods such as spherical wave expansion, which is used to characterize the initial error state of the antenna array element; the spatial angle error component is to decompose the beam pointing deviation into error components in directions such as azimuth and pitch angle according to the spatial coordinate system, so as to facilitate more detailed analysis and calibration; the least squares method is used to find the best fitting parameters by minimizing the sum of squares of errors; the deep separable convolutional neural network refers to the combination of deep convolution layers and point convolution layers, which can efficiently extract features from multidimensional data.

[0026] The dynamic calibration weight matrix is ​​a matrix calibrated according to the environmental disturbance compensation factor and the initial error, and is used to adjust the amplitude and phase states of the antenna array elements in real time to adapt to the dynamically changing environment; distributed calibration nodes refer to multiple calibration equipment nodes deployed in the phased array antenna system, which can collect the working status parameters of the array elements in the area in real time; multi-base station collaboration refers to the information exchange and collaborative work between multiple base stations to achieve more extensive calibration and optimization; mutual coupling effect refers to the mutual influence between antenna array elements due to electromagnetic coupling, which will cause the amplitude and phase characteristics of the array elements to change, affecting the overall performance of the antenna; bias measurement is the process of quantitatively measuring the amplitude and phase bias of the array elements caused by mutual coupling effects, etc.; calibration decision is the process of determining whether calibration is needed and whether to perform calibration based on the bias measurement results and calibration requirements.

[0027] Error analysis is performed based on the initial error matrix and spatial angle error components. This analysis involves information such as antenna element phase difference, amplitude error, and beam pointing deviation. A phased array antenna has N elements, and the spatial angle error component of the initial error matrix is ​​an N×2 matrix (corresponding to azimuth and elevation angle errors). Based on the error data, the least squares method is used to determine the compensation coefficients. By defining the objective function as the mean square error between the actual and ideal beam patterns and constraining the adjustment range of the compensation coefficients (amplitude 0.5-2 times, phase -180-180°), a weighted least squares method is used to solve the overdetermined system of equations. The weight coefficients are dynamically assigned based on the near-far field of view attenuation characteristics of the elements from the center of the array. The compensation coefficients are then optimized using the Levenberg-Marquardt iterative method. The calculation is terminated when the iterative residual is ≤1e-6, generating an initial compensation table. This method accurately determines the compensation coefficients, providing key parameters for subsequent calibration. This effectively improves calibration accuracy, reduces the compensated beam pointing error to an acceptable range, and significantly enhances the antenna's radiation performance.

[0028] At the same time, the environmental disturbance compensation factor is extracted through a deep separable convolutional neural network. Furthermore, a neural network containing 3 deep convolutional layers and 2 point convolutional layers is constructed, and multi-dimensional time series data of environmental parameters (such as temperature range -40℃-85℃, vibration frequency 0Hz-50Hz, electromagnetic interference intensity -100dBm-0dBm) are input; the mapping relationship between environmental parameters and amplitude and phase errors in historical calibration data is used as training samples, and the Adam optimizer is used for model training, and the loss function is the root mean square error.

[0029] Three branches are set up in the network output layer to output the compensation factors corresponding to thermal drift, mechanical deformation, and electromagnetic interference respectively, to achieve the decoupling of the coupling effect, collect environmental data in real time and input the model, synchronously update the compensation factors, and improve the compensation accuracy. This effectively improves the stability and reliability of the antenna system in a dynamic environment, so that the calibration results can better adapt to changes in the actual working environment.

[0030] Distributed calibration nodes are deployed according to the dynamic calibration weight matrix and multi-base station collaborative optimization is performed. Furthermore, multiple calibration nodes are configured according to the beam pointing accuracy index. Synchronous communication between nodes is achieved through optical fiber (delay ≤ 10ns) to ensure the efficiency and real-time performance of data transmission. Each node integrates an amplitude and phase detector and processor, which can collect the working status parameters of the array elements in the area in real time, such as phase, amplitude and other information. A master-slave calibration architecture is set up. The master node summarizes the data of each slave node and generates global calibration instructions. The slave node performs local parameter adjustment to achieve refined calibration of the antenna array elements.

[0031] In a multi-base station collaborative scenario, timestamp alignment is used to synchronize cross-base station calibration parameters, ensuring consistency and coordination of antenna calibration across multiple base stations. Mutual coupling effect bias measurement and calibration decisions are simultaneously performed, and an array element mutual coupling model is established. The coupling coefficient between adjacent array elements is measured through port S parameters to generate a mutual coupling matrix. The bias of each array element is calculated based on the mutual coupling matrix, and targeted calibration is triggered when the bias exceeds the limit. Fuzzy decision-making is used to integrate beam pointing error, mutual coupling bias, and environmental disturbance level to generate calibration priorities. Calibration is performed on the highest priority calibration item to ensure antenna system availability. In the above steps, by rationally deploying calibration nodes and performing distributed collaborative optimization, calibration efficiency and accuracy are improved, effectively resolving the mutual coupling effect between array elements and multi-base station collaboration issues, further enhancing the performance and stability of phased array antenna systems in complex scenarios, enabling them to better meet the high-precision and high-reliability requirements of modern communication systems.

[0032] Furthermore, before performing mutual coupling effect bias measurement and calibration decision, the present application method includes: A first adaptive calibration indicator is configured based on the amplitude mutual coupling effect between phased array antenna elements; a second adaptive calibration indicator is configured based on the phase mutual coupling effect between phased array antenna elements; a test period is determined based on the first adaptive calibration indicator and the second adaptive calibration indicator; within the test period, the element phase difference is determined based on the received signal amplitude of the signal transmitted independently by each phased array element, and the mutual coupling effect offset measurement and calibration decision are performed: the coupling coefficient between adjacent array elements is calculated through port S parameter measurement, the offset of each phased array element is determined, and fuzzy decision making is used to perform adaptive calibration.

[0033] Specifically, the adaptive calibration index is based on the mutual coupling effect characteristics between phased array antenna elements, and dynamically measures the degree of deviation between the antenna phase and amplitude state and the ideal state. Specifically, the first adaptive calibration index reflects the impact of the amplitude mutual coupling effect on the amplitude of each element, and the second adaptive calibration index reflects the impact of the phase mutual coupling effect on the phase of each element. By real-time monitoring and analysis of the mutual coupling effect between elements, the adaptive calibration index can dynamically reflect the error state of the antenna system and provide a quantitative basis for subsequent calibration decisions.

[0034] Port S-parameters, or scattering parameters, are a parameter system that describes the characteristics of RF and microwave networks. In phased array antennas, port S-parameters are used to quantify the signal transmission and reflection characteristics between antenna ports. By measuring S-parameters, the coupling coefficient between adjacent array elements can be calculated, reflecting the degree of electromagnetic coupling between the elements. The coupling coefficient is a key indicator of the strength of the mutual coupling effect between array elements. A larger value indicates stronger mutual coupling. For example, when the coupling coefficient between two elements is large, the signal from one element will significantly affect the amplitude and phase characteristics of the other element, resulting in an increase in the mutual coupling effect offset.

[0035] Fuzzy decision-making is a decision-making method based on fuzzy logic, which is used to process uncertain information. In phased array antenna calibration, the fuzzy decision algorithm comprehensively considers multiple factors such as beam pointing error, mutual coupling offset, and environmental disturbance level to generate calibration priorities. These factors are often difficult to describe with precise mathematical models. Fuzzy decision-making transforms these factors into fuzzy sets through fuzzy rules and membership functions, thereby achieving a comprehensive evaluation of the priority of calibration items, effectively handling multi-factor and uncertain decision-making problems, and ensuring that reasonable calibration decisions are made in complex environments.

[0036] Execution steps: When configuring adaptive calibration indicators, two independent indicator systems are constructed based on the amplitude mutual coupling effect and phase mutual coupling effect of the phased array antenna elements. Furthermore, for the amplitude mutual coupling effect, the amplitude offset of each element is determined by measuring the coupling coefficient between the elements, thereby constructing a first adaptive calibration indicator. For the phase mutual coupling effect, a second adaptive calibration indicator is constructed based on the coupling coefficient and phase difference measurement. For example, if a phased array antenna has N elements, the coupling coefficients between each element and the other elements form an N×N mutual coupling matrix. By analyzing the eigenvalues ​​and eigenvectors of this matrix, the strength and impact range of the mutual coupling effect can be quantified. For example, in a typical phased array antenna, if the coupling coefficient amplitude between two adjacent elements is greater than 0.1, it is considered that there is a significant mutual coupling effect between the two adjacent elements and calibration is required. Based on these two calibration indicators, combined with historical calibration data and the operating status of the antenna, a machine learning algorithm is used to predict the optimal test cycle. The optimal test cycle for the phased array antenna is once an hour to ensure that changes in the mutual coupling effect are captured in a timely manner, thereby achieving adaptive calibration.

[0037] During the test cycle, each phased antenna array element is controlled to transmit a signal of known amplitude and phase. Other elements receive the signal and measure its amplitude and phase. Based on the amplitude change of the received signal, the element phase difference is determined using algorithms such as the mutual coupling model and the least squares method. For example, if an element transmits a signal with an amplitude of V0 and a phase of θ0, and an adjacent element receives a signal with an amplitude of V1 and a phase of θ1, the coupling coefficient and phase difference between the element and the other elements can be determined using the ratio of V1 to V0 and the difference between θ1 and θ0, combined with the mutual coupling model.

[0038] Through the port S parameter measurement, the coupling coefficient matrix between adjacent array elements is determined, and the offset of each phased antenna array element is determined. The element S in the coupling coefficient matrix is ij It represents the coupling effect of the i-th array element on the j-th array element. By analyzing S ij The amplitude and phase of the beam can be used to construct an offset model and quantify the offset of each element. Furthermore, if the offset of an element exceeds a preset threshold (e.g., 5%), a targeted calibration process is triggered. A fuzzy decision-making algorithm integrates multiple factors, such as beam pointing error, mutual coupling offset, and environmental disturbance level, to generate a calibration priority. Through real-time monitoring and analysis of mutual coupling effects, the calibration strategy is dynamically adjusted to ensure that the antenna system maintains high accuracy and stability in complex environments.

[0039] Furthermore, the phased antenna array elements are excited in groups, and the method of the present application includes: The phased antenna array plane is divided into N independent sub-arrays, each independent sub-array includes P phased antenna array elements, and the isolation between the independent sub-arrays meets the isolation limit condition; according to the timing control logic, the N independent sub-arrays are polled and excited by high-speed radio frequency switches.

[0040] Specifically, an independent subarray refers to a subarray with high isolation from each other formed by dividing the phased antenna array surface according to certain rules. Each independent subarray contains a certain number of phased antenna elements. Such a division helps to reduce mutual interference between subarrays and facilitates separate excitation and calibration; the isolation limitation condition refers to the minimum isolation requirement between subarrays specified to ensure that each independent subarray does not interfere with each other during operation, usually expressed in decibels (dB) to ensure that the signal leakage between subarrays is within an acceptable range; the timing control logic refers to the rules for controlling each independent subarray according to a pre-set time sequence and logical relationship, which is used to coordinate the action of high-speed RF switches so that each subarray can be excited in turn to avoid signal conflicts and mutual interference; polling excitation refers to the excitation of each independent subarray in turn according to a certain timing control logic, so that each subarray transmits a signal within a specified time so that its performance can be measured and analyzed separately.

[0041] Execution steps: After dividing the phased array antenna plane into N independent sub-arrays and performing polling excitation, further determine the number of sub-arrays N and the number of array elements P in each sub-array based on the array plane layout and performance requirements of the phased array antenna. For example, if the phased array antenna plane has 100 array elements, it is divided into 4 independent sub-arrays, each containing 25 array elements. During the division process, the arrangement and wiring of the array elements are optimized to ensure that the isolation between the independent sub-arrays meets the isolation limit conditions. For example, by increasing the spacing between the sub-arrays and adopting electromagnetic shielding measures, the isolation between the sub-arrays is increased to above 35dB, effectively reducing signal interference between the sub-arrays.

[0042] According to the timing control logic, a high-speed RF switch is used to perform polling excitation on these N independent subarrays. If the timing control logic requires that the excitation time for each subarray is 1ms and the excitation interval between adjacent subarrays is 0.5ms, the switching time of the high-speed RF switch must be less than 1μs to meet the timing requirements. Furthermore, by controlling the switching of the high-speed RF switch, an excitation signal is applied to each independent subarray in turn, causing it to transmit an RF signal. For example, after the first subarray is excited for 1ms, the high-speed RF switch is quickly switched to the second subarray, and it is excited after an interval of 0.5ms. This process is repeated to complete the polling excitation of all subarrays.

[0043] This polling excitation method allows the acquisition of radiation characteristic data for each subarray, including its directivity pattern, phase difference, and amplitude error, to analyze the performance and error of each subarray. This provides a grouped data foundation for subsequent error analysis and calibration parameter optimization, enabling the calibration process to be optimized for each subarray's specific conditions. This improves calibration efficiency and accuracy, while also facilitating the deployment of distributed calibration nodes and multi-base station collaborative optimization.

[0044] Furthermore, the present application method also includes: The adjacent sub-array excitation intervals corresponding to the N independent sub-arrays meet the time interval limitation condition; at the same time, a preset initial phase weight is applied to each independent sub-array according to the deviation between the independent sub-array beam pointing and the target direction.

[0045] Specifically, the adjacent subarray excitation interval refers to the time interval between two adjacent subarrays when polling and exciting each independent subarray. This time interval needs to meet the time interval limitation condition, that is, the minimum time interval requirement determined by factors such as the operating frequency of the phased array antenna, the signal propagation characteristics, and the switching speed of the equipment, to ensure that the high-speed RF switch has completed the switching and the latter subarray can work normally before the signal transmitted by the previous subarray interferes with the latter subarray; the preset initial phase weight refers to a phase compensation value set in advance for each independent subarray based on the deviation between the independent subarray beam pointing and the target direction. This phase weight is used to adjust the initial phase of its beam when exciting the subarray, so that the beam pointing is closer to the target direction, reducing the beam pointing deviation and improving the antenna's directivity performance.

[0046] Execution steps: To ensure that the excitation interval between adjacent subarrays meets the time interval constraint, a first time factor is further determined based on the operating frequency of the phased array antenna. For example, if the operating frequency is 10 GHz, the corresponding signal period is 0.1 μs. The free-space propagation model is used to calculate the distance from signal transmission to energy decay until the impact on adjacent subarray signal acquisition meets the set interference threshold value, and then the second time factor is determined. For example, if the interference signal is required to attenuate to below -60 dBm, it is found that at this operating frequency, the attenuation requirement is met when the signal propagation distance is 2 m. The corresponding second time factor is the time required for the signal to propagate 2 m in the air. The third time factor corresponding to the switching time of the high-speed RF switch and the fourth time factor corresponding to the signal processing delay are determined. Based on the above time factors, the excitation interval between adjacent subarrays is set to ensure that the next subarray is excited after the high-speed RF switch is switched and the signal interference is attenuated to an acceptable level, avoiding signal interference between subarrays and ensuring the accuracy of the measurement data.

[0047] At the same time, a preset initial phase weight is applied to each independent subarray based on the deviation between the beam pointing of the independent subarray and the target direction. Based on the relationship between beam pointing and phase weight, the preset initial phase weight required for the subarray is determined. Specifically, if measurement reveals that the beam pointing of an independent subarray deviates from the target direction by 2° in azimuth and 1.5° in elevation, each 1° deviation in azimuth corresponds to a phase adjustment of π / 90 radians (assuming a wavelength of λ and an array element spacing of λ / 2). The same applies to elevation. Therefore, the independent subarray needs to adjust its phase by 2×π / 90 radians in azimuth and 1.5×π / 90 radians in elevation. By adding corresponding phase compensation to the excitation signal, the beam pointing of the independent subarray is brought closer to the target direction. In the above steps, the preset initial phase weight reduces beam pointing deviation, improves the antenna's pattern gain and anti-interference capability, and provides more accurate initial conditions for subsequent dynamic calibration weight matrix generation and distributed calibration node optimization, ensuring the efficiency of the calibration process and the accuracy of the calibration results.

[0048] Furthermore, the adjacent sub-array excitation intervals corresponding to the N independent sub-arrays meet the time interval limitation condition. The method of the present application includes: A first time factor is determined based on the operating frequency of the phased array antenna. A free-space propagation model is used to determine the time required for the signal of the first adjacent subarray to be transmitted, decay in energy, and affect the signal acquisition of the second adjacent subarray to meet the set interference threshold, thereby obtaining a second time factor. A third time factor corresponding to the switching time of the high-speed RF switch and a fourth time factor corresponding to the signal processing delay are determined. The first and second time factors are combined to set a time interval limitation condition.

[0049] Specifically, the first time factor refers to the signal period or related time parameters calculated based on the operating frequency of the phased array antenna, which is used to evaluate the basic time characteristics of the signal; the second time factor refers to the time required for the adjacent sub-array signals to be transmitted and attenuated to the interference threshold value based on the free space propagation model, which is used to evaluate the attenuation time during the signal propagation process; the third time factor refers to the time required for the high-speed RF switch to complete switching, which is a key parameter of the equipment performance; the fourth time factor refers to the delay time generated in the signal processing process, including the sum of the time of signal detection, amplification, filtering and other processing links; the time interval limitation condition refers to the combination of all the above time factors to set the minimum time interval that must be met between adjacent sub-array excitations to ensure the accuracy of signal acquisition and avoid mutual interference between sub-arrays.

[0050] Execution steps: According to the formula f=1 / T, determine the signal period and the size of the first time factor. For example, at a working frequency of 10 GHz, the signal period is 0.1 μs. According to the free space propagation model and the interference threshold, calculate the signal propagation distance and convert it into time. For example, at a frequency of 10 GHz, the electromagnetic wave propagation speed C is 3×10 8 m / s, the time corresponding to the signal propagation distance of 2m is about 2 / (3×10 8 m / s), i.e. 6.67ns; if the switching time of the high-speed RF switch is 0.1μs; the signal processing delay is 0.2μs; the first time factor (0.1μs), the second time factor (6.67ns), the third time factor (0.1μs) and the fourth time factor (0.2μs) are added together to obtain a total time interval of 0.4667μs, which is used as the time interval limit for the excitation of adjacent sub-arrays.

[0051] In actual application, this also includes checking whether the calculated time interval can effectively avoid signal interference between subarrays, which can be verified through experiments or simulations; verifying whether the switching time and signal processing delay of the high-speed RF switch meet the actual device performance; and ensuring the accuracy and reliability of signal acquisition under the time interval constraints. Preferably, through these steps, the appropriate setting of the excitation interval between adjacent subarrays is ensured, signal interference is avoided, and calibration accuracy is improved.

[0052] Furthermore, the beam pointing deviation of each group of phased antenna array elements is determined, and the spatial angle error component is configured. The method of the present application includes: According to the pointing deviation value between the independent sub-array beam pointing and the target direction, the azimuth error component and the elevation error component are decomposed; and the space angle error component is configured through the azimuth error component and the elevation error component.

[0053] Specifically, the pointing deviation value refers to the deviation between the actual pointing of the beam and the target pointing, expressed as an angle; the azimuth error component refers to the angle between the beam pointing and the target direction in the horizontal plane; the pitch error component refers to the angle between the beam pointing and the target direction in the vertical plane; and the spatial angle error component integrates the azimuth and pitch errors to reflect a comprehensive indicator of spatial pointing deviation.

[0054] Execution steps: Determine the deviation from the target direction by measuring the beam pointing of independent subarrays. For example, if the target direction is 0° in azimuth and 45° in elevation, and the actual beam pointing is 2° in azimuth and 46.5° in elevation, the azimuth error is 2° and the elevation error is 1.5°. Configure the spatial angle error component. Further, according to the formula, add the square of the azimuth error to the square of the elevation error, and then take the square root to obtain the spatial angle error. This quantifies the spatial pointing deviation and provides accurate data support for subsequent calibration.

[0055] In the above steps, by decomposing and configuring the error components, key spatial error parameters are provided for the generation of the dynamic calibration weight matrix. The phase weights of the corresponding independent subarrays can be adjusted to reduce the deviation, thereby improving the antenna's directivity pattern gain and anti-interference capability. This helps to compensate for specific spatial angle errors in the subsequent calibration process, effectively improving calibration accuracy and antenna performance, and ensuring the quality of the communication link.

[0056] Furthermore, based on the phased array antenna configuration information, a three-dimensional near-field calibration benchmark is established. The method of the present application also includes: According to the number and arrangement of phased array antenna elements, a spherical scanning frame with a radius of U times the central wavelength of the operating frequency band is built in a microwave anechoic chamber. Based on the spherical scanning frame, a vector network analyzer and a double-ridged horn probe are configured to collect full-airspace field strength data through a preset scanning path. With the center of the array surface as the coordinate origin, the physical position parameters of each phased array antenna element are associated with the full-airspace field strength data to obtain a calibration reference database.

[0057] Specifically, the number and arrangement of phased array antenna elements means that the phased array antenna is composed of multiple antenna units (elements), and their number and arrangement (such as rectangular array, triangular array, etc.) determine the overall performance and calibration complexity of the antenna; the spherical scanning frame refers to a spherical structure built in a microwave anechoic chamber, which is used to support and move the measuring equipment. Its radius is determined according to the center wavelength of the working frequency band to ensure the accuracy of the measurement and the requirement of covering the entire airspace; the vector network analyzer is an instrument used to measure the amplitude and phase characteristics of RF and microwave components. It can accurately measure the amplitude and phase information of the antenna unit, providing key data support for calibration.

[0058] A double-ridged horn probe is a wide-band, low-loss antenna probe used to receive or transmit signals on a spherical scanning frame and collect full-airspace field strength data for phased array antennas. Full-airspace field strength data collection involves measuring the radiation field strength data of phased array antennas in all directions through a preset scanning path using a vector network analyzer in conjunction with a double-ridged horn probe to ensure the comprehensiveness of the calibration benchmark. The calibration benchmark database is a database constructed by associating the physical position parameters of each phased antenna element with the full-airspace field strength data. It is used to store and manage the basic data required for calibration, providing a basis for subsequent error analysis and calibration.

[0059] Execution steps: Based on the number and arrangement of the phased array antenna elements, a spherical scanning gantry with a radius of M times the center wavelength of the operating frequency band is constructed within a microwave anechoic chamber. For example, if the center wavelength of the phased array antenna's operating frequency band is λ and M is 2, the radius of the spherical scanning gantry is set to 2λ. Calibration is preferably performed in a darkroom environment to avoid external electromagnetic interference and ensure the accuracy of the measurement data. A vector network analyzer and a double-ridged horn probe are mounted on the spherical scanning gantry, and full-space field strength data is collected along a preset scanning path. The vector network analyzer accurately measures the amplitude and phase of the signal, while the double-ridged horn probe receives or transmits the signal. The two work together to ensure that the collected field strength data is comprehensive and accurate.

[0060] With the center of the array as the coordinate origin, the physical position parameters of each phased array antenna element are recorded, including the coordinate position, direction and other information of the element; the collected full-space field strength data is associated with these physical position parameters to form a calibration reference database. The calibration reference database can provide radiation characteristic data of the phased array antenna in different positions and directions, providing a basis for subsequent error analysis and calibration. For example, the calibration reference database can accurately analyze the radiation field strength of each element in different directions, and then determine the phase difference and amplitude error data in the initial error matrix, providing key support for subsequent calibration steps.

[0061] In the above steps, by precisely constructing a spherical scanning frame, configuring measurement equipment, and collecting full-airspace field strength data, a calibration benchmark database is built to provide reliable data support for subsequent error analysis, dynamic calibration weight matrix generation, and optimization of distributed calibration nodes, ensuring the scientific nature of the calibration process and the accuracy of the calibration results.

[0062] Furthermore, the least squares method is used to determine the compensation coefficient, and the environmental disturbance compensation factor is extracted by a deep separable convolutional neural network. The method of the present application also includes: An objective function is configured as the mean square error between a beam pattern and a target beam pattern, and a constraint condition of the objective function is an adjustment range of a compensation coefficient. Based on the objective function, a weighted least squares method is used to determine an overdetermined set of equations, and weight coefficients are dynamically allocated according to the near and far field of view attenuation characteristics of each phased antenna array element from the center of the array surface.

[0063] Specifically, the objective function is an indicator that measures the difference between the beam pattern and the target beam pattern, which is expressed here as the mean square error. The target beam pattern is the ideal pattern, while the beam pattern is the pattern obtained by actual measurement or simulation. The purpose of the objective function is to optimize the calibration effect by minimizing the difference between the two. The mean square error is a statistic that measures the difference between the predicted value and the true value. The calculation method is to square the difference between each predicted value and the true value and then take the average to measure the difference between the actual beam pattern and the target beam pattern.

[0064] The compensation coefficient adjustment range is the allowable range of compensation amplitude and phase adjustments during the calibration process. The compensation coefficient adjusts the amplitude and phase of the antenna elements to minimize the difference between the beam pattern and the target pattern. The adjustment range is typically expressed as an amplitude multiple and a phase angle, for example, between 0.5 and 2 for amplitude and -180° to 180° for phase. Weighted least squares is an optimization algorithm used to solve overdetermined equations in the presence of measurement errors or data uncertainties. By assigning a weight to each equation, data points with smaller errors have a greater impact on the result, thereby improving the accuracy of the solution.

[0065] An overdetermined system of equations is one in which the number of equations exceeds the number of unknowns. In this case, since the number of measured data points often exceeds the number of compensation coefficients required, a least squares method or other approach is required to solve the problem. The near-field attenuation characteristic describes how the radiation intensity of an antenna element varies with distance. The intensity in the near field decays more rapidly, while the intensity in the far field decays more slowly. Weight coefficients are dynamically assigned based on the element's distance from the array center and its field attenuation characteristic to reflect the influence of different elements on the overall beam pattern.

[0066] Execution steps: Configure the objective function as the mean square error between the beam pattern and the target beam pattern using the gain value of the actual beam pattern in the i-th direction and the gain value of the target beam pattern in the i-th direction. The constraint condition is the adjustment range of the compensation coefficient, that is, the amplitude is between 0.5 and 2 times, and the phase is between -180° and 180°.

[0067] Based on the configured objective function, a weighted least squares method is used to solve the overdetermined system of equations. Weight coefficients are dynamically assigned based on the field-of-view attenuation characteristics of each phased antenna element near or far from the array center. Specifically, the standard deviation of the field strength measurement error of the jth element is related to its distance from the array center. Near-field elements have faster field strength attenuation and larger measurement errors, resulting in smaller weight coefficients. Far-field elements have slower field strength attenuation and smaller measurement errors, resulting in larger weight coefficients. For example, in a phased array antenna containing 100 elements, the weight coefficient of the elements near the center may be 0.1, while the weight coefficient of the elements at the edge may be 1.0. Solving the overdetermined system of equations using the weighted least squares method yields the optimal compensation coefficient, minimizing the objective function and significantly improving the accuracy of the beam pattern.

[0068] In the above steps, the compensation coefficient is accurately determined through mathematical optimization methods to minimize the difference between the beam pattern and the target pattern. By rationally configuring the objective function and adopting the weighted least squares method, the contribution of different array elements and measurement errors are effectively considered, improving the accuracy and reliability of the calibration. This provides precise compensation coefficients for the subsequent dynamic calibration weight matrix generation and environmental disturbance compensation, ensuring that the phased array antenna can maintain high performance under various operating conditions.

[0069] Furthermore, a deep separable convolutional neural network is used to extract environmental disturbance compensation factors, separate the coupling effects of thermal drift, mechanical deformation, and electromagnetic interference, and generate a dynamic calibration weight matrix. The method of this application includes: The mapping relationship between environmental parameters and amplitude and phase errors in historical calibration data is used as a training sample. A separation thermal drift branch, a mechanical deformation branch, and an electromagnetic interference branch are set in the output layer of a deep separable convolutional neural network to decouple the coupling effect and formulate the dynamic calibration weight matrix.

[0070] Specifically, historical calibration data refers to data accumulated during past calibrations of phased array antennas. This data records the antenna's amplitude and phase errors under different environmental conditions, as well as the corresponding calibration parameters. This data is crucial for training machine learning models because it provides a mapping between environmental parameters and amplitude and phase errors. Environmental parameters refer to external conditions that affect the performance of phased array antennas, such as temperature, humidity, air pressure, vibration, and electromagnetic interference. Changes in these parameters can alter the antenna's amplitude and phase characteristics, affecting beam pointing accuracy and system stability.

[0071] Amplitude and phase errors refer to the deviations between the actual amplitude and phase of an antenna element and their ideal values. Amplitude errors are typically expressed in decibels (dB). These errors affect the antenna's beamforming and radiation pattern and require calibration to reduce. The deep separable convolutional neural network (DCNN) is a lightweight convolutional neural network architecture that reduces computational effort and the number of model parameters while maintaining performance by decomposing the standard convolution into depthwise and pointwise convolutions. This network architecture is suitable for resource-constrained environments, such as embedded systems.

[0072] Separating the thermal drift branch, mechanical deformation branch, and electromagnetic interference branch refers to setting up three independent branches in the output layer of the deep separable convolutional neural network, corresponding to the three coupling effects of thermal drift, mechanical deformation, and electromagnetic interference. Each branch outputs a corresponding compensation factor to decouple the influence of these three effects. Decoupling the coupling effects refers to separating the three intertwined influences of thermal drift, mechanical deformation, and electromagnetic interference, and determining their contributions to amplitude and phase errors. The decoupled compensation factors can be used more accurately for calibration, improving the calibration effect. The dynamic calibration weight matrix is ​​a matrix dynamically generated based on real-time environmental conditions and calibration requirements. It is used to adjust the amplitude and phase of the antenna array elements to compensate for the influence of environmental disturbances and mutual coupling effects. The dynamic calibration weight matrix plays a core role in the calibration process of phased array antennas, ensuring the high-performance operation of the antenna in different environments.

[0073] Implementation steps: Collect environmental parameters and corresponding amplitude and phase errors from historical calibration data. Record the amplitude and phase error data at different temperatures, vibration frequencies, and electromagnetic interference intensities to establish a training sample set. A deep separable convolutional neural network is constructed. The input layer receives the environmental parameter vector. The middle layer extracts features through depthwise and pointwise convolution. The output layer has three branches, corresponding to compensation factors for thermal drift, mechanical deformation, and electromagnetic interference, respectively. For example, the network structure can be: input layer (3 neurons) → depthwise convolution layer (32 filters, filter size 1×3) → pointwise convolution layer (64 filters) → depthwise convolution layer (64 filters, filter size 1×3) → pointwise convolution layer (128 filters) → output layer (3 branches, each corresponding to a compensation factor). During training, the root mean square error (RMSE) is used as the loss function, and the Adam optimizer is used for optimization. The training data can be used for training (80%), validation (10%), and testing (10%) to ensure the model's generalization ability.

[0074] Through the trained deep separable convolutional neural network, real-time environmental parameters are input, and the three branches of the network output layer output compensation factors for thermal drift, mechanical deformation, and electromagnetic interference respectively. Based on the decoupled compensation factors, the initial error matrix and the spatial angle error component are combined to generate a dynamic calibration weight matrix. The elements of the dynamic calibration weight matrix represent the amplitude and phase adjustment of each array element and are used to calibrate the amplitude and phase state of the antenna. In the above steps, historical data is used to train the machine learning model to achieve real-time compensation for environmental disturbances. The branch structure of the deep separable convolutional neural network can effectively decouple the effects of thermal drift, mechanical deformation, and electromagnetic interference, and generate an accurate dynamic calibration weight matrix, which provides key support for the high-performance operation of phased array antennas in complex dynamic environments and significantly improves the adaptability and reliability of the antenna system.

[0075] In summary, the beneficial effects of the embodiments of the present application are: Due to the adoption of a three-dimensional near-field calibration benchmark based on the phased array antenna configuration information, the phase difference and amplitude error of the receiving antenna unit are received, and the initial error matrix is ​​formulated; the phased antenna array elements are grouped and excited, and the beam pointing deviation of each group of phased antenna array elements is determined through the far-field direction vector, and the spatial angle error component is configured; based on the initial error matrix and the spatial angle error component, the compensation coefficient is determined using the least squares method, and the environmental disturbance compensation factor is extracted through the deep separable convolutional neural network, the coupling effect of thermal drift, mechanical deformation and electromagnetic interference is separated, and a dynamic calibration weight matrix is ​​generated; according to the dynamic calibration weight matrix, the distributed calibration nodes are deployed in combination with the beam pointing accuracy requirements, and the calibration parameters are iteratively optimized in a multi-base station collaborative scenario. At the same time, the mutual coupling effect bias measurement and calibration decision are performed. This application provides a phased array antenna calibration method and calibration system. By establishing a three-dimensional near-field calibration benchmark, the initial error matrix is ​​formulated by combining the phase difference and amplitude error, and the group excitation and angle error are collaboratively optimized. The least squares method and the deep separable convolutional neural network are combined to effectively separate the environmental disturbances and dynamically compensate for them. Distributed calibration nodes are deployed to improve the mutual coupling calibration efficiency between array elements, thereby ensuring the technical effects that affect the stability of the communication link.

[0076] The second embodiment is based on the same inventive concept as the phased array antenna calibration method in the previous embodiment. Figure 2 As shown, an embodiment of the present application provides a phased array antenna calibration system, wherein the system includes: Initial error matrix preparation module M100: Based on the phased array antenna configuration information, it establishes a three-dimensional near-field calibration benchmark, receives the phase difference and amplitude error of the antenna unit, and prepares the initial error matrix.

[0077] Group excitation module M200: Group excitation of phased antenna array elements, determines the beam pointing deviation of each group of phased antenna array elements through the far-field direction vector, and configures the spatial angle error component.

[0078] Coupling analysis module M300: Based on the initial error matrix and spatial angle error components, the least squares method is used to determine the compensation coefficient, and the environmental disturbance compensation factor is extracted through a deep separable convolutional neural network. The coupling effects of thermal drift, mechanical deformation and electromagnetic interference are separated to generate a dynamic calibration weight matrix.

[0079] Calibration decision module M400: Based on the dynamic calibration weight matrix and combined with the beam pointing accuracy requirements, distributed calibration nodes are deployed, calibration parameters are iteratively optimized in a multi-base station collaboration scenario, and mutual coupling effect bias measurement and calibration decision are performed at the same time.

[0080] Furthermore, the calibration decision module M400 is further configured to execute the following method: A first adaptive calibration indicator is configured based on the amplitude mutual coupling effect between phased array antenna elements; a second adaptive calibration indicator is configured based on the phase mutual coupling effect between phased array antenna elements; a test period is determined based on the first adaptive calibration indicator and the second adaptive calibration indicator; within the test period, the element phase difference is determined based on the received signal amplitude of the signal transmitted independently by each phased array element, and the mutual coupling effect offset measurement and calibration decision are performed: the coupling coefficient between adjacent array elements is calculated through port S parameter measurement, the offset of each phased array element is determined, and fuzzy decision making is used to perform adaptive calibration.

[0081] Furthermore, the group incentive module M200 is used to perform the following method: The phased antenna array plane is divided into N independent sub-arrays, each independent sub-array includes P phased antenna array elements, and the isolation between the independent sub-arrays meets the isolation limit condition; according to the timing control logic, the N independent sub-arrays are polled and excited by high-speed radio frequency switches.

[0082] Furthermore, the group incentive module M200 is further configured to execute the following method: The adjacent sub-array excitation intervals corresponding to the N independent sub-arrays meet the time interval limitation condition; at the same time, a preset initial phase weight is applied to each independent sub-array according to the deviation between the independent sub-array beam pointing and the target direction.

[0083] Furthermore, the group incentive module M200 is further configured to execute the following method: A first time factor is determined based on the operating frequency of the phased array antenna. A free-space propagation model is used to determine the time required for the signal of the first adjacent subarray to be transmitted, decay in energy, and affect the signal acquisition of the second adjacent subarray to meet the set interference threshold, thereby obtaining a second time factor. A third time factor corresponding to the switching time of the high-speed RF switch and a fourth time factor corresponding to the signal processing delay are determined. The first and second time factors are combined to set a time interval limitation condition.

[0084] Furthermore, the group incentive module M200 is further configured to execute the following method: According to the pointing deviation value between the independent sub-array beam pointing and the target direction, the azimuth error component and the elevation error component are decomposed; and the space angle error component is configured through the azimuth error component and the elevation error component.

[0085] Furthermore, the initial error matrix preparation module M100 is further configured to perform the following method: According to the number and arrangement of phased array antenna elements, a spherical scanning frame with a radius of U times the central wavelength of the operating frequency band is built in a microwave anechoic chamber. Based on the spherical scanning frame, a vector network analyzer and a double-ridged horn probe are configured to collect full-airspace field strength data through a preset scanning path. With the center of the array surface as the coordinate origin, the physical position parameters of each phased array antenna element are associated with the full-airspace field strength data to obtain a calibration reference database.

[0086] Furthermore, the coupling analysis module M300 is further configured to execute the following method: An objective function is configured as the mean square error between a beam pattern and a target beam pattern, and a constraint condition of the objective function is an adjustment range of a compensation coefficient. Based on the objective function, a weighted least squares method is used to determine an overdetermined set of equations, and weight coefficients are dynamically allocated according to the near and far field of view attenuation characteristics of each phased antenna array element from the center of the array surface.

[0087] Furthermore, the coupling analysis module M300 is further configured to execute the following method: The mapping relationship between environmental parameters and amplitude and phase errors in historical calibration data is used as a training sample. A separation thermal drift branch, a mechanical deformation branch, and an electromagnetic interference branch are set in the output layer of a deep separable convolutional neural network to decouple the coupling effect and formulate the dynamic calibration weight matrix.

[0088] In summary, any step can be stored as a computer instruction or program in an unlimited computer memory and can be called and recognized by an unlimited computer processor, without any unnecessary restrictions.

[0089] Furthermore, the above technical solution only reflects the preferred technical solution of the technical solution of the embodiment of the present application. Some changes that may be made to certain parts thereof by technical personnel in this technical field all reflect the novel principles of the embodiment of the present application. Obviously, technical personnel in this field can make various changes and modifications to the present application without departing from the scope of the present application.

Claims

1. A phased array antenna calibration method, characterized in that: The method comprises: Based on the phased array antenna configuration information, a three-dimensional near-field calibration benchmark is established to receive the phase difference and amplitude error of the antenna unit and formulate the initial error matrix. The phased antenna array elements are excited in groups, and the beam pointing deviation of each group of phased antenna array elements is determined through the far-field direction vector, and the spatial angle error component is configured; Based on the initial error matrix and the spatial angle error component, the least squares method is used to determine the compensation coefficient, and the environmental disturbance compensation factor is extracted through a deep separable convolutional neural network to separate the coupling effects of thermal drift, mechanical deformation and electromagnetic interference, and generate a dynamic calibration weight matrix; According to the dynamic calibration weight matrix, distributed calibration nodes are deployed in combination with the beam pointing accuracy requirements, and calibration parameters are iteratively optimized in a multi-base station collaboration scenario. At the same time, mutual coupling effect bias measurement and calibration decision-making are performed.

2. The phased array antenna calibration method according to claim 1, wherein: Before making mutual coupling effect bias measurements and calibration decisions, the method includes: According to the amplitude mutual coupling effect between the phased array antenna elements, a first adaptive calibration index is configured; according to the phase mutual coupling effect between the phased array antenna elements, a second adaptive calibration index is configured; determining a test period based on the first adaptive calibration indicator and the second adaptive calibration indicator; During the test cycle, the phase difference of each phased antenna element is determined based on the received signal amplitude of the signal transmitted by each phased antenna element, and mutual coupling effect offset measurement and calibration decisions are made. The coupling coefficient between adjacent elements is calculated through port S-parameter measurement, the offset of each phased antenna element is determined, and adaptive calibration is performed using fuzzy decision making.

3. The phased array antenna calibration method according to claim 2, wherein: The phased antenna array elements are excited in groups, the method comprising: Dividing the phased antenna array into N independent sub-arrays, each independent sub-array comprising P phased antenna elements, wherein the isolation between the independent sub-arrays meets the isolation limit condition; According to the timing control logic, the N independent sub-arrays are polled and excited by a high-speed radio frequency switch.

4. The phased array antenna calibration method according to claim 3, wherein: The adjacent sub-array excitation intervals corresponding to the N independent sub-arrays meet the time interval limitation condition; At the same time, a preset initial phase weight is applied to each independent sub-array according to the deviation between the independent sub-array beam pointing direction and the target direction.

5. The phased array antenna calibration method according to claim 4, wherein: The adjacent sub-array excitation intervals corresponding to the N independent sub-arrays meet a time interval limitation condition, and the method includes: Determine the first time factor based on the operating frequency of the phased array antenna; Using a free-space propagation model, determine the time required for the first adjacent subarray signal to be transmitted, decay in energy, and affect the acquisition of the second adjacent subarray signal to meet the set interference threshold, thereby obtaining a second time factor. Determine a third time factor corresponding to the switching time of the high-speed RF switch and a fourth time factor corresponding to the signal processing delay, and set a time interval limiting condition in combination with the first time factor and the second time factor.

6. The phased array antenna calibration method according to claim 4, wherein: Determining the beam pointing deviation of each group of phased antenna array elements and configuring a spatial angle error component, the method includes: According to the deviation between the independent sub-array beam pointing and the target direction, it is decomposed into azimuth error component and elevation error component; The spatial angle error component is configured by using the azimuth error component and the elevation error component.

7. The phased array antenna calibration method according to claim 2, wherein: Establishing a three-dimensional near-field calibration benchmark based on the phased array antenna configuration information, the method further comprising: According to the number and arrangement of phased array antenna elements, a spherical scanning frame with a radius of U times the central wavelength of the working frequency band is built in the microwave darkroom; Based on the spherical scanning frame, a vector network analyzer and a double-ridged horn probe are configured to collect full-space field strength data through a preset scanning path; Taking the center of the array as the coordinate origin, the physical position parameters of each phased antenna array element are associated with the full airspace field strength data to obtain a calibration reference database.

8. The phased array antenna calibration method according to claim 7, wherein: Determining the compensation coefficient using the least squares method, and extracting the environmental disturbance compensation factor using a deep separable convolutional neural network, the method further comprising: An objective function is configured as a mean square error between a beam pattern and a target beam pattern, wherein a constraint condition of the objective function is an adjustment range of a compensation coefficient; Based on the objective function, a weighted least square method is used to determine an overdetermined set of equations, and weight coefficients are dynamically allocated according to the near and far field of view attenuation characteristics of each phased antenna array element from the center of the array surface.

9. The phased array antenna calibration method according to claim 8, wherein: The environmental disturbance compensation factor is extracted by a deep separable convolutional neural network, the coupling effects of thermal drift, mechanical deformation and electromagnetic interference are separated, and a dynamic calibration weight matrix is ​​generated. The method includes: The mapping relationship between environmental parameters and amplitude and phase errors in historical calibration data is used as training samples; A separation thermal drift branch, a mechanical deformation branch, and an electromagnetic interference branch are set in the output layer of the deep separable convolutional neural network to decouple the coupling effect and formulate the dynamic calibration weight matrix.

10. Phased array antenna calibration system, characterized in that, The system for implementing the phased array antenna calibration method according to any one of claims 1 to 9 comprises: Initial error matrix preparation module: Based on the phased array antenna configuration information, a three-dimensional near-field calibration benchmark is established, the phase difference and amplitude error of the receiving antenna unit are received, and the initial error matrix is ​​prepared; Group excitation module: Group excitation of phased antenna array elements, determines the beam pointing deviation of each group of phased antenna array elements through the far-field direction vector, and configures the spatial angle error component; Coupling analysis module: Based on the initial error matrix and the spatial angle error component, the least squares method is used to determine the compensation coefficient, and the environmental disturbance compensation factor is extracted through a deep separable convolutional neural network to separate the coupling effects of thermal drift, mechanical deformation and electromagnetic interference, and generate a dynamic calibration weight matrix; Calibration decision module: Based on the dynamic calibration weight matrix and combined with the beam pointing accuracy requirements, distributed calibration nodes are deployed, calibration parameters are iteratively optimized in a multi-base station collaboration scenario, and mutual coupling effect bias measurement and calibration decision are performed at the same time.

Images