Multi-channel passive antenna array high-efficiency amplitude-phase test system

Abstract

The invention discloses a multichannel passive antenna array high-efficiency amplitude-phase test system, and relates to the technical field of amplitude-phase tests.According to the multichannel passive antenna array high-efficiency amplitude-phase test system, a robustness analysis and error modeling module is introduced, the influence of environmental disturbance factors on an antenna array is evaluated and compensated in real time, a global error model is constructed, static errors and time-varying errors can be effectively recognized, and the accuracy of amplitude-phase testing is improved. An error compensation strategy is dynamically adjusted, high precision can be kept in a dynamic environment, meanwhile, a deep learning auto-encoder is adopted to extract time-varying error features, a self-adaptive correction strategy is combined, real-time error compensation and dynamic adjustment are achieved, the system adjusts amplitude-phase data of each channel according to real-time feedback, and in various platform dynamic states, the real-time error compensation and dynamic adjustment are achieved. The test result is kept continuous and accurate, error compensation and correction can be automatically completed in the whole process through automatic error feature extraction, self-adaptive correction and a feedback mechanism, the requirement for human intervention is reduced, and a correction strategy is continuously optimized in the platform test process.

Description

Technical Field

[0001] The invention relates to the technical field of amplitude and phase testing, in particular to a high-efficiency amplitude and phase testing system for a multi-channel passive antenna array. Background Art

[0002] In the amplitude and phase test of multi-channel passive antenna arrays, with the widespread application of arrays on high-speed dynamic platforms, such as aircraft and radar systems, the requirements for their test accuracy are gradually increasing. When array antennas are operating on these platforms, they are affected by environmental factors such as tremor, vibration, and temperature fluctuations. The amplitude and phase errors of the channels may frequently fluctuate greatly, showing time-varying and randomness. This error change directly affects the accuracy of array beamforming, and then affects the detection and positioning capabilities of the signal. Therefore, accurate and rapid correction of real-time amplitude and phase errors has become an important technical problem.

[0003] However, most existing amplitude and phase test methods rely on offline calibration or real-time adjustment through fixed reference sources, which makes it difficult to cope with changing high-dynamic environments. When errors change rapidly, these methods lack real-time adaptive adjustment capabilities, resulting in insufficient test efficiency and accuracy.

[0004] Some traditional solutions use offline calibration methods, and preset test sources and reference models to calibrate each channel one by one to obtain amplitude and phase error information. However, this method is too time-consuming in dynamic scenarios, requires frequent corrections, and has low overall efficiency. In addition, some solutions place a small number of standard calibration elements in the array for online real-time adjustment. Although this method can provide reference values, it is limited to dealing with certain specific error types and cannot handle complex and frequently changing errors. In the above solutions, due to the complex sources of array errors and the combined influence of multiple environmental factors, a single reference source or a single calibration cannot effectively deal with the time-varying nature of complex errors. Therefore, a multi-channel passive antenna array high-efficiency amplitude and phase test system is urgently needed to solve such problems. Summary of the invention

[0005] In view of the above existing problems, the present invention is proposed.

[0006] The present invention provides a multi-channel passive antenna array high-efficiency amplitude and phase test system to solve the problem of complex disturbances caused by the superposition of multiple error sources. Traditional algorithms mostly process different error sources separately in error modeling, making it difficult to construct a global characteristic model of the error, and therefore it is difficult to effectively ensure the stability of the array beam in a dynamic environment.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] On the one hand, the present invention provides a multi-channel passive antenna array high-efficiency amplitude and phase test system, which includes:

[0009] The signal acquisition module is responsible for collecting signals from the multi-channel antenna array and performing signal preprocessing on the collected raw data;

[0010] Robustness analysis module, based on preprocessed signals, models environmental factors that may affect array performance, including tremor, temperature change and vibration; tremor, temperature change and vibration factors may cause array errors; identify and analyze the sources of static errors and time-varying errors, and generate a global error model containing different error characteristics;

[0011] The error feature extraction module uses the trained autoencoder to perform real-time analysis on signal data and automatically extract time-varying error features;

[0012] The correction strategy generation module generates the correction strategy for each channel in real time based on the error features extracted by the autoencoder;

[0013] Adaptive correction feedback module, using real-time adaptive correction algorithm, applies the correction strategy to the amplitude and phase data of each channel in real time;

[0014] The global correction evaluation module evaluates the correction effect of the correction strategy; the global correction evaluation module checks the correction effect of all channels and readjusts the correction strategy as needed;

[0015] A global evaluation and retraining module, after the global evaluation, re-evaluates the error pattern based on the evaluation results and retrains the autoencoder; the error pattern includes errors caused by environmental disturbances, equipment changes or other external factors.

[0016] Furthermore, in the signal acquisition module, preliminary processing includes noise suppression, signal filtering, and removal of external interference, and the signal is collected and converted by hardware devices such as receivers and analog-to-digital converters (ADCs);

[0017] The autoencoder uses a deep learning model to extract amplitude and phase error features from the input signal provided by the signal acquisition module.

[0018] Furthermore, the correction strategy is dynamically adjusted to cope with changes in the platform state, including array vibration and temperature fluctuations; through this strategy, the system can automatically generate adaptive correction schemes for different error sources (such as vibration, ambient temperature changes, etc.) based on real-time data, thereby ensuring that amplitude and phase errors can be effectively corrected at any time.

[0019] On the other hand, the present invention provides a high-efficiency amplitude and phase testing method for a multi-channel passive antenna array, comprising:

[0020] Step S1, array error modeling and signal acquisition,

[0021] The signal acquisition module collects signals from the multi-channel antenna array. The collected raw signals are first converted by the analog-to-digital converter ADC, and then undergo preliminary signal preprocessing, including noise suppression, signal filtering, and interference removal.

[0022] The robustness analysis module models possible environmental disturbance factors based on the preprocessed signals, analyzes the impact of disturbance factors on array performance, and generates a global error model containing static error and time-varying error characteristics based on the analysis results.

[0023] Step S2, real-time extraction of error features,

[0024] The error feature extraction module uses a deep learning autoencoder to perform real-time analysis on the pre-processed data obtained from the signal acquisition module. The autoencoder is trained to automatically extract time-varying error features from the signal, including amplitude and phase error features. The real-time output of the autoencoder includes the amplitude and phase error mode of each channel of the array.

[0025] Step S3, dynamically generate a correction strategy,

[0026] Based on the time-varying error features extracted by the autoencoder, the correction strategy generation module generates the error correction strategy for each channel in real time;

[0027] In step S3, the error correction strategy combines the global error model to dynamically generate correction parameters for error types caused by tremor, temperature fluctuation and vibration environment factors; the generated correction strategy includes amplitude and phase adjustment values ​​and adaptive control parameters for each channel, and performs personalized correction for different error sources;

[0028] Step S4, adaptive correction and real-time feedback,

[0029] In this step, the adaptive correction feedback module applies the correction strategy generated in step S3 to the amplitude and phase data of each channel in real time, and uses an adaptive algorithm to continuously optimize the correction strategy according to the dynamic changes of the platform state;

[0030] During the calibration process, the amplitude and phase data of each channel are dynamically adjusted based on continuous error evaluation and feedback mechanism;

[0031] Step S5, global correction effect evaluation and retraining,

[0032] The global correction evaluation module evaluates the data after the correction strategy is applied in step S4 to detect the deviation between the correction effect and the standard data;

[0033] Based on the evaluation results, if it is found that the correction effect does not meet expectations, the global evaluation and retraining module re-evaluates the error pattern and retrains the autoencoder. The retraining process models new error patterns (such as equipment changes, external disturbances, etc.);

[0034] Step 6: Continuous optimization and online adjustment,

[0035] During the entire testing process, the test system uses real-time adaptive correction and global evaluation mechanisms to continuously optimize the correction strategy. Whenever environmental factors change or the platform status changes, the error compensation strategy is automatically adjusted. Finally, through feedback and evaluation, the model and strategy are continuously adjusted to improve the robustness and adaptability of the test system.

[0036] Furthermore, in step S1, the array error modeling and signal acquisition method is:

[0037] Assume that the signal collected from the multi-channel antenna array after preprocessing is y i (t), where i represents different array channels and t is the measurement time. For each channel’s signal: y i (t) = A i (t)+n i (t)+∈ i (t), where y i (t) is the signal of the i-th channel at time i, A i (t) is the ideal signal, representing the ideal amplitude and phase response without environmental disturbance, n i (t) is the noise term, ∈ i (t) is the environmental disturbance term, which represents the impact of tremor, temperature change and vibration dynamic factors on the signal;

[0038] Assume that the perturbation of each channel ∈ i (t) consists of disturbance sources of different frequencies: Among them, α i,k is the amplitude of the i-th channel on the k-th disturbance source, ω k is the frequency of the kth disturbance source, φ i,k is the phase of the kth disturbance source, δ i (t) is the error term caused by other random disturbances in the environment, and K represents the number of disturbance sources;

[0039] Introduce the time-varying error matrix E(t), the matrix elements are E ij (t), represents the error correlation between the i-th channel and the j-th channel: Where C is the conventional static error matrix of the system, representing the long-term fixed error of the array, β k is the influence weight of the kth disturbance source on the error, ω k and θk are the frequency and phase of the kth disturbance source, respectively, and δ(t) is the error caused by other random disturbances;

[0040] Combine the static error matrix C and the time-varying error matrix E(t) to generate the global error model M(t), that is:

[0041] Where M(t) is the global error matrix, which includes static error and time-varying error;

[0042] Preprocessing signal y i (t) is compared with the global error model M(t) to extract the time-varying error characteristics of each channel:

[0043] in, is the fitted signal, M(t) is the calculated global error;

[0044] In step S1, the error matrix is ​​updated regularly: Among them, K(t) is the updated gain matrix, which indicates the degree of adjustment of the error matrix at the current moment. is the error between the current signal and the fitted signal.

[0045] Furthermore, in step S2, the error feature is extracted in real time in the following manner:

[0046] The preprocessed signal y(t) obtained by the signal acquisition module is input into the autoencoder, where t represents the time step. is a multi-channel signal, N represents the number of channels in the array, represents the transposition operation. The signal is processed by the encoder and mapped to the low-dimensional latent space z(t). Let the encoder function be f enc , then z(t)=f enc (y(t),θ enc ), where θ enc is the encoder parameter, z(t)∈R M represents the low-dimensional representation in the latent space, and M<<N represents the dimension of the latent space;

[0047] The encoder output z(t) is passed to the decoder, and the decoder function is f dec , then the reconstructed signal Expressed as Among them, θ dec are the decoder parameters, is the signal reconstructed by the decoder, and the reconstruction error L reconstruction (t) is the mean square error MSE between the input signal and the reconstructed signal, that is, Among them, y i (t) and are the i-th channels of the original signal and the reconstructed signal, respectively. N is the number of channels. The back propagation algorithm is used to optimize the parameters θ of the autoencoder. enc and θ dec Minimize the reconstruction error so that the autoencoder can capture the time-varying error characteristics in the signal. During the training process, the changes in the latent space representation z(t) reflect the dynamic characteristics of the time-varying error.

[0048] After the autoencoder is trained, the latent space representation z(t) is analyzed to extract the amplitude and phase error features. Assuming that the autoencoder captures the amplitude and phase error features Δφ(t), the amplitude and phase error is extracted from the latent space: Δφ(t) = g(z(t),θ φ ), where g is the mapping function from the latent space to the amplitude and phase error, θ φ is the parameter of the mapping function. Through the autoencoder architecture, the real-time signal data is encoded and decoded, and the amplitude and phase error features are automatically extracted in the latent space. The amplitude and phase error features Δφ(t) of each time step t are output as the input of the subsequent correction strategy generation module.

[0049] Furthermore, in step S3, the adaptive correction and real-time feedback method is:

[0050] Assume that the time-varying error feature extracted from the autoencoder is Δφ i (t), where i represents the error feature of the i-th channel, t is the time step, and the error feature extracted by the encoder is expressed as: Among them, Φ i (t) is the real signal phase of the i-th channel, is the estimated value of the signal after preprocessing of the i-th channel;

[0051] The global error model is constructed, and E(t) represents the total impact of all disturbance factors on the system, including tremor, temperature fluctuation and vibration, and η(t) is the environmental disturbance factor: Among them, η k (t) represents the impact of the kth environmental disturbance on the signal, w k is the weight factor of each disturbance on the signal;

[0052] According to the extracted error feature Δφ i (t) and the global error model E(t), the correction strategy generation module uses the optimization algorithm to generate the amplitude and phase correction parameters of each channel Set the calibration strategy to:

[0053] Among them, λ i The adaptive correction gain for each channel represents the response strength of each channel to error compensation, Δφ i(t) is the time-varying error feature extracted from the autoencoder, and E(t) is the environmental disturbance effect generated by the global error model;

[0054] The correction strategy generation module introduces the adaptive control parameter α i (t), dynamically adjust the correction gain λ i : Among them, β i is the adaptive control coefficient of channel i, |Δφ i (t)| represents the absolute value of the error characteristic, reflecting the current error size, according to the adaptive control parameter α i (t), correction gain λ i Further adjusted to λ i (t) = λ i α i (t);

[0055] Based on the adaptively adjusted correction gain λ i (t) and the time-varying error characteristic Δφ i (t), generate the correction strategy: The strategy is applied to the signal of each channel, and the correction is performed in real time.

[0056] Furthermore, in step S4, the correction strategy is dynamically generated as follows:

[0057] According to the correction strategy generated in step S3, the real-time correction amplitude and phase value of each channel is calculated, assuming is the corrected amplitude and phase data of the i-th channel, Φ i (t) is the original amplitude and phase data of the i-th channel, then the amplitude and phase of each channel after real-time correction is for: Among them, Φ i (t) is the original amplitude and phase data of the i-th channel, is the amplitude and phase correction value of the ith channel at time t, is the corrected amplitude and phase data of the i-th channel at time t;

[0058] After applying the correction strategy, the continuous error evaluation and feedback mechanism calculates the error between the corrected amplitude and the reference amplitude to evaluate the current correction effect, and sets the reference amplitude to Then the error of the i-th channel ∈ i (t) is: in, is the amplitude and phase data after real-time correction calculated in step S3, is the reference amplitude and phase data, that is, the amplitude and phase under ideal conditions, ∈ i (t) is the amplitude and phase error of the i-th channel;

[0059] After error evaluation, an adaptive algorithm is used to optimize the correction strategy for each channel, setting the adaptive gain to α i (t), then the adaptive correction strategy update formula is: in, is the initial correction amplitude and phase of the i-th channel, α i (t) is the adaptive gain, ∈ i (t) is the error calculated in step S2;

[0060] Adaptive gain α i (t) Adjust according to the change of platform status. When the platform status changes, recalculate the gain α i (t), and the state feedback control method is used to adjust: Among them, λ i (t) is the dynamic adjustment factor caused by the change of platform status, which is estimated in real time according to different disturbance factors. i (y) is the adaptive gain, dynamically adjust α i (t) Achieve continuous optimization of the correction strategy;

[0061] After each correction and evaluation, the amplitude and phase data applied to each channel are fed back in real time. When the platform status changes, continuous adjustments are made to finally obtain the optimal amplitude and phase values:

[0062] Furthermore, in step S5, the global correction evaluation and retraining method is:

[0063] Set the corrected amplitude and phase data to The standard amplitude and phase data is The evaluation process is based on the overall evaluation of the amplitude and phase errors of each channel, and the deviation of the i-th channel is defined as: Calculate the weighted mean square error of each channel deviation to evaluate the global correction effect and set the weight function w i (t) Weighting the importance of different channels: Among them, w i (t) is the weight function of the ith channel, δ i (t) is the deviation of the ith channel, i.e., the difference between the corrected amplitude and phase data and the standard data;

[0064] If the global error evaluation result shows a deviation value δ i If (t) is too large and the correction effect does not meet expectations, the error mode re-evaluation process is entered. The error mode is re-evaluated according to environmental changes and equipment fluctuations. Here, a dynamic error model based on high-order moments is introduced to dynamically update the error mode:

[0065] M i (t) = α·M i (t-1)+(1-α)·(δ i (t)) 2 , where M i (t) is the error matrix of the ith channel, updated at each time step, α is the smoothing factor in the range [0,1], and δ i (t) is the deviation of the ith channel, indicating the amplitude and phase error at the current moment;

[0066] The autoencoder is retrained based on the results of global error evaluation and error pattern re-evaluation. Assuming that the loss function of the autoencoder is L(θ), the loss function is optimized:

[0067] in, is the corrected amplitude and phase data of the i-th channel, is the corrected amplitude and phase data regenerated after retraining the autoencoder, λ is the regularization parameter, and θ is the parameter of the autoencoder; through this loss function, the autoencoder will be retrained for the new error pattern and gradually optimize the model weights so that the model can adapt to the new disturbance source and error characteristics.

[0068] Furthermore, in step S5, the global correction evaluation and retraining method further includes:

[0069] After retraining, the new autoencoder model can be used to generate a dynamic correction strategy for the current error pattern. The autoencoder generates more accurate correction values ​​based on the updated error pattern and feedback signal, and adjusts the amplitude and phase data of each channel in real time.

[0070] Assume that during the error pattern re-evaluation process, the system uses a feedback mechanism to dynamically adjust each channel and update the correction strategy. The dynamic adjustment formula is:

[0071] in, is the updated amplitude and phase data, indicating the final result after correction, γ i (t) is the adaptive adjustment factor of each channel, which is adjusted according to the dynamic state of the platform, δ i (t) is the current error, which indicates the difference between the corrected data and the standard data.

[0072] The beneficial effects of the present invention are:

[0073] The present invention introduces a robustness analysis and error modeling module to evaluate and compensate for the impact of environmental disturbance factors on the antenna array in real time, build a global error model, effectively identify static errors and time-varying errors, dynamically adjust the error compensation strategy, and maintain high precision in a dynamic environment. At the same time, a deep learning autoencoder is used to extract time-varying error features, and an adaptive correction strategy is combined to achieve real-time error compensation and dynamic adjustment. The autoencoder captures the error patterns of different channels and generates real-time error correction parameters through an optimization algorithm. The system adjusts the amplitude and phase data of each channel according to real-time feedback, keeps the test results continuously accurate under various platform dynamic conditions, and significantly improves the flexibility and robustness of the system.

[0074] The present invention can automatically complete error compensation and correction throughout the process through automated error feature extraction, adaptive correction and feedback mechanisms, reducing the need for human intervention. The real-time trained autoencoder model and global error evaluation module can continuously optimize the correction strategy during the platform testing process, significantly improving the testing efficiency, and enabling the system to complete high-precision amplitude and phase testing in a short time.

[0075] The present invention introduces a global error evaluation and retraining module, which can dynamically adjust the error compensation model according to the error pattern and deviation evaluation results. Based on the dynamic error model of high-order moments, the system can update the error pattern in real time according to environmental changes and equipment fluctuation factors, and retrain the autoencoder. When encountering new disturbances or equipment changes, it can quickly adapt to and optimize the test process.

[0076] The present invention introduces a global error model and a time-varying error matrix, which can effectively model the error correlation between channels. The adaptive correction strategy can dynamically adjust the compensation gain according to the error characteristics of each channel, so that the error of each channel can be personalized. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0078] Figure 1 It is a schematic diagram of the structure of a high-efficiency amplitude and phase test system for a multi-channel passive antenna array of the present invention;

[0079] Figure 2 It is a schematic flow chart of a high-efficiency amplitude and phase testing method for a multi-channel passive antenna array according to the present invention. DETAILED DESCRIPTION

[0080] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0081] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0082] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0083] Example 1, reference Figure 1 , this embodiment provides a multi-channel passive antenna array high-efficiency amplitude and phase test system, including:

[0084] The signal acquisition module is responsible for collecting signals from the multi-channel antenna array and performing signal preprocessing on the collected raw data;

[0085] In the signal acquisition module, preliminary processing includes noise suppression, signal filtering and removal of external interference. The signal is collected and converted by hardware devices such as receivers and analog-to-digital converters (ADCs);

[0086] Robustness analysis module, based on preprocessed signals, models environmental factors that may affect array performance, including tremor, temperature change and vibration; tremor, temperature change and vibration factors may cause array errors; identify and analyze the sources of static errors and time-varying errors, and generate a global error model containing different error characteristics;

[0087] The error feature extraction module uses the trained autoencoder to perform real-time analysis on signal data and automatically extract time-varying error features;

[0088] The autoencoder uses a deep learning model to extract amplitude and phase error features from the input signal provided by the signal acquisition module;

[0089] The correction strategy generation module generates the correction strategy for each channel in real time based on the error features extracted by the autoencoder;

[0090] The correction strategy is dynamically adjusted to cope with changes in the platform state, including array vibration and temperature fluctuations. Through this strategy, the system can automatically generate adaptive correction schemes for different error sources (such as vibration, ambient temperature changes, etc.) based on real-time data, thereby ensuring that amplitude and phase errors can be effectively corrected at any time.

[0091] Adaptive correction feedback module, using real-time adaptive correction algorithm, applies the correction strategy to the amplitude and phase data of each channel in real time;

[0092] The global correction evaluation module evaluates the correction effect of the correction strategy; the global correction evaluation module checks the correction effect of all channels and readjusts the correction strategy as needed;

[0093] A global evaluation and retraining module, after the global evaluation, re-evaluates the error pattern based on the evaluation results and retrains the autoencoder; the error pattern includes errors caused by environmental disturbances, equipment changes or other external factors.

[0094] Example 2, reference Figure 2 This embodiment provides a high-efficiency amplitude and phase test method for a multi-channel passive antenna array, comprising the following steps:

[0095] Step S1, array error modeling and signal acquisition,

[0096] The signal acquisition module collects signals from the multi-channel antenna array. The collected raw signals are first converted by the analog-to-digital converter ADC, and then undergo preliminary signal preprocessing, including noise suppression, signal filtering, and interference removal.

[0097] The robustness analysis module models possible environmental disturbance factors based on the preprocessed signals, analyzes the impact of disturbance factors on array performance, and generates a global error model containing static error and time-varying error characteristics based on the analysis results.

[0098] In step S1, array error modeling and signal acquisition methods are as follows:

[0099] Assume that the signal collected from the multi-channel antenna array after preprocessing is y i (t), where i represents different array channels and t is the measurement time. For each channel’s signal: y i (t) = A i (t)+n i (t)+∈ i (t), where y i (t) is the signal of the i-th channel at time i, A i (t) is the ideal signal, representing the ideal amplitude and phase response without environmental disturbance, n i (t) is the noise term, ∈i (t) is the environmental disturbance term, which represents the impact of tremor, temperature change and vibration dynamic factors on the signal;

[0100] Assume that the perturbation of each channel ∈ i (t) consists of disturbance sources of different frequencies: Among them, α i,k is the amplitude of the i-th channel on the k-th disturbance source, ω k is the frequency of the kth disturbance source, φ i,k is the phase of the kth disturbance source, δ i (t) is the error term caused by other random disturbances in the environment, and K represents the number of disturbance sources;

[0101] Introduce the time-varying error matrix E(t), the matrix elements are E ij (t), represents the error correlation between the i-th channel and the j-th channel: Where C is the conventional static error matrix of the system, representing the long-term fixed error of the array, β k is the influence weight of the kth disturbance source on the error, ω k and θ k are the frequency and phase of the kth disturbance source, respectively, and δ(t) is the error caused by other random disturbances;

[0102] Combine the static error matrix C and the time-varying error matrix E(t) to generate the global error model M(t), that is:

[0103] Where M(t) is the global error matrix, which includes static error and time-varying error;

[0104] Preprocessing signal y i (t) is compared with the global error model M(t) to extract the time-varying error characteristics of each channel:

[0105] in, is the fitted signal, M(t) is the calculated global error;

[0106] In step S1, the error matrix is ​​updated regularly: Among them, K(t) is the updated gain matrix, which indicates the degree of adjustment of the error matrix at the current moment. is the error between the current signal and the fitting signal;

[0107] Specifically, the robustness analysis module can not only identify and analyze the error sources of array performance, but also generate a global error model that adapts to different environmental disturbances. The error model contains the characteristics of static errors and time-varying errors.

[0108] Step S2, real-time extraction of error features,

[0109] The error feature extraction module uses a deep learning autoencoder to perform real-time analysis on the pre-processed data obtained from the signal acquisition module. The autoencoder is trained to automatically extract time-varying error features from the signal, including amplitude and phase error features. The real-time output of the autoencoder includes the amplitude and phase error mode of each channel of the array.

[0110] In step S2, the error feature is extracted in real time in the following manner:

[0111] The preprocessed signal y(t) obtained by the signal acquisition module is input into the autoencoder, where t represents the time step. is a multi-channel signal, N represents the number of channels in the array, represents the transposition operation. The signal is processed by the encoder and mapped to the low-dimensional latent space z(t). Let the encoder function be f enc , then z(t)=f enc (y(t),θ enc ), where θ enc is the encoder parameter, z(t)∈R M represents the low-dimensional representation in the latent space, and M<<N represents the dimension of the latent space;

[0112] The encoder output z(t) is passed to the decoder, and the decoder function is f dec , then the reconstructed signal Expressed as Among them, θ dec are the decoder parameters, is the signal reconstructed by the decoder, and the reconstruction error L reconstruction (t) is the mean square error MSE between the input signal and the reconstructed signal, that is, Among them, y i (t) and are the i-th channels of the original signal and the reconstructed signal, respectively. N is the number of channels. The back propagation algorithm is used to optimize the parameters θ of the autoencoder. enc and θ dec Minimize the reconstruction error so that the autoencoder can capture the time-varying error characteristics in the signal. During the training process, the changes in the latent space representation z(t) reflect the dynamic characteristics of the time-varying error.

[0113] After the autoencoder is trained, the latent space representation z(t) is analyzed to extract the amplitude and phase error features. Assuming that the autoencoder captures the amplitude and phase error features Δφ(t), the amplitude and phase error is extracted from the latent space: Δφ(t) = g(z(t),θ φ ), where g is the mapping function from the latent space to the amplitude and phase error, θ φis the parameter of the mapping function. Through this autoencoder architecture, the real-time signal data is encoded and decoded, and the amplitude and phase error features are automatically extracted in the latent space. The amplitude and phase error features Δφ(t) at each time step t are output as the input of the subsequent correction strategy generation module.

[0114] Specifically, the autoencoder method maps the input signal through the encoder, captures the low-dimensional representation of the signal in the latent space, and then reconstructs the signal through the decoder and minimizes the reconstruction error. The dynamic changes in the latent space representation obtained through training can effectively reflect the time-varying amplitude and phase error characteristics.

[0115] Step S3, dynamically generate a correction strategy,

[0116] Based on the time-varying error features extracted by the autoencoder, the correction strategy generation module generates the error correction strategy for each channel in real time;

[0117] In step S3, the error correction strategy combines the global error model to dynamically generate correction parameters for error types caused by tremor, temperature fluctuation and vibration environment factors; the generated correction strategy includes amplitude and phase adjustment values ​​and adaptive control parameters for each channel, and performs personalized correction for different error sources;

[0118] In step S3, the adaptive correction and real-time feedback method is:

[0119] Assume that the time-varying error feature extracted from the autoencoder is Δφ i (t), where i represents the error feature of the i-th channel, t is the time step, and the error feature extracted by the encoder is expressed as: Among them, Φ i (t) is the real signal phase of the i-th channel, is the estimated value of the signal after preprocessing of the i-th channel;

[0120] The global error model is constructed, and E(t) represents the total impact of all disturbance factors on the system, including tremor, temperature fluctuation and vibration, and η(t) is the environmental disturbance factor: Among them, η k (t) represents the impact of the kth environmental disturbance on the signal, w k is the weight factor of each disturbance on the signal;

[0121] According to the extracted error feature Δφ i (t) and the global error model E(t), the correction strategy generation module uses the optimization algorithm to generate the amplitude and phase correction parameters of each channel Set the calibration strategy to:

[0122] Among them, λ iThe adaptive correction gain for each channel represents the response strength of each channel to error compensation, Δφ i (t) is the time-varying error feature extracted from the autoencoder, and E(t) is the environmental disturbance effect generated by the global error model;

[0123] The correction strategy generation module introduces the adaptive control parameter α i (t), dynamically adjust the correction gain λ i : Among them, β i is the adaptive control coefficient of channel i, |Δφ i (t)| represents the absolute value of the error characteristic, reflecting the current error size, according to the adaptive control parameter α i (t), correction gain λ i Further adjusted to λ i (t) = λ i α i (t);

[0124] Based on the adaptively adjusted correction gain λ i (t) and the time-varying error characteristic Δφ i (t), generate the correction strategy: This strategy is applied to the signal of each channel, and correction is performed in real time;

[0125] Specifically, through the above steps, the correction strategy generation module can generate a correction strategy for each channel in real time according to the time-varying error characteristics and environmental factors. The correction strategy includes not only the amplitude and phase adjustment values An adaptive control parameter (α i (t)) to dynamically adjust the response and ensure the error correction effect when the platform status changes.

[0126] Step S4, adaptive correction and real-time feedback,

[0127] In this step, the adaptive correction feedback module applies the correction strategy generated in step S3 to the amplitude and phase data of each channel in real time, and uses an adaptive algorithm to continuously optimize the correction strategy according to the dynamic changes of the platform state;

[0128] During the calibration process, the amplitude and phase data of each channel are dynamically adjusted based on continuous error evaluation and feedback mechanism;

[0129] In step S4, the correction strategy is dynamically generated as follows:

[0130] According to the correction strategy generated in step S3, the real-time correction amplitude and phase value of each channel is calculated, assuming is the corrected amplitude and phase data of the i-th channel, Φ i(t) is the original amplitude and phase data of the i-th channel, then the amplitude and phase of each channel after real-time correction is for: Among them, Φ i (t) is the original amplitude and phase data of the i-th channel, is the amplitude and phase correction value of the ith channel at time t, is the corrected amplitude and phase data of the i-th channel at time t;

[0131] After applying the correction strategy, the continuous error evaluation and feedback mechanism calculates the error between the corrected amplitude and the reference amplitude to evaluate the current correction effect, and sets the reference amplitude to Then the error of the i-th channel ∈ i (t) is: in, is the amplitude and phase data after real-time correction calculated in step S3, is the reference amplitude and phase data, that is, the amplitude and phase under ideal conditions, ∈ i (t) is the amplitude and phase error of the i-th channel;

[0132] After error evaluation, an adaptive algorithm is used to optimize the correction strategy for each channel, setting the adaptive gain to α i (t), then the adaptive correction strategy update formula is: in, is the initial correction amplitude and phase of the i-th channel, α i (t) is the adaptive gain, ∈ i (t) is the error calculated in step S2;

[0133] Adaptive gain α i (t) Adjust according to the change of platform status. When the platform status changes, recalculate the gain α i (t), and the state feedback control method is used to adjust: Among them, λ i (t) is the dynamic adjustment factor caused by the change of platform status, which is estimated in real time according to different disturbance factors. i (t) is the adaptive gain, dynamically adjust α i (t) Achieve continuous optimization of the correction strategy;

[0134] After each correction and evaluation, the amplitude and phase data applied to each channel are fed back in real time. When the platform status changes, continuous adjustments are made to finally obtain the optimal amplitude and phase values:

[0135] Step S5, global correction effect evaluation and retraining,

[0136] The global correction evaluation module evaluates the data after the correction strategy is applied in step S4 to detect the deviation between the correction effect and the standard data;

[0137] Based on the evaluation results, if it is found that the correction effect does not meet expectations, the global evaluation and retraining module re-evaluates the error pattern and retrains the autoencoder. The retraining process models new error patterns (such as equipment changes, external disturbances, etc.);

[0138] In step S5, the global correction evaluation and retraining method is:

[0139] Set the corrected amplitude and phase data to The standard amplitude and phase data is The evaluation process is based on the overall evaluation of the amplitude and phase errors of each channel, and the deviation of the i-th channel is defined as: Calculate the weighted mean square error of each channel deviation to evaluate the global correction effect and set the weight function w i (t) Weighting the importance of different channels: Among them, w i (t) is the weight function of the ith channel, δ i (t) is the deviation of the ith channel, i.e., the difference between the corrected amplitude and phase data and the standard data;

[0140] If the global error evaluation result shows a deviation value δ i If (t) is too large and the correction effect does not meet expectations, the error mode re-evaluation process is entered. The error mode is re-evaluated according to environmental changes and equipment fluctuations. Here, a dynamic error model based on high-order moments is introduced to dynamically update the error mode:

[0141] M i (t) = α·M i (t-1)+(1-α)·(δ i (t)) 2 , where M i (t) is the error matrix of the ith channel, updated at each time step, α is the smoothing factor in the range [0,1], and δ i (t) is the deviation of the ith channel, indicating the amplitude and phase error at the current moment;

[0142] The autoencoder is retrained based on the results of global error evaluation and error pattern re-evaluation. Assuming that the loss function of the autoencoder is L(θ), the loss function is optimized:

[0143] in, is the corrected amplitude and phase data of the i-th channel, is the corrected amplitude and phase data regenerated after retraining the autoencoder, λ is the regularization parameter, and θ is the parameter of the autoencoder; through this loss function, the autoencoder will be retrained for the new error pattern, gradually optimizing the model weights so that the model adapts to the new disturbance source and error characteristics;

[0144] In step S5, the global correction evaluation and retraining method further includes:

[0145] After retraining, the new autoencoder model can be used to generate a dynamic correction strategy for the current error pattern. The autoencoder generates more accurate correction values ​​based on the updated error pattern and feedback signal, and adjusts the amplitude and phase data of each channel in real time.

[0146] Assume that during the error pattern re-evaluation process, the system uses a feedback mechanism to dynamically adjust each channel and update the correction strategy. The dynamic adjustment formula is:

[0147] in, is the updated amplitude and phase data, indicating the final result after correction, γ i (t) is the adaptive adjustment factor of each channel, which is adjusted according to the dynamic state of the platform, δ i (t) is the current error, which indicates the difference between the corrected data and the standard data. Specifically, in step S5, all error evaluation and retraining steps are based on real-time error data. The optimization and retraining of the correction strategy are guaranteed by calculating the weighted mean square error, the dynamic error matrix and the loss function optimization. Finally, through the adaptive mechanism of feedback adjustment and retraining, the system can cope with dynamically changing error sources, continuously optimize and achieve more accurate amplitude and phase correction.

[0148] Step 6: Continuous optimization and online adjustment,

[0149] During the entire testing process, the test system uses real-time adaptive correction and global evaluation mechanisms to continuously optimize the correction strategy. Whenever environmental factors change or the platform status changes, the error compensation strategy is automatically adjusted. Finally, through feedback and evaluation, the model and strategy are continuously adjusted to improve the robustness and adaptability of the test system.

[0150] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A multi-channel passive antenna array high-efficiency amplitude and phase test system, characterized by: include, The signal acquisition module is responsible for collecting signals from the multi-channel antenna array and performing signal preprocessing on the collected raw data; Robustness analysis module, which models environmental factors that may affect array performance, including chatter, temperature changes, and vibrations based on preprocessed signals; identifies and analyzes the sources of static and time-varying errors, and generates a global error model that includes different error characteristics; The error feature extraction module uses the trained autoencoder to perform real-time analysis on signal data and automatically extract time-varying error features; The correction strategy generation module generates the correction strategy for each channel in real time based on the error features extracted by the autoencoder; Adaptive correction feedback module, using real-time adaptive correction algorithm, applies the correction strategy to the amplitude and phase data of each channel in real time; The global correction evaluation module evaluates the correction effect of the correction strategy; the global correction evaluation module checks the correction effect of all channels and readjusts the correction strategy as needed; Global evaluation and retraining module,After the global evaluation, the error pattern is re-evaluated according to the evaluation results, and the autoencoder is retrained.

2. A multi-channel passive antenna array high-efficiency amplitude and phase test system according to claim 1, characterized in that: In the signal acquisition module, preliminary processing includes noise suppression, signal filtering, and removal of external interference; The autoencoder uses a deep learning model to extract amplitude and phase error features from the input signal provided by the signal acquisition module.

3. A multi-channel passive antenna array high-efficiency amplitude and phase test system according to claim 2, characterized in that: The correction strategy is dynamically adjusted to cope with changes in the platform state, including vibration and temperature fluctuations of the array.

4. A high-efficiency amplitude and phase test method for a multi-channel passive antenna array, characterized in that: A multi-channel passive antenna array high-efficiency amplitude and phase test system based on any one of claims 1 to 3, The following steps are included: Step S1: array error modeling and signal acquisition. The signal acquisition module collects signals from the multi-channel antenna array. The collected raw signals are first converted by the analog-to-digital converter ADC, and then undergo preliminary signal preprocessing, including noise suppression, signal filtering, and interference removal. The robustness analysis module models possible environmental disturbance factors based on the preprocessed signals, analyzes the impact of disturbance factors on array performance, and generates a global error model containing static error and time-varying error characteristics based on the analysis results. Step S2, real-time extraction of error features, The error feature extraction module uses a deep learning autoencoder to perform real-time analysis on the pre-processed data obtained from the signal acquisition module. The autoencoder is trained to automatically extract time-varying error features from the signal, including amplitude and phase error features. The real-time output of the autoencoder includes the amplitude and phase error mode of each channel of the array. Step S3, dynamically generate a correction strategy, Based on the time-varying error features extracted by the autoencoder, the correction strategy generation module generates the error correction strategy for each channel in real time; In step S3, the error correction strategy combines the global error model to dynamically generate correction parameters for error types caused by tremor, temperature fluctuation and vibration environment factors; the generated correction strategy includes amplitude and phase adjustment values ​​and adaptive control parameters for each channel, and performs personalized correction for different error sources; Step S4, adaptive correction and real-time feedback, In this step, the adaptive correction feedback module applies the correction strategy generated in step S3 to the amplitude and phase data of each channel in real time, and uses an adaptive algorithm to continuously optimize the correction strategy according to the dynamic changes of the platform state; During the calibration process, the amplitude and phase data of each channel are dynamically adjusted based on continuous error evaluation and feedback mechanism; Step S5, global correction effect evaluation and retraining, The global correction evaluation module evaluates the data after the correction strategy is applied in step S4 to detect the deviation between the correction effect and the standard data; Based on the evaluation results, if it is found that the correction effect does not meet expectations, the global evaluation and retraining module re-evaluates the error pattern and retrains the autoencoder. The retraining process models the new error pattern. Step 6: Continuous optimization and online adjustment, During the entire testing process, the test system uses real-time adaptive correction and global evaluation mechanisms to continuously optimize the correction strategy; whenever environmental factors change or the platform status changes, the error compensation strategy is automatically adjusted.

5. A multi-channel passive antenna array high-efficiency amplitude and phase testing method according to claim 4, characterized in that: In step S1, array error modeling and signal acquisition methods are as follows: Assume that the signal collected from the multi-channel antenna array after preprocessing is y i (t), where i represents different array channels and t is the measurement time. For each channel’s signal: y i (t) = A i (t)+n i (t)+∈ i (t), where y i (t) is the signal of the i-th channel at time i, A i (t) is the ideal signal, representing the ideal amplitude and phase response without environmental disturbance, n i (t) is the noise term, ∈ i (t) is the environmental disturbance term, which represents the impact of tremor, temperature change and vibration dynamic factors on the signal; Assume that the perturbation of each channel ∈ i (t) consists of disturbance sources of different frequencies: Among them, α i,k is the amplitude of the i-th channel on the k-th disturbance source, ω k is the frequency of the kth disturbance source, φ i,k is the phase of the kth disturbance source, δ i (t) is the error term caused by other random disturbances in the environment, and K represents the number of disturbance sources; Introduce the time-varying error matrix E(t), the matrix elements are E ij (t), represents the error correlation between the i-th channel and the j-th channel: Where C is the conventional static error matrix of the system, representing the long-term fixed error of the array, β k is the influence weight of the kth disturbance source on the error, ω k and θ k are the frequency and phase of the kth disturbance source, respectively, and δ(t) is the error caused by other random disturbances; Combine the static error matrix C and the time-varying error matrix E(t) to generate the global error model M(t), that is: Where M(t) is the global error matrix, which includes static error and time-varying error; Preprocessing signal y i (t) is compared with the global error model M(t) to extract the time-varying error characteristics of each channel: in, is the fitted signal, M(t) is the calculated global error; In step S1, the error matrix is ​​updated regularly: Among them, K(t) is the updated gain matrix, which indicates the degree of adjustment of the error matrix at the current moment. is the error between the current signal and the fitted signal.

6. A multi-channel passive antenna array high-efficiency amplitude and phase testing method according to claim 5, characterized in that: In step S2, the error feature is extracted in real time in the following manner: The preprocessed signal y(t) obtained by the signal acquisition module is input into the autoencoder, where t represents the time step, y(t) = [y1(t), y2(t), …, y N (t)] T is a multi-channel signal, N represents the number of channels in the array, T represents the transposition operation, the signal is processed by the encoder and mapped to the low-dimensional latent space z(t), let the encoder function be f enc , then z(t)=f enc (y(t),θ enc ), where θ enc is the encoder parameter, z(t)∈R M represents the low-dimensional representation in the latent space, and M<<N represents the dimension of the latent space; The encoder output z(t) is passed to the decoder, and the decoder function is f dec , then the reconstructed signal Expressed as Among them, θ dec are the decoder parameters, is the signal reconstructed by the decoder, and the reconstruction error L reconstruction (t) is the mean square error MSE between the input signal and the reconstructed signal, that is, Among them, y i (t) and are the i-th channels of the original signal and the reconstructed signal, respectively. N is the number of channels. The back propagation algorithm is used to optimize the parameters θ of the autoencoder. enc and θ dec Minimize reconstruction error; After the autoencoder is trained, the latent space representation z(t) is analyzed to extract the amplitude and phase error features. Assuming that the autoencoder captures the amplitude and phase error features Δφ(t), the amplitude and phase error is extracted from the latent space: Δφ(t) = g(z(t),θ φ ), where g is the mapping function from the latent space to the amplitude and phase error, θ φ are the parameters of the mapping function.

7. A multi-channel passive antenna array high-efficiency amplitude and phase testing method according to claim 6, characterized in that: In step S3, the adaptive correction and real-time feedback method is: Assume that the time-varying error feature extracted from the autoencoder is Δφ i (t), where i represents the error feature of the i-th channel, t is the time step, and the error feature extracted by the encoder is expressed as: Among them, Φ i (t) is the real signal phase of the i-th channel, is the estimated value of the signal after preprocessing of the i-th channel; The global error model is constructed, and E(t) represents the total impact of all disturbance factors on the system, including tremor, temperature fluctuation and vibration, and η(t) is the environmental disturbance factor: Among them, η k (t) represents the impact of the kth environmental disturbance on the signal, w k is the weight factor of each disturbance on the signal; According to the extracted error feature Δφ i (t) and the global error model E(t), the correction strategy generation module uses the optimization algorithm to generate the amplitude and phase correction parameters of each channel Set the calibration strategy to: Among them, λ i The adaptive correction gain for each channel represents the response strength of each channel to error compensation, Δφ i (t) is the time-varying error feature extracted from the autoencoder, and E(t) is the environmental disturbance effect generated by the global error model; The correction strategy generation module introduces the adaptive control parameter α i (t), dynamically adjust the correction gain λ i : Among them, β i is the adaptive control coefficient of channel i, |Δφ i (t)| represents the absolute value of the error characteristic, reflecting the current error size, according to the adaptive control parameter α i (t), correction gain λ i Further adjusted to λ i (t) = λ i α i (t); Based on the adaptively adjusted correction gain λ i (t) and the time-varying error characteristic Δφ i (t), generate the correction strategy: The strategy is applied to the signal of each channel, and the correction is performed in real time.

8. A high-efficiency amplitude and phase testing method for a multi-channel passive antenna array according to claim 7, characterized in that: In step S4, the correction strategy is dynamically generated in the following manner: According to the correction strategy generated in step S3, the real-time correction amplitude and phase value of each channel is calculated, assuming is the corrected amplitude and phase data of the i-th channel, Φ i (t) is the original amplitude and phase data of the i-th channel, then the amplitude and phase of each channel after real-time correction is for: Among them, Φ i (t) is the original amplitude and phase data of the i-th channel, is the amplitude and phase correction value of the ith channel at time t, is the corrected amplitude and phase data of the i-th channel at time t; After applying the correction strategy, the continuous error evaluation and feedback mechanism calculates the error between the corrected amplitude and the reference amplitude to evaluate the current correction effect. The reference amplitude is set to Then the error of the i-th channel ∈ i (t) is: in, is the amplitude and phase data after real-time correction calculated in step S3, is the reference amplitude and phase data, that is, the amplitude and phase under ideal conditions, ∈ i (t) is the amplitude and phase error of the i-th channel; After error evaluation, an adaptive algorithm is used to optimize the correction strategy for each channel, setting the adaptive gain to α i (t), then the adaptive correction strategy update formula is: in, is the initial correction amplitude and phase of the i-th channel, α i (t) is the adaptive gain, ∈ i (t) is the error calculated in step S2; Adaptive gain α i (t) Adjust according to the change of platform status. When the platform status changes, recalculate the gain α i (t), and the state feedback control method is used to adjust: Among them, λ i (t) is the dynamic adjustment factor caused by the change of platform status, which is estimated in real time according to different disturbance factors. i (t) is the adaptive gain; After each correction and evaluation, the amplitude and phase data applied to each channel are fed back in real time. When the platform status changes, continuous adjustments are made to finally obtain the optimal amplitude and phase values:

9. A high-efficiency amplitude and phase testing method for a multi-channel passive antenna array according to claim 8, characterized in that: In step S5, the global correction evaluation and retraining method is: Set the corrected amplitude and phase data to The standard amplitude and phase data is The evaluation process is based on the overall evaluation of the amplitude and phase errors of each channel, and the deviation of the i-th channel is defined as: Calculate the weighted mean square error of each channel deviation to evaluate the global correction effect and set the weight function w i (t) Weighting the importance of different channels: Among them, w i (t) is the weight function of the ith channel, δ i (t) is the deviation of the ith channel, i.e., the difference between the corrected amplitude and phase data and the standard data; If the global error evaluation result shows a deviation value δ i If (t) is too large and the correction effect does not meet expectations, the error mode re-evaluation process is entered. The error mode is re-evaluated according to environmental changes and equipment fluctuations. Here, a dynamic error model based on high-order moments is introduced to dynamically update the error mode: M i (t) = α·M i (t-1)+(1-α)·(δ i (t)) 2 , where M i (t) is the error matrix of the ith channel, updated at each time step, α is the smoothing factor in the range [0,1], and δ i (t) is the deviation of the ith channel, indicating the amplitude and phase error at the current moment; The autoencoder is retrained based on the results of global error evaluation and error pattern re-evaluation. Assuming that the loss function of the autoencoder is L(θ), the loss function is optimized: in, is the corrected amplitude and phase data of the i-th channel, is the corrected amplitude and phase data regenerated after retraining the autoencoder, λ is the regularization parameter, and θ is the parameter of the autoencoder.

10. A high-efficiency amplitude and phase testing method for a multi-channel passive antenna array according to claim 9, characterized in that: In step S5, the global correction evaluation and retraining method further includes: Assume that during the error pattern re-evaluation process, the system uses a feedback mechanism to dynamically adjust each channel and update the correction strategy. The dynamic adjustment formula is: in, is the updated amplitude and phase data, indicating the final result after correction, γ i (t) is the adaptive adjustment factor of each channel, which is adjusted according to the dynamic state of the platform, δ i (t) is the current error, which indicates the difference between the corrected data and the standard data.

Images