A data processing method for a wave control module

Abstract

This invention discloses a data processing method for a wave control module, belonging to the field of radar data processing. The method includes: acquiring spatial electromagnetic signals; constructing the original data sequence of the spatial electromagnetic signals; sampling the original data to obtain standardized data of the sampling points; dynamically suppressing Gaussian white noise and impulse interference using an improved adaptive Kalman filter algorithm to output the denoised useful target signal; extracting three-dimensional features in the time, frequency, and spatial domains to obtain the weighted fusion features of the sampling points; constructing an LSTM-Attention hybrid model to predict the core parameters of the wave control module; correcting the core parameters; and using the corrected core parameters as the core parameters for the future output of the wave control module to modify the data processing process of the wave control module. This invention effectively improves the anti-interference capability of the wave control module in the data processing process, enhances environmental adaptability, and is suitable for complex electromagnetic environments, providing a wave control module data processing process with high precision, low latency, and strong anti-interference characteristics.

Description

Technical Field

[0001] This invention relates to the field of radar data processing, and more specifically to a data processing method for a beam control module. Background Technology

[0002] The beam control module is a core component of phased array radar and adaptive communication systems. Its data processing performance directly determines beam pointing accuracy, signal response speed, and system anti-interference capability. Existing beam control modules have the following main shortcomings in data processing:

[0003] 1. Data preprocessing using fixed-coefficient filtering (such as mean filtering and static Kalman filtering) cannot adapt to time-varying noise (such as impulse noise and broadband interference) in complex electromagnetic environments, resulting in limited improvement in signal-to-noise ratio.

[0004] 2. The processing and correction of beam parameters rely on experience or single-dimensional characteristics, which generally result in large errors, making it difficult to meet the requirements of high-precision applications and having poor environmental adaptability. Summary of the Invention

[0005] To address the aforementioned shortcomings in the prior art, this invention provides a data processing method for a wave control module.

[0006] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows:

[0007] A data processing method for a wave control module is provided, comprising:

[0008] Step S1: Acquire spatial electromagnetic signals through the ADC sampling unit of the wave control module, construct the original data sequence of spatial electromagnetic signals, extract the target useful signal, sample the original data, and standardize the original data at the sampling points to obtain standardized data of the sampling points. ;

[0009] Step S2: Based on the target useful signal at each sampling point and standardized data An improved adaptive Kalman filter algorithm is used to dynamically suppress Gaussian white noise and impulse interference, and output the denoised target useful signal. ;

[0010] Step S3: From the target useful signal Three-dimensional features in the time domain, frequency domain, and spatial domain are extracted, and three-dimensional feature fusion is achieved through a dynamic weight matrix to obtain the weighted fused features of the sampling points. ;

[0011] Step S4: Construct an LSTM-Attention hybrid model to fuse weights and features. As input, the core parameters of the predictive wave control module The core parameters include the pointing angle. Beamwidth and transmit gain ;

[0012] Step S5: Utilize the core parameters predicted by the wave control module The core parameters are corrected based on the real-time output of the wave control module, and the corrected core parameters are used as the core parameters for future output of the wave control module to correct the data processing process of the wave control module.

[0013] Further, step S1 includes:

[0014] Step S11: Acquire spatial electromagnetic signals through the ADC sampling unit of the wave control module to construct the original data sequence of the spatial electromagnetic signals. The original data sequence satisfies... , The raw data collected, Useful signals for the target It is Gaussian white noise. This is an electromagnetic interference signal. t For signal timing;

[0015] Useful signals of the target , A The signal amplitude, The center frequency of the signal. This is the initial phase;

[0016] Electromagnetic interference signals Including pulse interference and bandwidth interference , , ; B For the amplitude of interference, C The interference intensity coefficient, For rectangular window functions, The pulse width. This is a function for generating random numbers;

[0017] Step S12: Sample from the original data sequence to obtain the original data corresponding to each sampling point. , k Number the sampling points; based on the minimum value in the original data sequence. and maximum value For raw data Standardization processing is performed to obtain standardized data from the sampling points. ;

[0018] .

[0019] Further, step S2 includes:

[0020] Step S21: Utilize the target's useful signal and the corresponding center frequency and initial phase Constructing the system state vector And construct the state equations;

[0021] ;

[0022] in, Here is the state transition matrix. Sampling points k -1 system state vector This is the input control matrix; it is a zero matrix when there is no external input. , Sampling points k -1 control input, Sampling points k -1 process noise, The process noise has a Gaussian distribution. The process noise covariance matrix is... These are the attenuation coefficients for the target useful signal, center frequency, and initial phase, respectively. Sampling points k The covariance of the target useful signal, center frequency, and initial phase;

[0023] Step S22: Utilize standardized data Construct observation equations;

[0024] ;

[0025] in, Sampling points k The observed values, Sampling points k The observation matrix To observe the noise, To observe the noise variance, To observe the Gaussian distribution of the noise;

[0026] Step S23: Construct an adaptive adjustment mechanism for process noise covariance and observation noise covariance;

[0027] ;

[0028] in, M The sliding window length is set during the adaptive adjustment process. i The number of the sampling point in the sliding window. Sampling pointsk A prediction error of -1, For adaptive adjustment coefficients, Sampling points k The process noise covariance matrix is ​​-1. Sampling points k -1 observation value, Sampling points k The observation matrix of -1 Sampling points k The prior system state estimation vector is -1. Sampling points i The prediction error;

[0029] Step S24: Correct the system state vector based on the adaptive adjustment mechanism, collaboratively suppress Gaussian white noise and impulse interference at the sampling points, recursively update the system state vector at the sampling points using Kalman filtering, and output the denoised target useful signal. ;

[0030] ;

[0031] in, Sampling points k The prior system state estimation vector is -1. Sampling points k of

[0032] Prior covariance matrix, Sampling points k The prior covariance matrix of -1 Sampling points k Kalman gain, For updated sampling points k The system state vector, For updated sampling points k The covariance matrix, I It is an identity matrix.

[0033] Further, step S3 includes:

[0034] Step S31: From the target useful signal Extract three-dimensional features in the time domain, frequency domain, and spatial domain;

[0035] Time-domain characteristics are ; They are respectively N The target useful signal at each sampling point The mean, standard deviation, skewness, and kurtosis;

[0036] Frequency domain characteristics are ; The useful signals of the target are respectively The center frequency, frequency bandwidth, maximum spectral power, and total harmonic distortion;

[0037] airspace characteristics are , These are the azimuth angle, elevation angle, rate of change of azimuth angle, and rate of change of elevation angle of the signal, respectively.

[0038] Step S32: Calculate the weights of time-domain features, frequency-domain features, and spatial-domain features. , obtain sampling points k dynamic weight vector ;

[0039] ;

[0040] in, Weights for time-domain features, frequency-domain features, or frequency-domain features. The ratio of signal-to-noise ratio to time-domain features, frequency-domain features, or frequency-domain features. Signal-to-noise ratio (SNR) representing time-domain, frequency-domain, or frequency-domain features. Power as a time-domain feature, frequency-domain feature, or frequency-domain feature. Noise power with time-domain, frequency-domain, or frequency-domain characteristics;

[0041] Step S33: Calculate the weighted fusion features of the sampling points based on the dynamic weight vector. ;

[0042] .

[0043] Further, step S4 includes:

[0044] Step S41: Construct an LSTM-Attention hybrid model to fuse weights and features. The data processing procedure of the LSTM-Attention hybrid model is as follows:

[0045] ;

[0046] ;

[0047] ;

[0048] ;

[0049] in, The output vector of the input gate. It is the sigmoid activation function. Here is the input weight matrix of the input gate. Sampling points k -1 is a hidden state. Let be the hidden state weight matrix of the input gate. Sampling points k -1 cell state vector The cell state weight matrix is ​​the input gate. For the input gate bias; This is the output vector of the forget gate. Here is the input weight matrix for the forget gate. Here is the hidden state weight matrix for the forget gate. This is the cell state weight matrix for the forgetting gate. For the input gate bias; Sampling points k The cell state vector, The input weight matrix represents the cell state. The hidden state weight matrix represents the cell state. Bias for cell state; The input weight matrix of the output gate. Let the hidden state weight matrix be the output gate. This is the cell state weight matrix for the output gate. For the output gate bias;

[0050] Step S42: Based on the hidden state corresponding to different sampling points Weight the hidden states and output the weighted hidden states. ;

[0051] ;

[0052] in, For the average hidden state, This is the attention weight matrix. , For the set of real numbers, Sampling points k The corresponding hidden state coefficients, Sampling points k The corresponding attention weights;

[0053] Step S43: Weight the hidden state Input to a fully connected layer, output the core parameters of the prediction. ;

[0054] ;

[0055] in, These are the predicted output weight matrix and the output bias, respectively. ;

[0056] The loss function of the LSTM-Attention hybrid model is:

[0057] ;

[0058] in, Sampling points k The corresponding true values ​​of the core parameters of the wave control module. The regularization coefficient is set.

[0059] Further, step S5 includes:

[0060] Step S51: Utilize the predicted core parameters and sampling points k The core parameters output in real time by the wave control module Error in calculating core parameters ;

[0061] ;

[0062] Step S52: Based on the error The core parameters are then corrected to obtain the corrected core parameters. ;

[0063] ;

[0064] in, To correct the step size, It is a symbolic function;

[0065] Step S53: Utilize the corrected core parameters As the core parameter output by the future wave control module, the data processing process of the wave control module is corrected.

[0066] The beneficial effects of this invention are as follows: This invention effectively improves the anti-interference capability of the wave control module in the data processing process, enhances environmental adaptability, and is suitable for complex electromagnetic environments, while also providing a wave control module data processing process with high precision, low latency, and strong anti-interference characteristics. Attached Figure Description

[0067] Figure 1 This is a flowchart of the data processing method for the wave control module. Detailed Implementation

[0068] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0069] like Figure 1 As shown, a data processing method for a wave control module includes:

[0070] Step S1: Acquire spatial electromagnetic signals through the ADC sampling unit of the wave control module, construct the original data sequence of spatial electromagnetic signals, extract the target useful signal, sample the original data, and standardize the original data at the sampling points to obtain standardized data of the sampling points.

[0071] Step S1 specifically includes:

[0072] Step S11: Acquire spatial electromagnetic signals through the ADC sampling unit of the wave control module to construct the original data sequence of the spatial electromagnetic signals. The original data sequence satisfies... , The raw data collected, Useful signals for the target It is Gaussian white noise. This is an electromagnetic interference signal. t For signal timing;

[0073] Useful signals of the target , A The signal amplitude, The center frequency of the signal. This is the initial phase;

[0074] Electromagnetic interference signals Including pulse interference and bandwidth interference , , ; B For the amplitude of interference, C The interference intensity coefficient, For rectangular window functions, The pulse width. This is a function for generating random numbers;

[0075] Step S12: Sample from the original data sequence to obtain the original data corresponding to each sampling point. , kThe sampling points are numbered, and the sampling points are sampled based on the time sequence of the original data, corresponding to the signal time of the original data; based on the minimum value in the original data sequence. and maximum value For raw data Standardization processing is performed to obtain standardized data from the sampling points. ;

[0076] .

[0077] Step S2: Based on the target useful signal at each sampling point and standardized data An improved adaptive Kalman filter algorithm is used to dynamically suppress Gaussian white noise and impulse interference, and output the denoised target useful signal. .

[0078] Step S2 specifically includes:

[0079] Step S21: Utilize the target's useful signal and the corresponding center frequency and initial phase Constructing the system state vector And construct the state equations;

[0080] ;

[0081] in, Here is the state transition matrix. Sampling points k -1 system state vector This is the input control matrix; it is a zero matrix when there is no external input. , Sampling points k -1 control input, Sampling points k -1 process noise, The process noise has a Gaussian distribution. The process noise covariance matrix is... These are the attenuation coefficients for the target useful signal, center frequency, and initial phase, respectively. Sampling points k The covariance of the target useful signal, center frequency, and initial phase;

[0082] Step S22: Utilize standardized data Construct observation equations;

[0083] ;

[0084] in, Sampling points k The observed values, Sampling points k The observation matrix To observe the noise, To observe the noise variance, To observe the Gaussian distribution of the noise;

[0085] Step S23: Construct an adaptive adjustment mechanism for process noise covariance and observation noise covariance;

[0086] ;

[0087] in, M The sliding window length is set during the adaptive adjustment process. i The number of the sampling point in the sliding window. Sampling points k A prediction error of -1, For adaptive adjustment coefficients, Sampling points k The process noise covariance matrix is ​​-1. Sampling points k -1 observation value, Sampling points k The observation matrix of -1 Sampling points k The prior system state estimation vector is -1. Sampling points i The prediction error;

[0088] Step S24: Correct the system state vector based on the adaptive adjustment mechanism, collaboratively suppress Gaussian white noise and impulse interference at the sampling points, recursively update the system state vector at the sampling points using Kalman filtering, and output the denoised target useful signal. ;

[0089] ;

[0090] in, Sampling points k The prior system state estimation vector is -1. Sampling points k of

[0091] Prior covariance matrix, Sampling points k The prior covariance matrix of -1 Sampling points k Kalman gain, For updated sampling points k The system state vector, For updated sampling points k The covariance matrix, I It is an identity matrix.

[0092] This invention overcomes the limitations of traditional Kalman filtering with its fixed noise covariance by constructing an error-driven, dynamic covariance adjustment mechanism. Using the filter prediction error as feedback, the process noise covariance is adjusted in real time. When the error increases (e.g., due to sudden impulse interference), the process noise covariance matrix adaptively increases, improving the filter's ability to track time-varying noise. When the error decreases (due to environmental stability), the process noise covariance matrix automatically shrinks, ensuring filter stability. This avoids interference from single-moment errors in noise estimation, achieving a synergistic suppression of Gaussian white noise and impulse interference rather than a compromise.

[0093] This invention effectively improves the noise suppression capability of the beam control module during data processing, avoids beam pointing deviation caused by noise, enhances environmental adaptability, eliminates the need for manual adjustment of filtering parameters, and maintains low error in the filtered output signal under three typical scenarios: Gaussian white noise dominance, impulse interference dominance, and mixed interference.

[0094] Step S3: From the target useful signal Three-dimensional features in the time domain, frequency domain, and spatial domain are extracted, and three-dimensional feature fusion is achieved through a dynamic weight matrix to obtain the weighted fused features of the sampling points. .

[0095] Step S3 specifically includes:

[0096] Step S31: From the target useful signal Extract three-dimensional features in the time domain, frequency domain, and spatial domain;

[0097] Time-domain characteristics are ; They are respectively N The target useful signal at each sampling point The mean, standard deviation, skewness, and kurtosis;

[0098] Frequency domain characteristics are ; The useful signals of the target are respectively The center frequency, frequency bandwidth, maximum spectral power, and total harmonic distortion;

[0099] airspace characteristics are , These are the azimuth angle, elevation angle, rate of change of azimuth angle, and rate of change of elevation angle of the signal, respectively.

[0100] Step S32: Calculate the weights of time-domain features, frequency-domain features, and spatial-domain features. , obtain sampling points k dynamic weight vector ;

[0101] ;

[0102] in, Weights for time-domain features, frequency-domain features, or frequency-domain features. The ratio of signal-to-noise ratio to time-domain features, frequency-domain features, or frequency-domain features. Signal-to-noise ratio (SNR) representing time-domain, frequency-domain, or frequency-domain features. Power as a time-domain feature, frequency-domain feature, or frequency-domain feature. Noise power with time-domain, frequency-domain, or frequency-domain characteristics;

[0103] Step S33: Calculate the weighted fusion features of the sampling points based on the dynamic weight vector. ;

[0104] .

[0105] This invention addresses the problem of one-sided traditional single-dimensional feature information by establishing a feature signal-to-noise ratio-dynamic weight mapping to achieve optimal coupling of features in the time, frequency, and spatial domains. It quantifies the effective information proportion of features in different dimensions, highlights the contribution of frequency features to beam parameters, ensures that fused features always focus on high signal-to-noise ratio features, avoids ineffective features diluting effective information, and resolves the contradiction between feature redundancy and complementarity.

[0106] The utilization rate of effective feature information is improved, avoiding information overload caused by invalid features. This strengthens the foundation for subsequent prediction accuracy. The prediction model with fused features as input provides high-quality feature input for high-precision beam parameter prediction, reducing the pressure on subsequent model correction.

[0107] Step S4: Construct an LSTM-Attention hybrid model to fuse weights and features. As input, the core parameters of the predictive wave control module The core parameters include the pointing angle. Beamwidth and transmit gain .

[0108] Step S4 specifically includes:

[0109] Step S41: Construct an LSTM-Attention hybrid model to fuse weights and features. The data processing procedure of the LSTM-Attention hybrid model is as follows:

[0110] ;

[0111] ;

[0112] ;

[0113] ;

[0114] in, The output vector of the input gate. It is the sigmoid activation function. Here is the input weight matrix of the input gate. Sampling points k -1 is a hidden state. Let be the hidden state weight matrix of the input gate. Sampling points k -1 cell state vector The cell state weight matrix is ​​the input gate. For the input gate bias; This is the output vector of the forget gate. Here is the input weight matrix for the forget gate. Here is the hidden state weight matrix for the forget gate. This is the cell state weight matrix for the forgetting gate. For the input gate bias; Sampling points k The cell state vector, The input weight matrix represents the cell state. The hidden state weight matrix represents the cell state. Bias for cell state; The input weight matrix of the output gate. Let the hidden state weight matrix be the output gate. This is the cell state weight matrix for the output gate. For the output gate bias;

[0115] Step S42: Based on the hidden state corresponding to different sampling points Weight the hidden states and output the weighted hidden states. ;

[0116] ;

[0117] in, For the average hidden state, This is the attention weight matrix. , For the set of real numbers, Sampling points k The corresponding hidden state coefficients, Sampling points kThe corresponding attention weights;

[0118] Step S43: Weight the hidden state Input to a fully connected layer, output the core parameters of the prediction. ;

[0119] ;

[0120] in, These are the predicted output weight matrix and the output bias, respectively. ;

[0121] The loss function of the LSTM-Attention hybrid model is:

[0122] ;

[0123] in, Sampling points k The corresponding true values ​​of the core parameters of the wave control module. The regularization coefficient is set.

[0124] This invention combines the temporal modeling capabilities of LSTM with the feature focusing capabilities of the Attention mechanism to solve the problem that traditional empirical models cannot capture the dynamic correlation of parameters.

[0125] Step S5: Utilize the predicted core parameters of the wave control module The core parameters are corrected based on the real-time output of the wave control module, and the corrected core parameters are used as the core parameters for future output of the wave control module to correct the data processing process of the wave control module.

[0126] Step S5 specifically includes:

[0127] Step S51: Utilize the predicted core parameters and sampling points k The core parameters output in real time by the wave control module Error in calculating core parameters ;

[0128] ;

[0129] Step S52: Based on the error The core parameters are then corrected to obtain the corrected core parameters. ;

[0130] ;

[0131] in, To adjust the step size, it is generally set to , It is a symbolic function;

[0132] Step S53: Utilize the corrected core parameters As the core parameter output by the future wave control module, the data processing process of the wave control module is corrected.

[0133] This invention effectively improves the anti-interference capability of the wave control module in the data processing process, enhances environmental adaptability, and is suitable for wave control module data processing in complex electromagnetic environments, featuring high precision, low latency, and strong anti-interference characteristics.

Claims

1. A data processing method for a wave control module, characterized in that, include: Step S1: Acquire spatial electromagnetic signals through the ADC sampling unit of the wave control module, construct the original data sequence of spatial electromagnetic signals, extract the target useful signal, sample the original data, and standardize the original data at the sampling points to obtain standardized data of the sampling points. ; Step S2: Based on the target useful signal at each sampling point and standardized data An improved adaptive Kalman filter algorithm is used to dynamically suppress Gaussian white noise and impulse interference, and output the denoised target useful signal. ; Step S3: From the target useful signal Three-dimensional features in the time domain, frequency domain, and spatial domain are extracted, and three-dimensional feature fusion is achieved through a dynamic weight matrix to obtain the weighted fused features of the sampling points. ; Step S4: Construct an LSTM-Attention hybrid model to fuse weights and features. As input, the core parameters of the predictive wave control module The core parameters include the pointing angle. Beamwidth and transmit gain ; Step S5: Utilize the core parameters predicted by the wave control module The core parameters are corrected based on the real-time output of the wave control module, and the corrected core parameters are used as the core parameters for future output of the wave control module to correct the data processing process of the wave control module. Step S4 includes: Step S41: Construct an LSTM-Attention hybrid model to fuse weights and features. The data processing procedure of the LSTM-Attention hybrid model is as follows: ； ； ； ； in, The output vector of the input gate. It is the sigmoid activation function. Here is the input weight matrix of the input gate. Sampling points k -1 is a hidden state. Let be the hidden state weight matrix of the input gate. Sampling points k -1 cell state vector The cell state weight matrix is ​​the input gate. For the input gate bias; This is the output vector of the forget gate. Here is the input weight matrix for the forget gate. Here is the hidden state weight matrix for the forget gate. This is the cell state weight matrix for the forgetting gate. For the input gate bias; Sampling points k The cell state vector, The input weight matrix represents the cell state. The hidden state weight matrix represents the cell state. This is a bias for the cell state; The input weight matrix of the output gate. Let the hidden state weight matrix be the output gate. This is the cell state weight matrix for the output gate. For the output gate bias; Step S42: Based on the hidden state corresponding to different sampling points Weight the hidden states and output the weighted hidden states. ; ； in, For the average hidden state, Here is the attention weight matrix. , For the set of real numbers, Sampling points k The corresponding hidden state coefficients, Sampling points k The corresponding attention weights; Step S43: Weight the hidden state Input to a fully connected layer, output the core parameters of the prediction. ; ； in, These are the predicted output weight matrix and the output bias, respectively. ; The loss function of the LSTM-Attention hybrid model is: ； in, Sampling points k The corresponding true values ​​of the core parameters of the wave control module. The regularization coefficient is set.

2. The data processing method for the wave control module according to claim 1, characterized in that, Step S1 includes: Step S11: Acquire spatial electromagnetic signals through the ADC sampling unit of the wave control module to construct the original data sequence of the spatial electromagnetic signals. The original data sequence satisfies... , The raw data collected, Useful signals for the target It is Gaussian white noise. This is an electromagnetic interference signal. t For signal timing; Useful signals of the target , A The signal amplitude, The center frequency of the signal. This is the initial phase; Electromagnetic interference signals Including pulse interference and bandwidth interference , , ; B For the amplitude of interference, C The interference intensity coefficient, For rectangular window functions, The pulse width. This is a function for generating random numbers; Step S12: Sample from the original data sequence to obtain the original data corresponding to each sampling point. , k Number the sampling points; based on the minimum value in the original data sequence. and maximum value For raw data Standardization processing is performed to obtain standardized data from the sampling points. ; 。 3. The data processing method for the wave control module according to claim 2, characterized in that, Step S2 includes: Step S21: Utilize the target's useful signal and the corresponding center frequency and initial phase Constructing the system state vector And construct the state equations; ； in, Here is the state transition matrix. Sampling points k -1 system state vector This is the input control matrix; it is a zero matrix when there is no external input. , Sampling points k -1 control input, Sampling points k -1 process noise, The process noise has a Gaussian distribution. The process noise covariance matrix is... These are the attenuation coefficients for the target useful signal, center frequency, and initial phase, respectively. Sampling points k The covariance of the target useful signal, center frequency, and initial phase; Step S22: Utilize standardized data Construct observation equations; ； in, Sampling points k The observed values, Sampling points k The observation matrix To observe the noise, To observe the noise variance, To observe the Gaussian distribution of the noise; Step S23: Construct an adaptive adjustment mechanism for process noise covariance and observation noise covariance; ； in, M The sliding window length is set during the adaptive adjustment process. i The number of the sampling point in the sliding window. Sampling points k A prediction error of -1, For adaptive adjustment coefficients, Sampling points k The process noise covariance matrix is ​​-1. Sampling points k -1 observation value, Sampling points k The observation matrix of -1 Sampling points k The prior system state estimation vector is -1. Sampling points i The prediction error; Step S24: Correct the system state vector based on the adaptive adjustment mechanism, collaboratively suppress Gaussian white noise and impulse interference at the sampling points, recursively update the system state vector at the sampling points using Kalman filtering, and output the denoised target useful signal. ; ； in, Sampling points k The prior system state estimation vector is -1. Sampling points k of Prior covariance matrix, Sampling points k The prior covariance matrix of -1 Sampling points k Kalman gain, For updated sampling points k The system state vector, For updated sampling points k The covariance matrix, I It is an identity matrix.

4. The data processing method for the wave control module according to claim 3, characterized in that, Step S3 includes: Step S31: From the target useful signal Extract three-dimensional features in the time domain, frequency domain, and spatial domain; Time-domain characteristics are ; They are respectively N The target useful signal at each sampling point The mean, standard deviation, skewness, and kurtosis; Frequency domain characteristics are ; The useful signals of the target are respectively The center frequency, frequency bandwidth, maximum spectral power, and total harmonic distortion; airspace characteristics are , These are the azimuth angle, elevation angle, rate of change of azimuth angle, and rate of change of elevation angle of the signal, respectively. Step S32: Calculate the weights of time-domain features, frequency-domain features, and spatial-domain features. , obtain sampling points k dynamic weight vector ; ； in, Weights for time-domain features, frequency-domain features, or frequency-domain features. The ratio of signal-to-noise ratio to time-domain features, frequency-domain features, or frequency-domain features. Signal-to-noise ratio (SNR) representing time-domain, frequency-domain, or frequency-domain features. Power as a time-domain feature, frequency-domain feature, or frequency-domain feature. Noise power with time-domain, frequency-domain, or frequency-domain characteristics; Step S33: Calculate the weighted fusion features of the sampling points based on the dynamic weight vector. ; 。 5. The data processing method for the wave control module according to claim 4, characterized in that, Step S5 includes: Step S51: Utilize the predicted core parameters and sampling points k The core parameters output in real time by the wave control module Error in calculating core parameters ; ； Step S52: Based on the error The core parameters are then corrected to obtain the corrected core parameters. ; ； in, To correct the step size, It is a symbolic function; Step S53: Utilize the corrected core parameters As the core parameter output by the future wave control module, the data processing process of the wave control module is corrected.

Images