GNSS anti-interference closed-loop phase self-correction method based on deep learning

Abstract

The invention discloses a GNSS anti-interference closed-loop phase self-correction method based on deep learning, and relates to the technical field of satellite navigation and artificial intelligence, and the method comprises the steps: obtaining an anti-interference array output signal and weight information through an array antenna and space-time adaptive processing; extracting a carrier-to-noise ratio, a weight phase feature and historical positioning quality, and constructing a deep learning feature vector; reasoning an array element-level phase correction reference quantity by using a multi-layer perceptron model, and realizing accurate phase compensation of an array output signal by combining phase error layered modeling, quick-change disturbance prediction and weight sensitivity estimation; an anti-interference perception observation covariance dynamic scaling mechanism is introduced into RTK positioning calculation; on-line learning and model updating are triggered through a positioning quality evaluation result, so that the system has an adaptive evolution capability. According to the method, the GNSS carrier phase quality and the positioning precision under the strong interference condition can be remarkably improved, deep fusion of anti-interference processing and high-precision positioning is realized, and the method has important engineering application value.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of satellite navigation and artificial intelligence, in particular to a GNSS anti-jamming closed-loop phase self-correction method based on deep learning. BACKGROUND

[0002] With the increasing density of various types of communication base stations, surveillance radars and security / countermeasure electromagnetic equipment deployed on the ground, the electromagnetic interference received by satellite navigation signals is increasing in intensity and frequency. In order to improve the anti-interference performance, GNSS receivers usually integrate array antennas and space-time adaptive processing (STAP), beamforming, space domain interference suppression and other anti-interference technologies to significantly improve signal reception quality and anti-interference ability.

[0003] However, there is a difficult technical contradiction in the existing anti-interference methods: the change of anti-interference weights will directly introduce additional array phase distortion, which makes the GNSS carrier phase no longer satisfy the phase continuity assumption of the traditional positioning model, thereby causing the differential positioning accuracy to decrease significantly. Especially in the scenes of strong interference, frequent weight update and low signal-to-noise ratio, the receiver carrier phase observation will appear mutation, drift and additional noise, making it difficult to correctly fix the integer ambiguity of the RTK high-precision positioning method, and causing the technical contradiction of "anti-interference but precision decrease". Traditional filtering and other technologies cannot quantize the influence of anti-interference weights on phase in real time, lack array-level phase compensation mechanism, and it is difficult to distinguish the instantaneous phase jitter caused by fast-changing interference from the slow-changing bias caused by hardware drift. SUMMARY

[0004] In order to overcome the defects in the prior art, the present application provides a GNSS anti-jamming closed-loop phase self-correction method based on deep learning, which constructs a closed-loop system from anti-interference output to phase repair, to high-precision positioning, and finally online learning, aiming to solve the problems of serious carrier phase distortion after array anti-interference processing, difficulty in fixing RTK ambiguity and significant decrease in positioning accuracy.

[0005] To achieve the above purpose, the present application adopts the following technical scheme, comprising: A GNSS anti-jamming closed-loop phase self-correction method based on deep learning, comprising the following steps: S1, acquiring the complex baseband signals of each array element of the array antenna, and obtaining an anti-interference array output signal vector using a space-time adaptive processing anti-interference algorithm and an anti-interference weight vector ; wherein the subscript represents an epoch, i.e. a time point; S2, constructing a signal quality feature vector and extracting an anti-interference weight phase feature vector and historical positioning quality feature vector The feature vectors are concatenated to obtain the input feature vectors of the deep learning model. ; S3, input feature vector Inputting the deep learning model yields the phase correction reference vector. ; S4. Decompose the measured phase residual to obtain the phase slowly varying offset. and phase rapid drift The phase rapid change drift prediction is obtained by using the phase rapid change drift of the most recent w epochs. ; S5, Based on phase correction reference vector Phase slow-varying bias and phase rapid change drift prediction The phase compensation vector is obtained. Used to counter interference array output signal vector Apply phase correction to obtain the phase-corrected array output signal vector. ; S6, based on the phase-corrected array output signal vector Perform RTK positioning calculations to obtain the positioning results. With positioning quality indicators ; S7. Confidence level is determined based on ambiguity fixation rate and phase noise index; quality is evaluated based on positioning quality index; and data pairs meeting the set confidence criteria are selected as sample pairs. The deep learning model is trained online when the positioning quality meets the set conditions.

[0006] Preferably, step S1 is as follows: S11. Receive the original radio frequency time domain signal from each element of the array antenna. The original radio frequency time domain signal contains satellite navigation signal, interference signal and noise. Perform down-conversion, bandpass filtering and gain control on the original radio frequency time domain signal of each element to shift the effective components in the working frequency band to intermediate frequency or baseband to obtain the complex intermediate frequency / complex baseband signal of each element. S12. Perform analog-to-digital conversion on the complex intermediate frequency / complex baseband signals of each array element to obtain the discrete complex baseband sampling sequence of each array element; in each epoch... The complex baseband sampled values ​​of each array element are arranged into an array data vector according to the array element index. : ; in, Indicates the first Individual elements in the historical period The complex baseband sampled values; the total number of array elements is ; S13. Within a preset time or snapshot window, based on the array data vector Estimate the interference plus noise covariance matrix covariance matrix The diagonal elements represent the received power of each array element, and the off-diagonal elements represent the spatial correlation between array elements. S14. Based on the estimated interference plus noise covariance matrix and the steering vector of the desired satellite signal An optimization problem for space-time adaptive processing is constructed, which minimizes interference and noise power while satisfying the desired satellite signal gain constraint, and the anti-interference weight vector is obtained by solving the problem. : ; ; in, Indicates the first The weight range of each array element, Indicates the first Individual elements in the historical period The weighted phase adjustment amount is the weighted phase characteristic. Indicates the epoch For the Unit amplitude complex phase rotation applied to each array element channel, Indicates the first Individual elements in the historical period The anti-interference weight, where j represents the imaginary unit; S15. Utilizing anti-interference weight vectors For array data vectors Weighted synthesis is performed to obtain the output signal vector of the anti-interference array. : ; in, Indicates the first Individual elements in the historical period The complex baseband signal after space-time adaptive processing, i.e., the first... Individual elements in the historical period The anti-interference output signal .

[0007] Preferably, step S2 is as follows: S21, Output signal vector of the anti-jamming array Conduct signal quality assessments for each satellite. The carrier-to-noise ratio is calculated using the relevant output power and noise estimates, thus constructing a signal quality feature vector. : ; wherein, denotes the carrier-to-noise ratio of the satellite at the epoch ; denotes the total number of satellites; denotes the useful signal power of the relevant output, denotes the equivalent noise power spectral density or noise estimate value; S22, extracting the anti-interference weight phase feature, extracting the weight phase feature of each array element, i.e. the weight phase adjustment amount from the anti-interference weight vector , to form an anti-interference weight phase feature vector : ; wherein the total number of array elements is ; S23, obtaining the ambiguity fixing rate and the phase noise index of the previous epoch from the RTK positioning solution module , to form a historical positioning quality feature vector : ; S24, concatenating the three feature vectors obtained in the predetermined order to form the input feature vector of the deep learning model: ; wherein, denotes the concatenation operation of the feature vectors; denotes the concatenated feature vector; performing normalization or standardization processing on the concatenated feature vector, so that various features are in a unified dimension and range, to form the final input feature vector .

[0008] Preferably, step S3 is specifically as follows: S31, inputting the input feature vector to the deep learning model of the multi-layer perceptron structure , and sequentially performing linear mapping and activation operation according to the network level: ; and obtaining the output of the deep learning model through the linear mapping of the output layer ; wherein, , denote the weight matrix and bias vector of the i-th layer; ​represents the hidden layer output; the deep learning model has s layers in total; represents a linear rectifier function, i.e., an activation function; represents the current epoch ; represents the model parameters; S32, the output of the deep learning model is a phase correction reference quantity of the s th array element, and a phase correction reference vector is obtained; wherein, represents a phase correction reference quantity of the s th array element obtained by model inference; S33, in order to ensure that the output of the deep learning model does not damage the stability of subsequent phase compensation, an optional consistency check is performed: first, amplitude limiting is performed on the amplitude, and the limiting condition is , is an amplitude threshold value; second, a stability limiting condition is performed on the adjacent epoch change, and the limiting condition is , is a change threshold value; if the limiting condition is triggered, the output of the deep learning model is clipped to obtain a final phase correction reference vector .

[0009] Preferably, step S4 is specifically as shown below: S41, the measured carrier phase observation value of the current epoch is extracted from the RTK observation generation module , and the geometric expected phase is calculated according to the satellite geometric model and the last epoch coordinate prediction value , and the measured phase residual is calculated ; S42, according to the time variation characteristics of the measured phase residual , the phase residual is divided into a slow varying bias term and a fast varying drift term, the Kalman filter is used to extract the slow varying bias term, and the phase slow varying bias quantity is obtained: ; , so as to obtain the phase fast varying drift quantity : ; wherein, represents the historical epoch index before the current epoch ; represents a window set of the Kalman filter; represents the measured phase residual at the historical epoch ; represents an expectation operator; S43, according to the anti-interference weight vector of the current epoch and the anti-interference weight vector of the last epoch ​, calculate weight increment , characterize the weight adjustment of the space-time adaptive processing anti-jamming algorithm due to the change of the current interference environment; S44, construct the approximate linear relationship between the weight increment and the measured phase residual , wherein is the weight-phase sensitivity vector of the current epoch , used to describe the degree of influence of weight change on the output phase; a hybrid algorithm combining Kalman filtering algorithm and expectation maximization algorithm is used to estimate the weight-phase sensitivity vector online: ; represents the weight-phase sensitivity vector S45, construct the phase fast-changing drift amount of the last w epochs as a sequence construct the phase fast-changing drift amount of the last w epochs as a sequence input the timing prediction model to predict the phase fast-changing drift prediction amount of the current epoch : ; wherein represents the timing prediction model.

[0010] Preferably, step S5 is specifically as follows: S51, take the phase slow-changing bias amount as the slow-changing compensation amount ; take the phase fast-changing drift prediction amount as the fast-changing compensation amount ; according to the weight-phase sensitivity vector and the current weight increment , calculate the phase change estimation amount ; S52, weight and fuse the phase correction reference vector , the slow-changing compensation amount , the fast-changing compensation amount , and the phase change estimation amount to form the final phase compensation vector: ; wherein is the fusion weight; ; wherein represents the phase compensation amount of the i-th array element at epoch ; the total number of array elements is ; ; S53, output the signal vector of the anti-jamming array The complex phase rotation is applied element by element to obtain the phase-corrected array output signal , and a phase-corrected array output signal vector is formed: ; wherein, represents an anti-interference array output signal vector of the th element in the anti-interference array output signal vector.

[0011] Preferably, in step S6, a double-difference observation model is used in the RTK positioning solution process, and an observation covariance dynamic scaling mechanism is introduced for ambiguity fixing, as shown below: S61, according to the signal quality feature vector and the anti-interference weight phase feature vector and the phase compensation amount , a weight adaptive covariance matrix is constructed; S62, in the double-difference observation model, the weight adaptive covariance matrix is used to weight the phase-corrected array output signal vector for solving, and combined with ambiguity fixing, a positioning result is obtained, and a quality index is output; wherein, represents the ambiguity fixing rate of the current epoch ; represents the phase noise index of the current epoch ; represents the compensation consistency index of the current epoch , which is determined by the adjacent epoch phase compensation change amount ; represents the satellite geometry index, which is determined by the geometry distribution index output by the receiver.

[0012] Preferably, step S7 is as shown below: S71, the high confidence criterion is that the ambiguity fixing rate satisfies and the phase noise index satisfies , wherein, , are the threshold values of the ambiguity fixing rate and the phase noise index; if the high confidence criterion is satisfied, a sample pair for online learning is constituted; S72, the quality evaluation criterion is that the positioning quality index satisfies , wherein, Positioning quality index threshold value; if the quality evaluation criterion is met, online learning is triggered, and model parameter updating is performed; otherwise, online learning is not triggered, and model parameter updating is not performed S73, using the sample pair of online learning , define the supervised learning loss :

[0013] wherein, indicates the output of the deep learning model; is the model parameter of the current epoch; The random gradient descent optimizer is used to update the model parameters according to the loss gradient:

[0014] wherein, is the online learning rate, indicates the loss gradient; S74, monitor the ambiguity fixing rate in the continuous epochs after the deep learning model is updated , phase noise index , phase compensation change and positioning residual index, If any of the ambiguity fixing rate decreases beyond the set range, the phase noise index rises beyond the set range, the phase compensation change exceeds the set value, and the positioning residual index exceeds the set value, it is determined that the update of the model parameters has a negative effect, and the model rollback mechanism is executed to restore the model parameters to the values before the update, that is as the current optimal model parameter, and the sample pairs in the recent period of time, i.e., the failed sample pairs, are emptied; otherwise, keep as the current optimal model parameter; S75, for the successful sample pairs that are not emptied, add them to the experience pool; S76, use the current optimal model parameter as the model parameter of the next epoch.

[0015] The application also provides an electronic device, which includes a processor, a memory, and a computer program stored on the memory and executable on the processor, wherein the processor implements the GNSS anti-jamming closed-loop phase self-correction method based on deep learning when executing the computer program.

[0016] The application also provides a computer program product, which includes computer programs / instructions, wherein the computer programs / instructions are executed by a processor to implement the GNSS anti-jamming closed-loop phase self-correction method based on deep learning.

[0017] The application has the advantages that: (1) Existing array anti-jamming methods usually suppress interference sources through STAP, MVDR, PI, etc. algorithms, but these weight updates inevitably introduce array element level phase distortion, making the carrier phase observation no longer satisfy the continuity assumption of the RTK model, resulting in a significant decrease in ambiguity fixing rate. Existing technologies mainly rely on analytical modeling, fixed filtering or simple smoothing to compensate for phase deviation, which is difficult to handle the instantaneous fast-changing error caused by weight changes and the slow-changing bias caused by hardware drift at the same time, and also cannot adapt to complex dynamic interference environment. The present application proposes a deep learning driven array element level phase reference reasoning method, which can learn the nonlinear mapping relationship between weight changes and phase distortion by only using the receiver internal observable quantities (carrier-to-noise ratio, anti-jamming weight phase, historical positioning quality), and at the same time, combined with phase error hierarchical modeling and weight sensitivity term, realize high-precision recovery of complex phase disturbance.

[0018] (2) Existing anti-jamming positioning technologies are mostly open processing links, and the anti-jamming module and the RTK positioning module lack interaction, and cannot form effective feedback. The present application innovatively constructs a closed-loop structure of "anti-jamming - phase compensation - RTK positioning quality feedback - online learning", enhances the RTK solving capability through the observation covariance dynamic scaling mechanism of anti-jamming perception, and takes the ambiguity fixing rate, phase noise, compensation consistency and satellite geometry as the confidence control signal of model learning, so that the system can perform self-distillation online update at high quality moments, and through the rollback mechanism to ensure the safety and reliability of the update, realize the adaptive evolution of the anti-jamming positioning system.

[0019] (3) In existing technologies, deep learning is mainly used for signal detection, interference identification, multipath mitigation, etc. and is not used for array element level phase compensation, and more not form a deep learning and RTK ambiguity fixing closed loop coupling system. In the present application, the phase correction reference quantity output by the deep learning model is fused with slow-changing bias compensation and fast-changing disturbance prediction to form a real-time running high-precision phase recovery mechanism, so that the damage of array weight update to RTK performance is effectively eliminated, and the ambiguity fixing rate and positioning stability in strong interference environment are significantly improved. BRIEF DESCRIPTION OF DRAWINGS

[0020] Figure 1 is a flow chart of the method of the present application.

[0021] Figure 2 is an experimental effect diagram (positioning accuracy under anti-jamming condition) of the present application. DETAILED DESCRIPTION

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Depend on Figure 1 As shown, a deep learning-based GNSS anti-interference closed-loop phase self-calibration method includes the following steps: S1. By receiving and processing the array signal through the anti-interference module, the complex baseband signal of each element of the array antenna is obtained, and the output signal vector of the anti-interference array is obtained by using the space-time adaptive processing anti-interference algorithm. With anti-interference weight vector .

[0024] Step S1 is as follows: S11. Receive the original radio frequency time domain signals from each element of the array antenna. The original radio frequency time domain signals simultaneously contain satellite navigation signals, interference signals, and noise. Perform down-conversion, bandpass filtering, and gain control on the original radio frequency time domain signals of each element respectively, and shift the effective components in the BeiDou / GNSS working frequency band to the intermediate frequency or baseband to obtain the complex intermediate frequency / complex baseband signals of each element.

[0025] S12. Perform analog-to-digital conversion on the analog complex intermediate frequency / complex baseband signals of each array element to obtain the discrete complex baseband sampling sequence of each array element; at each epoch (time point). The complex baseband sampled values ​​of each array element are organized into an array data vector according to the element index, providing input for subsequent space-time adaptive processing. As shown below: ; in, Indicates the first Individual elements in the historical period The complex baseband sampled values; in this embodiment, a 4-element matrix array is selected, that is, the total number of array elements is... =4.

[0026] Array data vector Each element in The current complex signal value corresponding to an array element includes complex form signals of satellite navigation signals, interference signals, and noise: ; in, Indicates the first Individual elements in the historical period a real number representing the in-phase component of the signal at epoch T, reflecting the projection of the signal in the cosine (cos) basis direction; a real number representing the in-phase component of the signal at epoch T, reflecting the projection of the signal in the cosine (cos) basis direction; a real number representing the in-phase component of the signal at epoch T, reflecting the projection of the signal in the cosine (cos) basis direction; a real number representing the in-phase component of the signal at epoch T, reflecting the projection of the signal in the cosine (cos) basis direction; a real number representing the in-phase component of the signal at epoch T, reflecting the projection of the signal in the cosine (cos) basis direction;

[0027] S13, estimating the interference-plus-noise covariance matrix R based on the array data vector X within a preset time or snapshot window; S13, estimating the interference-plus-noise covariance matrix R based on the array data vector X within a preset time or snapshot window; S13, estimating the interference-plus-noise covariance matrix R based on the array data vector X within a preset time or snapshot window; S13, estimating the interference-plus-noise covariance matrix R based on the array data vector X within a preset time or snapshot window;

[0028] S14, constructing an optimization problem of space-time adaptive processing based on the estimated interference-plus-noise covariance matrix R and the steering vector of the desired satellite signal a, and solving the anti-interference weight vector w under the premise of satisfying the gain constraint of the desired satellite signal: S14, constructing an optimization problem of space-time adaptive processing based on the estimated interference-plus-noise covariance matrix R and the steering vector of the desired satellite signal a, and solving the anti-interference weight vector w under the premise of satisfying the gain constraint of the desired satellite signal: S14, constructing an optimization problem of space-time adaptive processing based on the estimated interference-plus-noise covariance matrix R and the steering vector of the desired satellite signal a, and solving the anti-interference weight vector w under the premise of satisfying the gain constraint of the desired satellite signal: ; ; wherein, a real number representing the in-phase component of the signal at epoch T, reflecting the projection of the signal in the cosine (cos) basis direction; a real number representing the in-phase component of the signal at epoch T, reflecting the projection of the signal in the cosine (cos) basis direction; a real number representing the in-phase component of the signal at epoch T, reflecting the projection of the signal in the cosine (cos) basis direction; a real number representing the in-phase component of the signal at epoch T, reflecting the projection of the signal in the cosine (cos) basis direction; a real number representing the in-phase component of the signal at epoch T, reflecting the projection of the signal in the cosine (cos) basis direction; a real number representing the in-phase component of the signal at epoch T, reflecting the projection of the signal in the cosine (cos) basis direction; a real number representing the in-phase component of the signal at epoch T, reflecting the projection of the signal in the cosine (cos) basis direction; a real number representing the in-phase component of the signal at epoch T, reflecting the projection of the signal in the cosine (cos) basis direction; a real number representing the in-phase component of the signal at epoch T, reflecting the projection of the signal in the cosine (cos) basis direction. S15, performing weighted synthesis on the array data vector X using the anti-interference weight vector w to obtain an anti-interference array output signal vector y: S15, performing weighted synthesis on the array data vector X using the anti-interference weight vector w to obtain an anti-interference array output signal vector y:

[0029] S15, performing weighted synthesis on the array data vector X using the anti-interference weight vector w to obtain an anti-interference array output signal vector y: S15, performing weighted synthesis on the array data vector X using the anti-interference weight vector w to obtain an anti-interference array output signal vector y: S15, performing weighted synthesis on the array data vector X using the anti-interference weight vector w to obtain an anti-interference array output signal vector y: S15, performing weighted synthesis on the array data vector X using the anti-interference weight vector w to obtain an anti-interference array output signal vector y: S15, performing weighted synthesis on the array data vector X using the anti-interference weight vector w to obtain an anti-interference array output signal vector y: S15, performing weighted synthesis on the array data vector X using the anti-interference weight vector w to obtain an anti-interference array output signal vector y: S15, performing weighted synthesis on the array data vector X using the anti-interference weight vector w to obtain an anti-interference array output signal vector y: S15, performing weighted synthesis on the array data vector X using the anti-interference weight vector w to obtain an anti-interference array output signal vector y: S15, performing weighted synthesis on the array data vector X using the anti-interference weight vector w to obtain an anti-interference array output signal vector y: S15, performing weighted synthesis on the array data vector X using the anti-interference weight vector w to obtain an anti-interference array output signal vector y: S15, performing weighted synthesis on the array data vector X using the anti-interference weight vector w to obtain an anti-interference array output signal vector y: S15, performing weighted synthesis on the array data vector X using the anti-interference weight vector w to obtain an anti-interference array output signal vector y: S15, performing weighted synthesis on the array data vector X using the anti-interference weight vector w to obtain an anti-interference array output signal vector y:

[0030] Each array element in the epoch Weighted phase adjustment amount That is, the weighted phase features are stored so that the anti-interference weighted phase feature vector can be extracted in the subsequent step S2. use.

[0031] The obtained anti-interference array output signal vector This serves as the input for subsequent carrier tracking and observation generation. The anti-interference weight vector... This serves as input for subsequent deep learning feature construction and phase error modeling, thereby bridging the gap between anti-interference and closed-loop phase self-correction.

[0032] S2. Constructing signal quality feature vectors And extract the phase component of the anti-interference weight. And the historical positioning quality feature vector formed by the ambiguity fixation rate and phase noise index of the previous epoch. The three feature vectors are concatenated to obtain the input feature vector of the deep learning model.

[0033] Step S2 is as follows: S21, Output signal vector of the anti-jamming array Conduct signal quality assessments for each satellite. The carrier-to-noise ratio is calculated using the relevant output power and noise estimates, thus constructing a signal quality feature vector. : ; in, Indicates the first A satellite in the era The carrier-to-noise ratio is used to reflect the degree to which the signals of each satellite are preserved after interference suppression; Indicates the total number of satellites; This indicates the useful signal power of the relevant output. This represents the equivalent noise power spectral density or noise estimate.

[0034] S22. Extract the anti-interference weight phase features from... Extract the weighted phase feature of each array element, i.e., the weighted phase adjustment amount. This constitutes the anti-interference weighted phase eigenvector. : ; This is used to reflect the one-time impact of changes in the space-time adaptive anti-interference weights on the array output phase.

[0035] S23. Obtain the ambiguity fixation rate of the previous epoch from the RTK positioning solution module. and phase noise index , constitute a historical positioning quality feature vector : ; wherein, reflects the reliability of the ambiguity fixed in the last epoch, characterizes the phase stability of the carrier observation.

[0036] S24, feature splicing and uniform formatting processing: The obtained three feature vectors are spliced in a predetermined order to form an input feature vector of the deep learning model: ; wherein, denotes a series splicing operation of the feature vector; denotes the spliced feature vector; and the spliced feature vector is normalized or standardized to make various features in a unified dimension and range, forming the final input feature vector .

[0037] S3, based on the deep learning model inference array element level phase correction reference quantity, the input feature vector is input to the deep learning model with multi-layer perception structure, and the phase correction reference vector is obtained through forward propagation.

[0038] Step S3 is as follows: S31, input the input feature vector to the deep learning model with multi-layer perception (MLP) structure , and perform linear mapping and activation operation in sequence according to the network level: ; and get the output of the deep learning model through linear mapping of the output layer; wherein, , denote the weight matrix and bias vector of the i-th layer; denote the hidden layer output; the deep learning model has s layers; denotes a linear rectification function, i.e. an activation function; denotes the model parameters of the current epoch.

[0039] S32, the output of the deep learning model is phase correction reference quantity (phase compensation reference quantity) of the array element, and the phase correction reference vector is obtained; wherein, denotes the i-th element of the phase correction reference vector A phase correction reference quantity of an array element, used to compensate for array phase disturbance introduced by space-time adaptive processing.

[0040] S33, to ensure that the output of the deep learning model does not destroy the stability of the subsequent phase compensation, an optional consistency check is performed on the output of the model: first, the amplitude is limited, and the limit condition is , is the amplitude threshold; second, the stability limit is performed on the adjacent epoch change, and the limit condition is , is the change threshold; if any of the above limits is triggered, the output of the deep learning model is clipped to obtain the final phase correction reference vector .

[0041] S4, phase error hierarchical modeling and short-term dynamic prediction are performed, the measured phase residual is decomposed into a slowly varying bias term and a fast varying drift term, the phase slowly varying bias and the phase fast varying drift are obtained, and the short-term prediction of the phase fast varying drift in the recent window is performed through a lightweight time series prediction model to obtain the phase fast varying drift prediction.

[0042] Step S4 is specifically as follows: S41, the measured carrier phase observation value of the current epoch is extracted from the RTK observation generation module , and the geometric expected phase is calculated according to the satellite geometric model and the last epoch coordinate prediction value , the measured phase residual is calculated , the measured phase residual reflects the combined influence of factors including ionosphere, electromagnetic environment disturbance, phase rotation introduced by anti-interference weight, etc.

[0043] S42, according to the time variation characteristics of the measured phase residual , the phase residual is divided into a slowly varying bias term (representing long-term drift) and a fast varying drift term (representing short-term mutation and transient disturbance), a long-time window (10 seconds) Kalman filter is used to extract the slowly varying bias term, and the phase slowly varying bias is obtained : ; Further, the phase fast varying drift is obtained : ; Among them, represents the historical epoch index before the current epoch ; represents a long-time window set used to extract the slowly varying bias term; represents the measured phase residual at the historical epoch ; an expected operator.

[0044] S43, the anti-jamming weight vector of the current epoch the anti-jamming weight vector of the previous epoch , calculate the weight increment , characterizes the weight adjustment of the space-time adaptive processing anti-jamming algorithm due to the change of the current interference environment.

[0045] S44, construct the approximate linear relationship between the weight increment and the measured phase residual wherein, is the weight-phase sensitivity vector of the current epoch , used to describe the degree of influence of weight change on the output phase; the KF-EM algorithm (a hybrid algorithm combining Kalman filter KF algorithm and expectation maximization EM algorithm) is used to estimate the weight-phase sensitivity vector online: wherein, represents the weight-phase sensitivity vector.

[0046] S45, construct the phase fast-varying drift amount of the last w epochs as a sequence input the sequence into a lightweight timing prediction model to predict the phase fast-varying drift prediction amount of the current epoch : ; wherein, represents the lightweight timing prediction model, which is a timing mapping function for short-term prediction of phase fast-varying disturbance.

[0047] S5, fuse the phase correction reference vector, the phase slow-varying bias amount and the phase fast-varying drift prediction amount to obtain a phase compensation vector, and apply a stability constraint to the anti-jamming array output signal vector apply phase correction to obtain the phase-corrected array output signal vector .

[0048] Step S5 is specifically as follows: S51, take the phase slow-varying bias amount as the slow-varying compensation amount ; take the phase fast-varying drift prediction amount of the transient phase disturbance as the fast-varying compensation amount ; according to the weight-phase sensitivity vector and the current weight increment , calculate the phase change estimation amount caused by the array element weight disturbance of the current epoch ; S52, Phase correction reference vector Slow-varying compensation amount Rapidly changing compensation amount Phase change estimator Weighted fusion is performed to form the final phase compensation vector: ; in, The fusion weights can be adaptively adjusted based on signal quality or interference intensity. ; in, Indicates the first Individual elements in the historical period Phase compensation amount.

[0049] S53, anti-jamming array output signal vector By applying a complex phase rotation to each array element, the phase-corrected array element output signal is obtained. This forms the phase-corrected array output signal vector: .

[0050] S6. Based on the phase-corrected array output signal vector RTK positioning calculation is performed, employing a double-difference observation model and introducing a dynamic scaling mechanism for observation covariance to fix ambiguity, thus obtaining the positioning result. With positioning quality indicators The observation covariance dynamic scaling mechanism adapts the covariance matrix by dynamically adjusting the weights. This enhances the robustness of the location solution under anti-interference conditions.

[0051] Step S6 is as follows: S61. Based on the signal quality feature vector and anti-interference weight phase eigenvector and phase compensation amount Construct a weighted adaptive covariance matrix :

[0052] in, This represents the noise covariance matrix of the observations used in the RTK positioning calculation process. When the positioning residual index is detected... When the noise covariance is large or the phase compensation is unstable, its weight is reduced by increasing the noise covariance of the corresponding observation; when the anti-interference effect is well restored and the phase compensation is stable, the noise covariance is automatically reduced to increase the contribution of effective observations to the solution. Among these, the positioning residual index... The observation residual or the position posterior residual statistics output from the RTK positioning solution process are obtained. represents the satellite elevation angle (i.e., satellite elevation angle); represents the covariance scaling function.

[0053] S62, under the double-difference observation model, the weight adaptive covariance matrix is used to correct the phase of the array output signal vector , and the ambiguity is fixed to obtain the positioning result of the current epoch , and output the positioning quality index ; wherein, represents the ambiguity fixing rate of the current epoch ; represents the phase noise index of the current epoch ; represents the compensation consistency index of the current epoch , which is determined by the adjacent epoch phase compensation change ; represents the satellite geometry index, which is determined by the geometry distribution index output by the receiver.

[0054] S7, according to the ambiguity fixing rate and the phase noise index, the confidence degree is judged, the positioning quality index is evaluated, and the data meeting the set conditions is selected as the sample pair for online learning of the deep learning model and model parameter updating when the positioning quality meets the set conditions.

[0055] Step S7 is as follows: S71, the high confidence criterion is that the ambiguity fixing rate satisfies and the phase noise index satisfies , wherein, , are the threshold values of the ambiguity fixing rate and the phase noise index; if the high confidence criterion is met, the current input feature vector of the deep learning model and the final phase compensation vector are obtained, and the phase correction reference vector output by the deep learning model is regarded as a self-supervised signal to form a sample pair for online learning; S72, the quality evaluation criterion is that the positioning quality index satisfies , wherein, is the threshold value of the positioning quality index; if the quality evaluation criterion is met, online learning is triggered, and the model parameter is updated; otherwise, online learning is not triggered, and the model parameter is not updated. ​This strategy is equivalent to self-imitation learning, that is, the model memorizes its best decision when it successfully learns. The way of constructing the self-supervised signal is equivalent to the self-imitation learning strategy, that is, when the positioning result of the system is determined to be high confidence, the current actual effective phase compensation amount is regarded as an approximate optimal decision, which is used to strengthen the consistency of the output of the model in a similar state.

[0056] S73, using the sample pair of online learning , defining the supervised learning loss :

[0057] wherein, represents the output of the deep learning model; is the model parameter of the current epoch; the loss function gradually strengthens the mapping relationship of “feature vector→optimal compensation amount”, and improves the robustness of future epochs.

[0058] The random gradient descent optimizer is used to update the model parameters according to the loss gradient:

[0059] wherein, is the online learning rate, represents the loss gradient.

[0060] The mini-batch strategy can be selected to cache the recent high-confidence samples to the queue to smooth the parameter update.

[0061] S74, monitoring the ambiguity fixing rate, the phase noise index, the phase compensation change amount and the positioning residual index in the next several epochs (10 epochs are selected in this embodiment) after the deep learning model is updated , If any of the following conditions occurs: the ambiguity fixing rate decreases beyond the set range, the phase noise index rises beyond the set range, the phase compensation change amount exceeds the set value, and the positioning residual index exceeds the set value, it is determined that the update of the model parameter has a negative effect on the system, and the model rollback mechanism is executed to restore the model parameter to the value before the update, that is, as the current optimal model parameter, and the failed sample pairs in the recent period of time are emptied to avoid polluting the model; if none of the above conditions occurs, then is retained as the current optimal model parameter.

[0062] S75, for the successful sample pairs (successful learning) that are not emptied, they are added to the experience pool , and the online learning rate is adjusted according to the current system stability ​​​, if the ambiguity fixing rate continues to rise, increase the online learning rate ; if the phase noise index, phase compensation change or positioning residual index significantly increases, reduce the online learning rate .

[0063] S76, the current optimal model parameters are used as the model parameters for inference at the next epoch.

[0064] Figure 2 The experimental effect diagram (positioning accuracy in the anti-interference case) of the application shows that the positioning accuracy ranges of north, east and sky (elevation) directions in the anti-interference case are all in the centimeter level.

[0065] The above is only a preferred embodiment of the present application, and is not intended to limit the present application. Any modification, equivalent replacement and improvement within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A GNSS anti-interference closed-loop phase self-calibration method based on deep learning, characterized in that, Includes the following steps: S1. Obtain the complex baseband signal of each element of the array antenna, and use the space-time adaptive processing anti-interference algorithm to obtain the anti-interference array output signal vector. With anti-interference weight vector ; where subscript Indicates an epoch, or a point in time; S2. Constructing signal quality feature vectors And extract the anti-interference weight phase feature vector. and historical positioning quality feature vector The feature vectors are concatenated to obtain the input feature vectors of the deep learning model. ; S3, input feature vector Inputting the deep learning model yields the phase correction reference vector. ; S4. Decompose the measured phase residual to obtain the phase slowly varying offset. and phase rapid drift The phase rapid change drift prediction is obtained by using the phase rapid change drift of the most recent w epochs. ; S5, Based on phase correction reference vector Phase slow-varying bias and phase rapid change drift prediction The phase compensation vector is obtained. Used to counter interference array output signal vector Apply phase correction to obtain the phase-corrected array output signal vector. ; S6, based on the phase-corrected array output signal vector Perform RTK positioning calculations to obtain the positioning results. With positioning quality indicators ; S7: Confidence is judged based on ambiguity fixation rate and phase noise index, and quality is evaluated based on positioning quality index. Data with confidence meeting the set conditions are selected as sample pairs, and the deep learning model is trained online when the positioning quality meets the set conditions.

2. The GNSS anti-interference closed-loop phase self-calibration method based on deep learning according to claim 1, characterized in that, Step S1 is as follows: S11. Receive the original radio frequency time domain signal from each element of the array antenna. The original radio frequency time domain signal contains satellite navigation signal, interference signal and noise. Perform down-conversion, bandpass filtering and gain control on the original radio frequency time domain signal of each element to shift the effective components in the working frequency band to intermediate frequency or baseband to obtain the complex intermediate frequency / complex baseband signal of each element. S12. Perform analog-to-digital conversion on the complex intermediate frequency / complex baseband signals of each array element to obtain the discrete complex baseband sampling sequence of each array element; in each epoch... The complex baseband sampled values ​​of each array element are arranged into an array data vector according to the array element index. : ； in, Indicates the first Individual elements in the historical period The complex baseband sampled values; the total number of array elements is ; S13. Within a preset time or snapshot window, based on the array data vector Estimate the interference plus noise covariance matrix covariance matrix The diagonal elements represent the received power of each array element, and the off-diagonal elements represent the spatial correlation between array elements. S14. Based on the estimated interference plus noise covariance matrix and the steering vector of the desired satellite signal An optimization problem for space-time adaptive processing is constructed, which minimizes interference and noise power while satisfying the desired satellite signal gain constraint, and the anti-interference weight vector is obtained by solving the problem. : ； ； in, Indicates the first The weight range of each array element, Indicates the first Individual elements in the historical period The weighted phase adjustment amount is the weighted phase characteristic. Indicates the epoch For the Unit amplitude complex phase rotation applied to each array element channel, Indicates the first Individual elements in the historical period The anti-interference weight, where j represents the imaginary unit; S15. Utilizing anti-interference weight vectors For array data vectors Weighted synthesis is performed to obtain the output signal vector of the anti-interference array. : ； in, Indicates the first Individual elements in the historical period The complex baseband signal after space-time adaptive processing, i.e., the first... Individual elements in the historical period The anti-interference output signal .

3. The GNSS anti-interference closed-loop phase self-correction method based on deep learning according to claim 1, characterized in that, Step S2 is detailed below: S21, Output signal vector of the anti-jamming array Conduct signal quality assessments for each satellite. The carrier-to-noise ratio is calculated using the relevant output power and noise estimates, thus constructing a signal quality feature vector. : ； in, Indicates the first A satellite in the era The carrier-to-noise ratio; Indicates the total number of satellites; This indicates the useful signal power of the relevant output. This represents the equivalent noise power spectral density or noise estimate. S22. Extract the phase features of the anti-interference weights from the anti-interference weight vector. Extract the weighted phase feature of each array element, i.e., the weighted phase adjustment amount. This constitutes the anti-interference weighted phase eigenvector. : ； The total number of array elements is ; S23. Obtain the ambiguity fixation rate of the previous epoch from the RTK positioning solution module. and phase noise index This constitutes the historical positioning quality feature vector. : ； S24. Concatenate the three feature vectors obtained in a predetermined order to form the input feature vector of the deep learning model: ； in, This represents the concatenation operation of eigenvectors; This represents the concatenated feature vector; Normalization or standardization is performed on the concatenated feature vectors to bring all features into a uniform dimension and range, thus forming the final input feature vector. .

4. The GNSS anti-interference closed-loop phase self-correction method based on deep learning according to claim 1, characterized in that, Step S3 is detailed below: S31. Input feature vector Deep learning models input into a multilayer perceptron structure Linear mapping and activation operations are performed sequentially according to the network hierarchy: ； The output of the deep learning model is obtained through a linear mapping of the output layer. ; in, , This represents the weight matrix and bias vector of the i-th layer; This represents the hidden layer output; the deep learning model has a total of s layers; This represents the linear rectification function, also known as the activation function. Indicates the current epoch. Model parameters; S32, the output of the deep learning model is The phase correction reference quantity of each array element is used to obtain the phase correction reference vector. ;in, The first inference obtained from the model represents the... Phase correction reference quantity for each array element; S33. To ensure that the output of the deep learning model does not compromise the stability of subsequent phase compensation, an optional consistency check is performed: First, an amplitude limit is applied to the amplitude, with the following condition: , The first step is to set an amplitude threshold; the second step is to apply stability constraints to changes in adjacent epochs, with the following constraints: , This is the threshold for the amount of change; if the constraint is triggered, the output of the deep learning model is pruned to obtain the final phase correction reference vector. .

5. The GNSS anti-interference closed-loop phase self-correction method based on deep learning according to claim 1, characterized in that, Step S4 is as follows: S41. Extract the measured carrier phase observation value of the current epoch from the RTK observation generation module. The expected geometric phase is calculated based on the satellite geometric model and the coordinate prediction values ​​from the previous epoch. Calculate the measured phase residual ; S42, Based on the measured phase residual The time-varying characteristics of the phase residual divide it into a slowly varying bias term and a rapidly varying drift term. A Kalman filter is used to extract the slowly varying bias term to obtain the slowly varying phase bias. : ； Thus, the phase rapid shift amount is obtained. : ； in, Indicates the current epoch. Previous historical epoch index; The set of windows representing the Kalman filter; Indicates the historical era Measured phase residual at the location; Represents the expectation operator; S43. Based on the anti-interference weight vector of the current epoch. Anti-interference weight vector of the previous epoch Calculate the weight increment , Characterizes the weight adjustment of the spatiotemporal adaptive processing anti-interference algorithm due to changes in the current interference environment; S44. Construct an approximate linear relationship between the weight increment and the measured phase residual. ,in, For the current epoch The weight-phase sensitivity vector is used to describe the degree of influence of weight changes on the output phase. A hybrid algorithm combining the Kalman filter algorithm and the expectation-maximization algorithm is used to estimate the weight-phase sensitivity vector online. ; Represents the weight-phase sensitivity vector; S45. Construct a sequence from the phase rapid shifts of the most recent w epochs. will sequence Input a time series prediction model to predict the current epoch. Phase fast drift prediction : ； in, This represents a time series prediction model.

6. The GNSS anti-interference closed-loop phase self-calibration method based on deep learning according to claim 5, characterized in that, Step S5 is as follows: S51, Change the phase slowly by the bias amount As a slowly varying compensation quantity ; Predicting phase fast drift As a rapid change compensation quantity According to the weight-phase sensitivity vector With current weight increment Calculate the phase change estimate ; S52, Phase correction reference vector Slow-varying compensation amount Rapidly changing compensation amount Phase change estimator Weighted fusion is performed to form the final phase compensation vector: ； in, For weighting; ； in, Indicates the first Individual elements in the historical period Phase compensation amount; total number of array elements is ; S53, anti-jamming array output signal vector By applying a complex phase rotation to each array element, the phase-corrected array element output signal is obtained. This forms the phase-corrected array output signal vector: ； in, Represents the output signal vector of the anti-interference array The Middle The anti-interference output signal of each array element.

7. The GNSS anti-interference closed-loop phase self-correction method based on deep learning according to claim 6, characterized in that, In step S6, a double-difference observation model is used in the RTK positioning solution process, and a dynamic scaling mechanism for observation covariance is introduced to fix ambiguity, as detailed below: S61. Based on the signal quality feature vector and anti-interference weight phase eigenvector and phase compensation amount Construct a weighted adaptive covariance matrix ; S62. In the double-difference observation model, the weighted adaptive covariance matrix is ​​used. Phase-corrected array output signal vector Weighted calculations are performed, and combined with fixed ambiguity, to obtain the positioning result. Output quality indicators ;in, Indicates the current epoch. The ambiguity fixation rate; Indicates the current epoch. Phase noise index; Indicates the current epoch. The compensation consistency index is determined by the phase compensation change between adjacent epochs. Decide; This represents the satellite's geometric parameters, determined by the geometric distribution parameters output by the receiver.

8. The GNSS anti-interference closed-loop phase self-calibration method based on deep learning according to claim 7, characterized in that, Step S7 is detailed below: S71. The high confidence criterion is: the ambiguity fixation rate satisfies... And the phase noise index meets ,in, , The thresholds for ambiguity fixation rate and phase noise index are used; if the high confidence criterion is met, then a sample pair for online learning is formed. ; S72. Quality evaluation criteria are: the positioning quality indicators meet the requirements. ,in, The threshold for locating quality indicators is set; if the quality evaluation criteria are met, online learning is triggered and the model parameters are updated; otherwise, online learning is not triggered and the model parameters are not updated. S73, using online learning samples to... Define supervised learning loss : in, This represents the output of a deep learning model; These are the model parameters for the current epoch; The stochastic gradient descent optimizer is used to update the model parameters based on the loss gradient. in, For online learning rate, Represents the loss gradient; S74. Monitor the ambiguity fixation rate within a consecutive epoch after the deep learning model is updated. Phase noise index Phase compensation change And positioning residual index, If any of the following occurs: ambiguity fixation rate decreases beyond the set range; phase noise index increases beyond the set range; phase compensation change exceeds the set value; or positioning residual index exceeds the set value, then the update of the model parameters is deemed to have a negative effect. The model rollback mechanism is then executed to restore the model parameters to their values ​​before the update. Use these as the current optimal model parameters and clear the sample pairs from the most recent period, i.e., failed sample pairs; otherwise, retain them. As the current optimal model parameters; S75. For successful sample pairs that have not been cleared, add them to the experience pool; S76. Use the current optimal model parameters as the model parameters for the next epoch.

9. An electronic device, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements a deep learning-based GNSS anti-interference closed-loop phase self-calibration method as described in any one of claims 1 to 8.

10. A computer program product, characterized in that, It includes a computer program / instruction that, when executed by a processor, implements the deep learning-based GNSS anti-interference closed-loop phase self-calibration method as described in any one of claims 1 to 8.

Images