Beidou satellite single-point positioning method based on non-direct signal interference suppression

Abstract

The invention provides a Beidou satellite single-point positioning method based on non-direct signal interference suppression. The Beidou satellite single-point positioning method comprises the steps that a pre-training sample data set and an NLOS signal recognition model based on a double-self-attention network are constructed, and pre-training is carried out; inputting the GNSS measurement data and ephemeris data into a pre-trained NLOS signal identification model to predict a signal type; performing correction processing on a weighting method based on an elevation angle by using a prediction result; generating a correction weight matrix; and calculating the position of the user by using a weighted least square single-point positioning algorithm. According to the NLOS signal identification model, two channels are established by using a double-self-attention network; the weighting based on the elevation angle is enhanced by introducing a correction coefficient, and the interference of the NLOS signal on positioning is reduced; and allocating a proper weight to the recognition result of the NLOS signal, and finally calculating the position of the user by using a weighted least square single-point positioning algorithm so as to improve the Beidou satellite positioning precision in the urban complex environment.

Description

Technical Field

[0001] The present invention relates to the field of satellite positioning technology, and in particular to a Beidou satellite single-point positioning method based on indirect signal interference suppression. Background Art

[0002] The BeiDou Global Navigation Satellite System can provide real-time absolute position information, which can realize high-precision positioning services for autonomous driving, smart mobile devices, IoT devices, etc. However, in urban areas, BeiDou satellite signals are easily blocked by buildings, resulting in the reception of non-direct (NLOS) signals. Due to the additional propagation path caused by reflection, NLOS signals will introduce serious deviations in BeiDou Global Navigation Satellite System measurements, thereby reducing positioning accuracy.

[0003] Traditional NLOS interference mitigation methods, such as ray tracing and shadow matching, require 3D building models or 3D maps to identify NLOS signals, and the 3D modeling cost is high. Another approach is to use fisheye cameras to exclude NLOS signals or rotate satellite antennas to mitigate NLOS multipath errors; however, the need for auxiliary equipment limits the scenario applicability of these methods.

[0004] In recent years, machine learning-based methods have been widely used in global navigation satellite system (GNSS) positioning to mitigate the interference of NLOS signals. Such methods, such as support vector machine (SVM) and decision tree (DT), classify NLOS signals by using machine learning models, remove identified NLOS signals from the observation equation, and eliminate NLOS interference. Using deep learning models to identify NLOS signals is a direct and effective method to mitigate NLOS interference and improve GNSS positioning accuracy; when using deep learning models to identify NLOS signals across multiple locations or scenes, environmental features and state changes of incident signals will seriously affect the recognition performance. However, these existing methods usually only consider single-level feature extraction, that is, the instant observation features of the signal itself; existing methods directly eliminate the identified NLOS signals, or directly multiply the predicted probability of the NLOS signal by the weight matrix to reduce NLOS interference, thereby improving positioning accuracy. However, since NLOS interference is mainly affected by the environment and the state of the incident signal, the existing NLOS signal recognition model fails to effectively model the environmental features and signal timing features, resulting in the inability to achieve high-precision NLOS recognition results across multiple locations or scenes.

[0005] Secondly, in the case of severe occlusion, directly eliminating NLOS signals may reduce the number of satellites available to the positioning algorithm, resulting in reduced positioning accuracy; if the predicted probability of the NLOS signal is directly multiplied by the weight matrix, since the classification prediction value is binary (0 or 1), the final assigned weight of the NLOS signal is much lower than the normal value, completely ignoring the role of the NLOS signal, which also leads to reduced positioning accuracy. Summary of the invention

[0006] In view of the shortcomings of the prior art, the present invention provides a Beidou satellite single-point positioning method based on indirect signal interference suppression. The present invention simultaneously models environmental characteristics and signal timing characteristics to improve the recognition performance of NLOS signals across multiple locations or scenes; and the present invention assigns appropriate weights to NLOS signals to ensure the actual number of positioning satellites while alleviating the interference of NLOS signals on satellite positioning accuracy.

[0007] The technical solution of the present invention is: a Beidou satellite single-point positioning method based on indirect signal interference suppression, comprising the following steps:

[0008] S1), obtain satellite GNSS measurement data of multiple locations or scenes and a sky map of the current location; and construct a pre-training sample data set;

[0009] S2), constructing a NLOS signal recognition model based on dual self-attention network;

[0010] S3), using the pre-training sample data set to pre-train the NLOS signal recognition model based on the dual self-attention network;

[0011] S4), inputting the acquired GNSS measurement data and ephemeris data into the pre-trained NLOS signal recognition model of the dual self-attention network to predict the signal type;

[0012] S5), using the prediction result in step S4) to correct the weight determination method based on the elevation angle; and generate a corrected weight matrix;

[0013] S6) Calculate the user's location using a weighted least squares point positioning algorithm.

[0014] Preferably, in step S1), a satellite receiver is used to record satellite GNSS measurement data of multiple locations or scenes, and a camera is used to capture a sky map of the current location. The position of each satellite projected onto the captured sky map is calculated based on the satellite's altitude angle and azimuth in the satellite GNSS measurement data. Whether the satellite is an NLOS signal is determined by whether the satellite projection position overlaps with a building, so that the satellite GNSS measurement data is marked, and finally a pre-training sample data set is constructed.

[0015] Preferably, in step S1), the satellite GNSS measurement data includes a characteristic sequence of observation values ​​of all satellites of multiple consecutive epochs, expressed as:

[0016]

[0017] Preferably, in step S1), the satellite GNSS measurement data includes observation value characteristics of the target satellite, characteristic sequences of historical epochs of the target satellite, and characteristic sequences of all satellites in the current epoch.

[0018] Preferably, in step S1), the observation value characteristics of the target satellite include satellite altitude angle, azimuth angle, carrier-to-noise ratio, and pseudorange residual.

[0019] Preferably, in step S1), the position of each satellite projected onto the captured sky map is calculated based on the altitude angle (elevation angle) and azimuth angle of the satellite in the satellite GNSS measurement data, specifically comprising the following steps:

[0020] S11), obtain the elevation angle of the satellite, and calculate the distance r between the satellite projection point and the image center in the captured image according to the equal solid angle projection calculation formula of the fisheye camera. The calculation formula is:

[0021]

[0022] Where f is the focal length of the camera lens and θ is the elevation angle of the satellite;

[0023] S12), calculating the orientation β of the satellite projection point on the captured image according to the azimuth angle of the satellite and the orientation of the camera lens;

[0024] S13), according to the value r of the satellite projection point from the image center and its orientation β on the image calculated in step S11), the specific position of the projection point on the captured sky map is calculated, that is:

[0025] (r*cosβ,r*sinβ);

[0026] The origin of the coordinates is the center point of the captured image.

[0027] Preferably, in step S2), the NLOS signal recognition model based on the dual self-attention network includes a spatial channel attention module and a temporal channel attention module, as well as a feedforward network and a classifier, and the spatial environment characteristics and the temporal characteristics of the signal are simulated by the spatial channel attention module and the temporal channel attention module.

[0028] Preferably, in step S3), the pre-training process of the NLOS signal recognition model based on the dual self-attention network is:

[0029] S31), inputting the observation value features of the target satellite, the satellite feature sequence of the current epoch, and the feature sequence of the historical epoch of the target satellite into the NLOS signal recognition model of the double self-attention network, and at the same time, the satellite feature sequence of the current epoch is input into the spatial channel attention module; the feature sequence of the historical epoch of the target satellite is fed into the time channel attention module;

[0030] S32), expanding the dimension of the satellite feature sequence and the feature sequence of the historical epoch of the target satellite through the embedding layer, and obtaining the corresponding representation vector X; that is:

[0031] X=W e S+b e ;

[0032] Among them, S is the input of the embedding layer, W e is the weight matrix of the neural network embedding layer, b e is the bias value of the neural network embedding layer;

[0033] S33), in the spatial channel attention module, generating a value vector V1 and a key vector K1 from the satellite feature sequence of the current epoch through two linear mapping layers;

[0034] In the time channel attention module, two linear mapping layers are used to generate the value vector V2 and the key vector K2 from the feature sequence of the historical epochs of the corresponding target satellite;

[0035] Another linear mapping layer is used to generate the query vector Q from the target satellite observation features;

[0036] S34), calculating the attention weight matrix between satellites in the current epoch and the attention weight matrix between historical epochs of the target satellites according to the obtained value vector, key vector, and query vector;

[0037] S35), connecting the attention weight matrix between satellites of the current epoch and the attention weight matrix between historical epochs of the target satellite and transferring them to a feed-forward network, wherein the feed-forward network has a higher dimensional intermediate layer representation, allowing the model to capture more complex patterns and relationships from the fused attention weights;

[0038] S36), input the output result of the feedforward network into the classifier for predicting the label type of the target satellite.

[0039] Preferably, in step S3), during the pre-training process of the NLOS signal recognition model of the dual self-attention network, a small batch gradient descent method is used to optimize the loss function, and the loss function adopts a cross entropy loss function l, which is expressed as:

[0040]

[0041] Among them, x i is the input of the NLOS signal recognition model, y i is the label corresponding to the input, and N is the number of samples.

[0042] Preferably, in step S35), the feedforward network includes two fully connected neural network layers; the attention weight matrix between satellites in the current epoch and the attention weight matrix between historical epochs of the target satellite are connected and transferred to the feedforward network; the calculation formula is:

[0043] F out =Relu([Z S ; Z T ]W1+b1)W2+b2

[0044] In the formula, F out is the output of the feedforward network; Z S and Z T They respectively represent the attention weight matrix between satellites in the current epoch and the attention weight matrix between historical epochs of the target satellite; Relu is the ReLU activation function between two fully connected neural network layers; W1 and W2 are the weight parameters of the two fully connected neural network layers; b1 and b2 are the bias values ​​of the two fully connected neural network layers.

[0045] Preferably, in step S36), the classifier consists of two fully connected layers, and a sigmoid function is used as the activation function of the last fully connected layer of the classifier.

[0046] Preferably, in step S5), the prediction result in step S4) is corrected using a weight determination method based on elevation angle; and the expression for generating the correction weight matrix is:

[0047]

[0048] Where ω is the modified weight matrix; P NLPS is the probability of NLOS predicted by the model; j is the correction coefficient; σ e Denotes weighting based on elevation angle.

[0049] Preferably, in step S5), The expression is:

[0050]

[0051] Where a and b are variable parameters; θ is the satellite altitude angle, that is, the elevation angle.

[0052] Preferably, in step S6), the user's position is calculated based on the modified weight matrix ω, the GNSS measurement data and the ephemeris data, and the weighted least squares point positioning algorithm, and the calculation expression is:

[0053]

[0054] In the formula, represents the user position estimation; A represents the coefficient matrix of unknown parameters; W represents the weight matrix; L represents the free term vector.

[0055] Preferably, in step S6), the user's position is calculated according to the modified weight matrix ω, the GNSS measurement data and the ephemeris data, and the weighted least squares point positioning algorithm, which specifically includes the following steps:

[0056] S61), assuming that the user's initial position is (x0, y0, z0); the offset is (△x, △y, △z);

[0057] S62), based on the GNSS measurement data of the satellite; calculate the estimated pseudorange ρ i , calculation expression:

[0058]

[0059] Among them, P i is the Euclidean distance between satellite i and the user, and c represents the speed of light; and δt r are the clock offsets of satellite i and GNSS receiver respectively; I i and T i Represent the ionospheric and tropospheric errors respectively;∈ represents the observation error;

[0060] S63), linearize equation (1) through Taylor series expansion to obtain equation (2):

[0061]

[0062] Where: is the Euclidean distance between satellite i and the user's initial position; is the unit LOS vector;

[0063] S64) Use l i To express the right side of formula (2), that is:

[0064]

[0065] Therefore, the matrix form of the pseudorange observation equation is obtained

[0066] ∈=AX-L

[0067] Where L represents the free term vector; X = [△x,△y,△z,c·δt r ] T ;

[0068] Where n is the number of captured satellites;

[0069] S65), using the least squares method to solve, the loss function of the least squares method is expressed as:

[0070] J(X)=∈ T ∈=(AX-L) T (AX-L)

[0071] To minimize the loss function, the gradient of J(X) should be:

[0072]

[0073] In the formula, represents the user location estimate;

[0074] At the same time, the measurement deviation between different satellites is considered, and the correction weight matrix ω is introduced to obtain the user position estimate

[0075]

[0076] S66), after calculating the solution of the user's position, replace the user's initial position, and then perform iterative calculation until the result meets the change threshold required for the solution, and finally stop the iteration and output the final position result.

[0077] Preferably, in step S63), the Euclidean distance between the satellite i and the user's initial position The expression is:

[0078]

[0079] Unit LOS vector It is expressed as:

[0080]

[0081] In the formula, (x i ,y i ,z i ) is the position of the satellite, which can be obtained from the ephemeris data.

[0082] The beneficial effects of the present invention are:

[0083] 1. The NLOS signal recognition model of the present invention uses a dual self-attention network to establish two channels: one to learn the space environment information from the satellite sequence of the current epoch, and the other to extract the signal timing information from the historical epoch; and by introducing a correction coefficient to enhance the weighting based on the elevation angle, the interference of the NLOS signal on the positioning is reduced; first, the NLOS signal is identified, and then a suitable weight is assigned to the identified signal, and finally the weighted least squares single point positioning algorithm is used to calculate the user's position, so as to improve the Beidou satellite positioning accuracy in complex urban environments;

[0084] 2. The present invention improves the NLOS recognition accuracy across multiple locations or scenes by simultaneously extracting spatial environment features and signal timing features, thereby suppressing the interference of NLOS signals on the Beidou satellite positioning accuracy;

[0085] 3. The present invention uses the NLOS recognition result to modify the elevation angle-based scheme, and can assign a more reasonable weight to the NLOS signal according to the recognition result. BRIEF DESCRIPTION OF THE DRAWINGS

[0086] Figure 1 It is a schematic diagram of the framework of the NLOS signal recognition model based on the dual self-attention network of the present invention;

[0087] Figure 2 It is a schematic diagram of the flow chart of the single point positioning method of the present invention. DETAILED DESCRIPTION

[0088] The specific implementation of the present invention will be further described below in conjunction with the accompanying drawings:

[0089] like Figure 1 and 2 As shown, this embodiment provides a Beidou satellite single-point positioning method based on indirect signal interference suppression, comprising the following steps:

[0090] S1), obtain satellite GNSS measurement data of multiple locations or scenes and a sky map of the current location; and construct a pre-training sample data set;

[0091] In this embodiment, a satellite receiver is used to record satellite GNSS measurement data of multiple locations or scenes, wherein the satellite GNSS measurement data includes the observation value characteristics of the target satellite, the characteristic sequence of the historical epoch of the target satellite, And the characteristic sequences of all satellites in the current epoch (S1, S2, ..., S n ), where n is the number of satellites in the current epoch; t is the current epoch, and k is the length of the historical epoch.

[0092] The observation value characteristics of the target satellite include satellite altitude angle, azimuth angle, carrier-to-noise ratio, and pseudorange residual.

[0093] This embodiment uses a camera to capture a sky map of the current location, calculates the position of each satellite projected onto the captured sky map based on the satellite's altitude angle and azimuth in the satellite GNSS measurement data, and determines whether the satellite is an NLOS signal by whether the satellite projection position overlaps with a building, so as to mark the satellite GNSS measurement data and finally construct a pre-training sample data set.

[0094] The position of each satellite projected onto the captured sky map is calculated based on the satellite's altitude angle and azimuth angle in the satellite GNSS measurement data, specifically including the following steps:

[0095] S11), obtain the elevation angle of the satellite, and calculate the distance r between the satellite projection point and the image center in the captured image according to the equal solid angle projection calculation formula of the fisheye camera. The calculation formula is:

[0096]

[0097] Where f is the focal length of the camera lens and θ is the elevation angle of the satellite;

[0098] S12), calculating the orientation β of the satellite projection point on the captured image according to the azimuth angle of the satellite and the orientation of the camera lens;

[0099] S13), according to the value r of the satellite projection point from the image center and its orientation β on the image calculated in step S11), the specific position of the projection point on the captured sky map is calculated, that is:

[0100] (r*cosβ,r·sinβ);

[0101] The origin of the coordinates is the center point of the captured image.

[0102] S2), constructing a NLOS signal recognition model based on dual self-attention network;

[0103] The framework diagram of the NLOS signal recognition model constructed in this implementation is shown in Figure 1 The NLOS signal recognition model based on the dual self-attention network described in this embodiment includes a spatial channel attention module and a temporal channel attention module, as well as a feedforward network and a classifier. The spatial channel attention module and the temporal channel attention module are used to simulate the spatial environment characteristics and the temporal characteristics of the signal, thereby improving the recognition ability of the NLOS signal recognition model across multiple locations or scenes.

[0104] S3), using the pre-training sample data set to pre-train the NLOS signal recognition model based on the dual self-attention network; in the pre-training process of the NLOS signal recognition model of the dual self-attention network in this embodiment, a small batch gradient descent method is used to optimize the loss function, and the loss function adopts a cross entropy loss function l, which is expressed as:

[0105]

[0106] Among them, x i is the input of the model, y i is the label corresponding to the input, and N is the number of samples.

[0107] The pre-training process of the NLOS signal recognition model based on the dual self-attention network specifically includes the following steps:

[0108] S31), the observation value characteristics of the target satellite, the satellite characteristic sequence of the current epoch (S1, S2, ..., S n ), the characteristic sequence of historical epochs of the target satellite The NLOS signal recognition model of the dual self-attention network is input, and the satellite feature sequence of the current epoch (S1, S2, ..., S n ) is input into the spatial channel attention module; the characteristic sequence of the historical epochs of the target satellite is fed into the temporal channel attention module;

[0109] S32), expanding the dimension of the satellite feature sequence and the feature sequence of the historical epoch of the target satellite through the embedding layer, and obtaining the corresponding representation vector X; that is:

[0110] X=W e S+b e ;

[0111] Among them, S is the input of the embedding layer, W e is the weight matrix of the neural network embedding layer, b e is the bias value of the neural network embedding layer;

[0112] S33), in the spatial channel attention module, generating a value vector V1 and a key vector K1 from the satellite feature sequence of the current epoch through two linear mapping layers;

[0113] In the time channel attention module, two linear mapping layers are used to generate the value vector V2 and the key vector K2 from the feature sequence of the historical epochs of the corresponding target satellite;

[0114] Another linear mapping layer is used to generate the query vector Q from the target satellite observation features;

[0115] S34), calculate the attention weight matrix between satellites in the current epoch and the attention weight matrix between historical epochs of the target satellites according to the obtained value vector, key vector, and query vector; the calculation expression is:

[0116]

[0117] In the formula, softmax is the activation function; is the scaling factor.

[0118] S35), connecting the attention weight matrix between satellites of the current epoch and the attention weight matrix between historical epochs of the target satellite and transferring them to a feedforward network, which has a higher dimensional intermediate layer representation, allowing the model to capture more complex patterns and relationships from the fused attention weights; wherein the feedforward network includes two fully connected neural network layers; by connecting the attention weight matrix between satellites of the current epoch and the attention weight matrix between historical epochs of the target satellite and transferring them to the feedforward network; its calculation formula is:

[0119] F out =Relu([Z S ; Z T ]W1+b1)W2+b2

[0120] In the formula, F out is the output of the feedforward network; Z S and Z T They respectively represent the attention weight matrix between satellites in the current epoch and the attention weight matrix between historical epochs of the target satellite; Relu is the ReLU activation function between two fully connected neural network layers; W1 and W2 are the weight parameters of the two fully connected neural network layers; b1 and b2 are the bias values ​​of the two fully connected neural network layers.

[0121] S36), inputting the output result of the feedforward network into a classifier for predicting the label type of the target satellite; the classifier is composed of two fully connected layers, and a sigmoid function is used as the activation function of the last fully connected layer of the classifier.

[0122] S4), inputting the acquired GNSS measurement data and ephemeris data into the pre-trained NLOS signal recognition model of the dual self-attention network to predict the signal type;

[0123] S5), using the prediction results in step S4) to correct the weight determination method based on the elevation angle; and generate a corrected weight matrix, the expression of which is:

[0124]

[0125] Where ω is the modified weight matrix; P NLOSis the probability of NLOS predicted by the model; j is the correction coefficient; σ e Denotes weighting based on elevation angle.

[0126] in, The expression is:

[0127]

[0128] Where a and b are variable parameters; θ is the satellite altitude angle, that is, the elevation angle.

[0129] This embodiment allocates more accurate weights to NLOS satellites during the positioning process through a weight determination method based on elevation angle, thereby reducing the impact of NLOS interference on satellite positioning accuracy.

[0130] S6), using the weighted least squares single point positioning algorithm to calculate the user's position. This embodiment uses the weighted least squares single point positioning algorithm to calculate the user's position based on the modified weight matrix ω, as well as the GNSS measurement data and ephemeris data, so as to improve the Beidou satellite positioning accuracy in a complex urban environment, specifically including the following steps:

[0131] S61), assuming that the user's initial position is (x0, y0, z0); the offset is (△x, △y, △z);

[0132] S62), based on the GNSS measurement data of the satellite; calculate the estimated pseudorange ρ i , calculation expression:

[0133]

[0134] Among them, P i is the Euclidean distance between satellite i and the user, and c represents the speed of light; and δt r are the clock offsets of satellite i and GNSS receiver respectively; I i and T i Represent the ionospheric and tropospheric errors respectively;∈ represents the observation error;

[0135] S63), linearize equation (1) through Taylor series expansion to obtain equation (2):

[0136]

[0137] Where: is the Euclidean distance between satellite i and the user's initial position; is the unit LOS vector;

[0138] Wherein, the Euclidean distance between the satellite i and the user's initial position is The expression is:

[0139]

[0140] Unit LOS vector It is expressed as:

[0141]

[0142] In the formula, (x i ,y i ,z i ) is the position of the satellite, which can be obtained from the ephemeris data;

[0143] S64) Use l i To express the right side of equation (2), that is:

[0144]

[0145] Therefore, the matrix form of the pseudorange observation equation is obtained

[0146] ∈=AX-L

[0147] Where L represents the free term vector; X = [△x,△y,△z,c·δt r ] T ;

[0148] Where n is the number of captured satellites;

[0149] S65), using the least squares method to solve, the loss function of the least squares method is expressed as:

[0150] J(X)=∈ T ∈=(AX-L) T (AX-L)

[0151] To minimize the loss function, the gradient of J(X) should be:

[0152]

[0153] In the formula, represents the user location estimate;

[0154] At the same time, the measurement deviation between different satellites is considered, and the correction weight matrix ω is introduced to obtain the user position estimate

[0155]

[0156] S66), after calculating the solution of the user's position, replace the user's initial position, and then perform iterative calculation until the result meets the change threshold required for the solution, and finally stop the iteration and output the final position result.

[0157] The above embodiments and descriptions are only for illustrating the principles and best embodiments of the present invention. Without departing from the spirit and scope of the present invention, the present invention may be subject to various changes and improvements, all of which fall within the scope of the present invention to be protected.

Claims

1. A Beidou satellite single-point positioning method based on indirect signal interference suppression, characterized in that: The following steps are involved: S1), obtaining satellite GNSS measurement data for multiple locations or scenes and a sky map of the current location; And build a pre-training sample dataset; S2), constructing a NLOS signal recognition model based on dual self-attention network; S3), using the pre-training sample data set to pre-train the NLOS signal recognition model based on the dual self-attention network; S4), inputting the acquired GNSS measurement data and ephemeris data into the pre-trained NLOS signal recognition model of the dual self-attention network to predict the signal type; S5), using the prediction result in step S4) to correct the weight determination method based on the elevation angle; and generate a corrected weight matrix; S6) Calculate the user's location using a weighted least squares point positioning algorithm.

2. The Beidou satellite single-point positioning method based on indirect signal interference suppression according to claim 1 is characterized in that: The satellite GNSS measurement data includes the observation value characteristics of the target satellite, the characteristic sequence of the historical epochs of the target satellite, and the characteristic sequence of all satellites in the current epoch.

3. The Beidou satellite single-point positioning method based on indirect signal interference suppression according to claim 2 is characterized in that: In step S1), a satellite receiver is used to record satellite GNSS measurement data of multiple locations or scenes, and a camera is used to capture a sky map of the current location. The position of each satellite projected onto the captured sky map is calculated based on the satellite's altitude angle and azimuth in the satellite GNSS measurement data. Whether the satellite is an NLOS signal is determined by whether the satellite projection position overlaps with a building, so that the satellite GNSS measurement data is marked, and finally a pre-training sample data set is constructed.

4. The Beidou satellite single-point positioning method based on indirect signal interference suppression according to claim 1 is characterized in that: In step S2), the NLOS signal recognition model based on the dual self-attention network includes a spatial channel attention module and a temporal channel attention module, as well as a feedforward network and a classifier. The spatial channel attention module and the temporal channel attention module are used to simulate the spatial environment characteristics and the temporal characteristics of the signal.

5. The Beidou satellite single-point positioning method based on indirect signal interference suppression according to claim 1 is characterized in that: In step S3), the pre-training process of the NLOS signal recognition model based on the dual self-attention network is: S31), inputting the observation value features of the target satellite, the satellite feature sequence of the current epoch, and the feature sequence of the historical epoch of the target satellite into the NLOS signal recognition model of the double self-attention network, and at the same time, the satellite feature sequence of the current epoch is input into the spatial channel attention module; the feature sequence of the historical epoch of the target satellite is fed into the time channel attention module; S32), expanding the dimension of the satellite feature sequence and the feature sequence of the historical epoch of the target satellite through the embedding layer, and obtaining the corresponding representation vector X; that is: X=W e S+b e ; Among them, S is the input of the embedding layer, W e is the weight matrix of the neural network embedding layer, b e is the bias value of the neural network embedding layer; S33), in the spatial channel attention module, generating a value vector V1 and a key vector K1 from the satellite feature sequence of the current epoch through two linear mapping layers; In the time channel attention module, two linear mapping layers are used to generate the value vector V2 and the key vector K2 from the feature sequence of the historical epochs of the corresponding target satellite; Another linear mapping layer is used to generate the query vector Q from the target satellite observation features; S34), calculating the attention weight matrix between satellites in the current epoch and the attention weight matrix between historical epochs of the target satellites according to the obtained value vector, key vector, and query vector; S35), connecting the attention weight matrix between satellites of the current epoch and the attention weight matrix between historical epochs of the target satellite and transferring them to the feedforward network; S36), input the output result of the feedforward network into the classifier for predicting the label type of the target satellite.

6. The Beidou satellite single-point positioning method based on indirect signal interference suppression according to claim 5 is characterized in that: In step S35), the feedforward network includes two fully connected neural network layers; the attention weight matrix between satellites in the current epoch and the attention weight matrix between historical epochs of the target satellite are connected and transferred to the feedforward network; the calculation formula is: F out =Relu([Z S ;WITH T ]W1+b1)W2+b2 In the formula, F out is the output of the feedforward network; Z S and Z T Respectively represent the attention weight matrix between satellites in the current epoch and the attention weight matrix between historical epochs of the target satellite; Relu is the ReLU activation function between two fully connected neural network layers; W1 and W2 are the weight parameters of the two fully connected neural network layers; b1 and b2 are the bias values ​​of the two fully connected neural network layers.

7. The Beidou satellite single-point positioning method based on indirect signal interference suppression according to claim 1 is characterized in that: In step S5), the prediction result in step S4) is corrected using a weight determination method based on elevation angle; and the expression of the correction weight matrix is ​​generated as follows: Where ω is the modified weight matrix; P NLOS is the probability of NLOS predicted by the model; j is the correction coefficient; σ e represents the weighting based on elevation angle, where The expression is: Where a and b are variable parameters; θ is the elevation angle of the satellite.

8. The Beidou satellite single-point positioning method based on indirect signal interference suppression according to claim 7 is characterized in that: In step S6), the user's position is calculated based on the modified weight matrix ω, the GNSS measurement data and the ephemeris data, and the weighted least squares point positioning algorithm, and the calculation expression is: In the formula, represents the user position estimation; A represents the coefficient matrix of unknown parameters; W represents the weight matrix; L represents the free term vector.

9. The Beidou satellite single-point positioning method based on indirect signal interference suppression according to claim 8 is characterized in that: In step S6), the user's position is calculated according to the modified weight matrix ω, the GNSS measurement data and the ephemeris data, and the weighted least squares point positioning algorithm, which specifically includes the following steps: S61), assuming that the user's initial position is (x0, y0, z0); the offset is (△x, △y, △z); S62), based on the GNSS measurement data of the satellite; calculate the estimated pseudorange ρ i , calculation expression: Among them, P i is the Euclidean distance between satellite i and the user, and c represents the speed of light; and δt r are the clock offsets of satellite i and GNSS receiver respectively; I i and T i Represent the ionospheric and tropospheric errors respectively;∈ represents the observation error; S63), linearize equation (1) through Taylor series expansion to obtain equation (2): Where; is the Euclidean distance between satellite i and the user's initial position; is the unit LOS vector; S64) Use l i To express the right side of formula (2), that is: Therefore, the matrix form of the pseudorange observation equation is obtained ∈=AX-L Where L represents the free term vector; X = [△x,△y,△z,c·δt r ] T ; Where n is the number of captured satellites; S65), using the least squares method to solve, the loss function of the least squares method is expressed as: J(X)=∈ T ∈=(AX-L) T (AX-L) To minimize the loss function, the gradient of J(X) should be: In the formula, represents the user location estimate; At the same time, the measurement deviation between different satellites is considered, and the correction weight matrix ω is introduced to obtain the user position estimate S66), after calculating the solution of the user's position, replace the user's initial position, and then perform iterative calculation until the result meets the change threshold required for the solution, and finally stop the iteration and output the final position result.

10. The Beidou satellite single-point positioning method based on indirect signal interference suppression according to claim 1, characterized in that: In step S63), the Euclidean distance between the satellite i and the user's initial position is The expression is: Unit LOS vector It is expressed as: In the formula, (x i ,y i ,z i ) is the position of the satellite, which can be obtained from the ephemeris data.

Images