Low-orbit navigation enhanced global rapid precision positioning method
By constructing a joint positioning model of Beidou satellite and low-Earth orbit satellite, combining Kalman filtering and machine learning methods, the problem of long convergence time of GNSS positioning is solved, and fast and precise positioning and high-precision solution are achieved.
Patent Information
- Application Number
- CN202510644064.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-19
- Publication Date
- 2025-07-18
AI Technical Summary
The existing GNSS precision single-point positioning technology has a long convergence time, limiting its application scenario range.
A joint positioning model based on the Beidou satellite navigation system and low-earth orbit satellite was constructed, and Kalman filtering combined with variance component estimation was used to determine the parameters to be estimated. Through machine learning, the fixed carrier phase full-circumference ambiguity is used, and the geometric image changes of low-orbit satellites are used to remove errors and improve positioning accuracy and success rate.
The GNSS low-orbit satellite fusion is realized quickly and precisely positioned, shortening the positioning convergence time, and improving the positioning accuracy and the success rate of ambiguity solution.
Smart Images
Figure CN120334959A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of low-earth orbit satellite enhanced global rapid precise positioning, and specifically relates to a global rapid precise positioning method based on low-earth orbit satellite information and signal enhancement. Background Art
[0002] At present, the precise point positioning technology of GNSS (Global Navigation Satellite System) uses multi-frequency pseudorange and carrier phase observations, combines satellite precise orbit and clock error correction information, adopts models to correct error sources such as relativity and earth rotation, and estimates the receiver position and tropospheric parameters in real time. Finally, global centimeter-level precision position service can be achieved. However, in the existing technology, due to the influence of various errors, the convergence time of the GNSS satellite precise point positioning technology takes more than 10 minutes, thus limiting its application scenario range. Summary of the Invention
[0003] Aiming at the deficiencies of the existing technology, the present invention proposes a global rapid precise positioning method based on low-earth orbit satellite information and signal enhancement to solve the technical problem of long GNSS positioning convergence time existing in the existing technology.
[0004] The technical solution adopted by the present invention is as follows:
[0005] In the first aspect, a low-orbit navigation enhanced global rapid precise positioning method is provided, including the following steps: constructing a joint positioning model based on the Beidou satellite navigation system and low-earth orbit satellites;
[0006] Performing error correction on the joint positioning model to obtain a precise positioning model;
[0007] Using Kalman filtering combined with variance component estimation for the precise positioning model to determine the parameters to be estimated, where the parameters to be estimated include the receiver position, clock error, and carrier phase integer ambiguity;
[0008] Extracting a ratio threshold for the carrier phase integer ambiguity through machine learning, and fixing the integer ambiguity according to the ratio threshold;
[0009] Updating the baseline result according to the fixed integer ambiguity to obtain the receiver position coordinates.
[0010] Furthermore, when constructing the joint positioning model based on the Beidou satellite navigation system and low-earth orbit satellites, a dual-frequency ionosphere-free model is adopted. The dual-frequency pseudorange and carrier phase observations of the Beidou satellite navigation system and low-earth orbit satellites are respectively formed into ionosphere-free observation equations; the error of the low-order term of the ionosphere delay is removed through the dispersion characteristic of the ionosphere dispersion medium on the propagation of electromagnetic waves, and an ionosphere-free model is respectively formed between the pseudorange and carrier phase observations of different frequencies.
[0011] Furthermore, for the satellites observed by the receiver, the observation equation expressions of the combined observations of the pseudorange P and the carrier phase L are as follows:
[0012]
[0013] In the above formula, P represents the pseudorange from the satellite to the receiver, and L represents the carrier phase; the subscript r represents the receiver, the subscript s represents the satellite, the superscript B represents the BeiDou satellite navigation system, and the superscript O represents the low Earth orbit satellite;
[0014] is the actual distance between the satellite and the receiver, c is the propagation speed of light in a vacuum, dt r is the receiver clock error, dt s is the satellite clock error, T represents the tropospheric delay along the slant path, b r,IF is the hardware delay of the receiver's code pseudorange, is the hardware delay of the code pseudorange at the satellite end, and e represents the pseudorange observation noise;
[0015] λ IF is the carrier wavelength of the ionosphere-free combination, B r,IF is the hardware delay of the phase at the receiver end, is the hardware delay of the phase at the satellite end, represents the ambiguity of the ionosphere-free combination, and ε represents the phase observation noise.
[0016] Furthermore, when correcting the errors of the joint positioning model, the correction parameters include orbit error, clock error, ionospheric delay, tropospheric delay, antenna phase center, relativistic effect, phase winding, and loading tide.
[0017] Furthermore, when using Kalman filtering combined with variance component estimation to determine the parameters to be estimated for the precise positioning model, the method of determining the weights of the low Earth orbit satellites and the BeiDou system in the precise positioning model by using the a posteriori variance estimation to determine the weights includes the following steps:
[0018] Construct the prediction residual and the variance-covariance matrix, as well as the system matrix of the carrier phase and the pseudorange;
[0019] According to the prediction residual and the variance-covariance matrix, and the system matrix of the carrier phase and the pseudorange, construct the error equations of the low Earth orbit satellites and the BeiDou system;
[0020] Set the weight matrix of the low Earth orbit satellites and the BeiDou system;
[0021] Adjust the weights of the low Earth orbit satellites and the BeiDou system by estimating the unit weight variance through variance component estimation.
[0022] Furthermore, extracting the ratio threshold of the carrier phase integer ambiguity through machine learning includes:
[0023] Obtain the optimal and sub - optimal integer ambiguity vectors;
[0024] Calculate the ratio threshold between the optimal and sub - optimal ambiguities;
[0025] Taking the ratio threshold as the output, and taking the solution success rate, number of satellites, frequency points, and number of epochs as the inputs, use the ratio threshold model based on convolutional neural network to extract the features of the input data at different nodes and the features of different time series respectively, fuse the extracted features, and extract the ratio threshold according to the fused features.
[0026] Furthermore, fixing the integer ambiguity according to the ratio threshold includes:
[0027] Calculate the optimal n - group ambiguity solution results;
[0028] Based on the ratio threshold, judge the reliability of the ambiguity solution according to the following formula:
[0029]
[0030] In the above formula, r is the ratio threshold obtained after machine learning, represents the floating - point solution of the ambiguity, is the optimal ambiguity value, is the sub - optimal ambiguity value, is the variance - covariance matrix;
[0031] Take the ambiguity that meets the requirements of the solution reliability as the fixed integer ambiguity.
[0032] Furthermore, update the baseline result according to the following formula:
[0033]
[0034] In the above formula, is the floating - point solution baseline vector, is the covariance between the baseline vector and the ambiguity, is the variance - covariance matrix, is the optimal ambiguity value, represents the floating - point solution of the ambiguity; is the receiver coordinate obtained by the solution.
[0035] In the second aspect, an electronic device is provided, including:
[0036] One or more processors; a storage device for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the low - earth - orbit navigation - enhanced global rapid precise positioning method provided in the first aspect.
[0037] In a third aspect, a computer program product is provided, including computer programs / instructions which, when executed by a processor, implement the steps of the low-earth orbit navigation enhanced global rapid precise positioning method provided in the first aspect.
[0038] As can be seen from the above technical solutions, the beneficial technical effects of the present invention are as follows:
[0039] Taking full advantage of the fast change of the geometric image of low-earth orbit satellites and the large amount of GNSS satellite observation data, a Kalman filtering method with variance component estimation is adopted to appropriately weight the observation data of low-earth orbit satellites and GNSS; a fuzzy ambiguity test method based on artificial intelligence is used to improve the positioning accuracy of GNSS low-earth orbit integrated rapid precise positioning and the success rate of ambiguity resolution. Description of the Drawings
[0040] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.
[0041] Figure 1 It is a schematic diagram of the LEO / BeiDou PPP system according to an embodiment of the present invention;
[0042] Figure 2 It is a schematic diagram of the flow of the positioning method according to an embodiment of the present invention. Detailed Embodiments
[0043] The following will describe in detail the embodiments of the technical solutions of the present invention with reference to the drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention, and thus are only examples and cannot be used to limit the protection scope of the present invention.
[0044] It should be noted that unless otherwise specified, the technical terms or scientific terms used in this application should have the ordinary meaning understood by those skilled in the art to which the present invention belongs.
[0045] Embodiment
[0046] As Figure 1 shown, low-earth orbit navigation enhanced BeiDou precise point positioning is to add a group of low-earth orbit satellites as navigation satellites on the basis of the BeiDou system, and use the fast geometric change of low-earth orbit satellites to accelerate the convergence of precise point positioning. Low-earth orbit satellites can simultaneously broadcast dual-frequency navigation signals and enhancement information, and low-earth orbit satellites have the characteristic of fast geometric configuration change in a short time, which can accelerate the convergence time of GNSS precise point positioning; at the same time, users receive the measurement signals and precise ephemerides broadcast by low-earth orbit satellites to achieve global precise point positioning.
[0047] Combined with the above LEO (Low Earth Orbit) / Beidou PPP system, this embodiment provides a global rapid precise positioning method based on low-orbit satellite information and signal enhancement, including the following steps:
[0048] S1. Construct a joint positioning model based on the Beidou satellite navigation system and low Earth orbit satellites
[0049] In this embodiment, a dual-frequency ionosphere-free model is adopted. The dual-frequency pseudorange and carrier phase observations of the Beidou satellite navigation system (hereinafter referred to as the Beidou system) and low Earth orbit satellites (hereinafter referred to as low-orbit satellites) are respectively composed into ionosphere-free observation equations. By using the dispersion characteristics of the ionospheric dispersive medium for electromagnetic wave propagation to remove the errors of the low-order terms of the ionospheric delay, an ionosphere-free model is respectively formed between the pseudorange and carrier phase observations of different frequencies. For the satellites observed by the receiver, the observation equation expressions of the combined observations of the pseudorange P and the carrier phase L are as follows:
[0050]
[0051] In the above formula, P represents the pseudorange from the satellite to the receiver, and L represents the carrier phase; the subscript r represents the receiver, the subscript s represents the satellite, the superscript B represents the Beidou satellite navigation system BDS, and the superscript O represents the low Earth orbit satellite LEO;
[0052] is the actual distance between the satellite and the receiver, c is the propagation speed of light in a vacuum, dt r is the receiver clock offset, dt s is the satellite clock offset, T represents the slant path tropospheric delay, b r,IF is the receiver's code pseudorange hardware delay, is the satellite-side code pseudorange hardware delay, e represents the pseudorange observation noise;
[0053] λ IF is the ionosphere-free combined carrier wavelength, B r,If is the receiver-side phase hardware delay, is the satellite-side phase hardware delay, represents the ambiguity of the ionosphere-free combination, and ε represents the phase observation noise.
[0054] Due to the different orbital altitudes of Beidou satellites and low-Earth orbit (LEO) satellites, the orbital perturbation forces, atmospheric drag, and atmospheric time delay during signal propagation they are subject to are all different. For integrated navigation and positioning, it is necessary to consider the differences in the orbital accuracy and satellite clock accuracy between Beidou and LEO satellites, as well as the differences in the noise of observation data, and construct a reasonable model by combining the contribution of LEO observation data to integrated positioning to maximize the characteristics of LEO augmentation. In the integrated positioning model of this embodiment, different weight settings for LEO satellites and the Beidou system will affect the final positioning accuracy of the solution; based on the existing orbital and clock error accuracies of LEO satellites, the initial weights of LEO satellites and the Beidou system can be determined as 1:1 during the initial solution.
[0055] For the measurement noise of the observations, a random model based on the elevation angle is generally used to describe the accuracy of the observations. In the integrated positioning model, the measurement noise of the observations is a function related to the satellite elevation angle, as follows:
[0056]
[0057] In the above formula, σ is the measurement noise of the observations, with the unit of m; a 2 represents the observation noise of high-elevation satellites, and h is the satellite elevation angle.
[0058] S2. Perform error correction on the integrated positioning model to obtain a precise positioning model
[0059] Analyze the positioning error terms of the integrated positioning model, and divide the error terms into two categories: space signal error and model error. The space signal error is the precise orbit and clock difference of the satellite, and the model error refers to the residual of the error terms after model-based correction. System error sources such as relativistic effects, phase wrapping errors, solid tides, ocean tides, and polar tides are considered to be precisely correctable.
[0060] The errors of the precise positioning model mainly include atmospheric time delay errors such as ionospheric and tropospheric errors, and the residual errors of other geophysical models such as antenna phase center and tidal correction are at the mm level. Usually, the ionospheric error in high-precision positioning is eliminated by dual-frequency correction, and the residual error is generally less than 2 mm. The real-time estimation accuracy of the tropospheric error is about 1 cm. Therefore, the comprehensive impact of the errors in the model is about 1 cm. Table 1 summarizes the main errors and correction methods for the combined precise point positioning of the Beidou system and LEO satellites.
[0061] Table 1 Error Model and Processing Strategy
[0062]
[0063]
[0064] Combined with the error terms in Table 1 and the correction methods for each error term, the error of the joint positioning model is modified, including orbit error, clock error, ionospheric delay, tropospheric delay, antenna phase center, relativistic effect, phase wrapping, and loading tide; the corrected joint positioning model is a precise positioning model.
[0065] S3. Use Kalman filtering combined with variance component estimation for the precise positioning model to determine the parameters to be estimated, including receiver position, clock error, and carrier phase integer ambiguity
[0066] The receiver position, clock error, troposphere, and other parameters to be estimated in the precise positioning model can be solved by using the Kalman filtering estimation method. The process of Kalman filtering is as follows:
[0067]
[0068] In the above formula, is the predicted state vector, T k|k-1 is the state transition matrix of Kalman filtering, is the state vector at epoch k-1, is the state vector at the current epoch k; y k is the ionosphere-free pseudorange and carrier phase combined observation value, H k is the coefficient matrix of the observation value, K k is the Kalman filter gain matrix.
[0069] The gain matrix K of Kalman filtering k is:
[0070]
[0071] In the above formula, P k|k-1 is the variance-covariance matrix of the one-step prediction state vector, H k is the coefficient matrix of the observation value, * represents transpose, R k is the observation noise matrix, which can be expressed in matrix form after obtaining the observation value measurement noise through formula (2).
[0072] The variance-covariance matrix P k|k-1 is:
[0073]
[0074] In the above formula, P k|k-1 is the variance-covariance matrix of the further prediction state vector, P k-1|k-1 is the variance-covariance matrix of the state vector at epoch k-1, P k|k is the variance-covariance matrix of the state vector at the current epoch k, * represents transpose; Q kis the process noise matrix, E is the identity matrix, and K k is the Kalman filter gain matrix, and H k is the coefficient matrix of the observed values.
[0075] In some embodiments, when using the Kalman filter estimation method for calculation, in order to achieve more accurate positioning, the a posteriori variance estimation method for determining weights is adopted to determine the weights of the low-earth orbit satellites and the Beidou system in the precise positioning model. Specifically, the a posteriori variance estimation method for determining weights calculates the residual vector, estimates the variances and covariances of different categories of observed values, so as to judge whether the weight distribution between different categories of observed values is reasonable; if the weight distribution of various observed values at the beginning of the calculation iteration is unreasonable, re-weighting can be performed according to the variances and covariances estimated a posteriori. When performing a posteriori variance estimation for determining weights, the predicted residual v k and its variance-covariance matrix are:
[0076]
[0077]
[0078] In the above formula, y k is the ionosphere-free pseudorange and carrier phase combined observed value, H k is the coefficient matrix of the observed values, is the predicted state vector, R k is the observation noise matrix, and P k|k-1 is the variance-covariance matrix of the further predicted state vector, and * represents the transpose.
[0079] Let H = [G1, G2] T , where G1 and G2 respectively correspond to the system matrices of the carrier phase and the pseudorange. Therefore, the error equations of the low-earth orbit satellites and the Beidou system are respectively:
[0080]
[0081] In the above formula, y1 is the carrier phase observed value, y2 is the pseudorange observed value, G1 is the carrier phase system matrix, G2 is the pseudorange system matrix, is the predicted state vector.
[0082] Since the two systems are independent of each other, their weight matrices can be respectively set as D1 and D2, and the complete weight matrix
[0083] The variance component estimation uses the quadratic form v of each category of observed quantities i T Dv i to estimate the variance of unit weight δ 0i , and the calculation formula for variance estimation is as follows:
[0084]
[0085] In the above formula, v1 is the error equation of the low-earth orbit satellite, v2 is the error equation of the Beidou system, D1 is the weight matrix of the low-earth orbit satellite, and D2 is the weight matrix of the Beidou system; in the calculation formula of parameter S, n represents the number of frequency observables in the Doppler positioning process of the low-earth orbit satellite, m represents the number of Beidou system satellites observed at the positioning moment, N1 = G1 T D1G1, N2 = G2 T D2G2, N = N1 + N2, and tr() is the trace function of the matrix.
[0086] In some embodiments, the calculation amount required by Equation (9) is very large. To reduce the calculation difficulty, Equation (9) can be simplified to establish a fast approximate calculation method, and the simplified calculation formula is as shown in Equation (10):
[0087]
[0088] In the above formula, the definitions of each parameter are the same as those in Equation (9).
[0089] When δ 01 and δ 02 are calculated, the weights of the two types of observations can be readjusted according to the following formula (11):
[0090]
[0091] In the above formula, θ is a constant, which can be taken as one of δ 01 and δ 02 , and D i is the initial weight of the low-earth orbit satellite and the Beidou system, with a value of 1:1.
[0092] Substitute the weight result obtained from Equation (11) into Equation (3) for Kalman filtering, and the floating-point solutions of the parameters to be estimated, the receiver position, clock error, and carrier phase ambiguity can be obtained and its variance-covariance matrix At this time, the LAMBDA ambiguity search method can be used to search for the optimal and sub-optimal integer ambiguity vectors is the integer ambiguity vector closest to the real ambiguity, is the integer ambiguity vector sub-closest to the real ambiguity.
[0093] S4. Extract the ratio threshold of the carrier phase integer ambiguity through machine learning, and fix the integer ambiguity according to the ratio threshold
[0094] After obtaining the optimal and sub-optimal integer ambiguity vectors, it is necessary to calculate the separability index Ratio value (ratio threshold) between the optimal and sub-optimal ambiguities:
[0095]
[0096] In the above formula, is the integer ambiguity vector closest to the real ambiguity, is the integer ambiguity vector that is the second closest to the real ambiguity.
[0097] The corresponding solution success rate is:
[0098]
[0099] In the above formula (13), φ is calculated in the following way:
[0100]
[0101] In the above formula (14): is the ambiguity variance matrix The conditional variance after LDL T decomposition, and z is a parameter.
[0102] is the ratio threshold for extracting the carrier phase integer ambiguity. In this embodiment, a RATIO threshold model based on a convolutional neural network is constructed. The RATIO value is used as the output, and the solution success rate, number of satellites, frequency points, and number of epochs are all input into the convolutional neural network model to extract the features of the input data at different nodes and the features of different time series respectively. Then, the extracted features are fused, and the RATIO threshold and fixed integer ambiguity are predicted based on the fused features to determine whether the solution of the integer ambiguity is correct.
[0103] The stable carrier phase integer ambiguity ratio threshold result is output through machine learning, specifically as follows:
[0104] (1) Data acquisition: Select several candidate values for the ambiguity threshold, and collect data such as RATIO values and the number of satellites at each target position multiple times. The data collected for the i-th time is xi.
[0105] (2) Random time series data fusion: Combine multiple one-dimensional data x i at the same position into two-dimensional data, denoted as X κ , X κ is the κ-th data of the input data matrix X, and the corresponding output data is y κ .
[0106] (3) Model training: First, the data X κFeed it into a convolutional neural network (CNN), and use the CNN to extract data features to obtain signal features. Then, use the fully connected layer of the CNN for prediction, and the output is the position where the grid is located.
[0107] In a specific implementation, the data input into the CNN passes through two different channels. The first channel includes 2 CNN layers (both with a convolutional kernel size of 2×2), 2 Dropout layers, and 1 flattening layer. The second channel is 1 flattening layer and 1 fully connected layer. The outputs of the two channels are added together and then classified through 3 fully connected layers to obtain the category to which the input data belongs. The second channel directly connects the output from the previous layer to the input of the next layer, thereby reducing the number of parameters in this channel. The fewer the number of parameters, the less likely it is to overfit.
[0108] The convolutional layer in the CNN is the core part. It extracts the features of the input data through convolutional operations and outputs a two-dimensional feature map. First, send the data into the convolutional layer. The convolutional operation can be regarded as a sliding window. Multiply the data within each position window element-wise with the convolutional kernel, and then add the results to obtain an element in the output feature map. The mathematical expression for the two-dimensional convolutional operation is:
[0109]
[0110] where: h i,j is an element in the output feature map, w m,n is the weight of the convolutional kernel, x i+m,j+n is the corresponding element in the input data, b is the bias term, A and B represent the size of the convolutional kernel, and i and j are the row and column indices of the calculation result. Then, pass the calculation result of the convolutional layer through the activation layer for non-linear transformation to obtain where σ is the activation function, is the calculation result of the activation layer. Finally, pass through the pooling layer to reduce the spatial size of the output of the convolutional layer to obtain the calculation result which is the output feature map of the convolutional layer.
[0111] Select 4 groups of data for fusion. Each group of data has 12 eigenvalues (the data within the dashed box is one group), that is, the original data is , and the fused data is , select the size of the convolutional kernel to be . The convolutional neural network localization model can not only extract the connections between different nodes, but also extract the connections between different time series. For example, the value of is the of the first group of data and It is obtained by multiplying with the convolution kernel, which includes the connections between data of different nodes. The value of is for the first group of data and It is obtained by multiplying with the convolution kernel, which contains the connections between data of different time series.
[0112] (4) Output the trained model
[0113] The convolutional neural network model performs feature extraction respectively. The convolutional layer is used to extract data such as RATIO values and the number of satellites and output the connections between RATIO values. The fully connected layer is used to map low-dimensional features to a high-dimensional feature space, and finally add them up to obtain the judgment of the RATIO threshold.
[0114] When fixing the integer ambiguity according to the RATIO threshold, first calculate the ambiguity according to the following formula:
[0115]
[0116] In the above formula, is the optimal solution result of n groups of ambiguities, z is the integer ambiguity candidate vector, represents the floating-point solution of the ambiguity, is the covariance matrix.
[0117] Then, based on the RATIO threshold, check and judge the reliability of the ambiguity calculation according to the following formula:
[0118]
[0119] In the above formula, r is the ratio threshold obtained after machine learning, represents the floating-point solution of the ambiguity, is the optimal ambiguity value, is the sub-optimal ambiguity value, is the variance-covariance matrix.
[0120] Take the ambiguity that meets the requirements of the calculation reliability as the fixed integer ambiguity.
[0121] S5. Update the baseline result according to the fixed integer ambiguity to obtain the receiver position coordinates
[0122] After the ambiguity test passes, update the baseline result according to the following formula:
[0123]
[0124] In the above formula, is the floating-point solution baseline vector, is the covariance of the baseline vector and the ambiguity, is the variance-covariance matrix, is the optimal ambiguity value, Indicates the float solution of ambiguity; Is the fixed solution of the baseline vector, that is, the receiver coordinates obtained by solution.
[0125] The technical solution of this embodiment makes full use of the advantages of fast geometric image change of low-earth orbit satellites and a large amount of GNSS satellite observation data. The Kalman filtering method of variance component estimation is adopted to appropriately weight the low-earth orbit satellite and GNSS observation data; the ambiguity testing method based on artificial intelligence is used to improve the positioning accuracy of GNSS-low-earth orbit fusion fast precise positioning and the success rate of ambiguity resolution, and the positioning convergence time is short.
[0126] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and the description of the present invention.
Claims
1. A method for global rapid precise positioning with low-orbit navigation enhancement, characterized in that, It includes the following steps: Construct a joint positioning model based on the Beidou satellite navigation system and low Earth orbit satellites; Perform error correction on the joint positioning model to obtain a precise positioning model; Use Kalman filtering combined with variance component estimation for the precise positioning model to determine the parameters to be estimated, where the parameters to be estimated include the receiver position, clock bias, and carrier phase integer ambiguity; Extract the ratio threshold for the carrier phase integer ambiguity through machine learning, and fix the integer ambiguity according to the ratio threshold; Update the baseline result according to the fixed integer ambiguity to obtain the receiver position coordinates.
2. The method for enhancing global rapid precise positioning of low-orbit navigation according to claim 1, wherein When constructing a joint positioning model based on the Beidou satellite navigation system and low Earth orbit satellites, a dual-frequency ionosphere-free model is adopted. The dual-frequency pseudorange and carrier phase observations of the Beidou satellite navigation system and low Earth orbit satellites are respectively composed into ionosphere-free observation equations; the error of the low-order term of the ionosphere delay is removed through the dispersion characteristic of the ionosphere dispersive medium on the electromagnetic wave propagation, and an ionosphere-free model is respectively formed between the pseudorange and carrier phase observations of different frequencies.
3. The method for low-orbit navigation enhanced global rapid precise positioning according to claim 2, wherein For the satellites observed by the receiver, the observation equation expression of the combined observation value of the pseudorange P and the carrier phase L is as follows: In the above formula, P represents the pseudorange from the satellite to the receiver, and L represents the carrier phase; the subscript r represents the receiver, the subscript s represents the satellite, the superscript B represents the Beidou satellite navigation system, and the superscript O represents the low Earth orbit satellite; is the actual distance from the satellite to the receiver, c is the propagation speed of light in vacuum, dt r is the receiver clock error, dt s is the satellite clock error, T represents the tropospheric delay along the slant path, b r,IF is the code pseudorange hardware delay of the receiver, is the code pseudorange hardware delay at the satellite end, and e represents the pseudorange observation noise; λ IF is the carrier wavelength of the ionosphere-free combination, B r,IF is the phase hardware delay at the receiver end, is the phase hardware delay at the satellite end, represents the ambiguity of the ionosphere-free combination, and ε represents the phase observation noise.
4. The low-orbit navigation enhanced global rapid precise positioning method according to claim 1, wherein When performing error correction on the joint positioning model, the correction parameters include orbit error, clock bias, ionospheric delay, tropospheric delay, antenna phase center, relativistic effect, phase wrapping, and loading tide.
5. The method for low-orbit navigation enhanced global rapid precise positioning according to claim 1, wherein When using Kalman filtering combined with variance component estimation for the precise positioning model to determine the parameters to be estimated, the method of determining the weights of the low-orbit satellites and the Beidou system in the precise positioning model by using the a posteriori variance estimation weight determination method includes the following steps: Construct the prediction residual and the variance-covariance matrix, and the system matrix of the carrier phase and the pseudorange; Construct the error equations of the low-orbit satellites and the Beidou system according to the prediction residual and the variance-covariance matrix, and the system matrix of the carrier phase and the pseudorange; Set the weight matrix of the low-orbit satellites and the Beidou system; Adjust the weights of the low-orbit satellites and the Beidou system through variance component estimation of the unit weight variance.
6. The low-orbit navigation enhanced global rapid precise positioning method according to claim 1, characterized in that, Extracting the ratio threshold for the carrier phase integer ambiguity through machine learning includes: Obtain the optimal and sub-optimal integer ambiguity vectors; Calculate the ratio threshold between the optimal and sub-optimal ambiguities; Taking the ratio threshold as the output, and taking the solution success rate, the number of satellites, the frequency points, and the number of epochs as the input, use the ratio threshold model based on the convolutional neural network to respectively extract the features of the input data at different nodes and the features of different time series, fuse the extracted features, and extract the ratio threshold according to the fused features.
7. The method for enhancing global rapid precise positioning of low-orbit navigation according to claim 6, wherein Fixing the integer ambiguity according to the ratio threshold includes: Calculate the optimal n groups of ambiguity solution results; Based on the ratio threshold, judge the reliability of the ambiguity solution according to the following formula: In the above formula, r is the proportional threshold obtained after machine learning, represents the floating-point solution of ambiguity, is the optimal ambiguity value, is the sub-optimal ambiguity value, is the variance-covariance matrix; Take the ambiguity that meets the requirement of solution reliability as the fixed integer ambiguity.
8. The method for low-orbit navigation enhanced global rapid precise positioning according to claim 1, characterized in that, Update the baseline result according to the following formula: In the above formula, is the floating solution baseline vector, is the covariance of the baseline vector and the ambiguity, is the variance-covariance matrix, is the optimal ambiguity value, represents the floating solution of the ambiguity; is the calculated receiver coordinate.
9. An electronic device, characterized in that, It includes: One or more processors; A storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the low-earth orbit navigation enhanced global rapid precise positioning method according to any one of claims 1-8.
10. A computer program product, comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, the steps of the low-earth orbit navigation enhanced global rapid precise positioning method according to any one of claims 1-8 are implemented.