High-precision positioning method of GNSS receiver based on multi-frequency point joint solution

Abstract

The application discloses a GNSS receiver high-precision positioning method based on multi-frequency point joint solution, which comprises the following steps: step one, acquiring frequency point observation data and positioning auxiliary data received by a target GNSS receiver; step two, performing frequency point quality calibration through a frequency point credibility dynamic calibration network; step three, constructing a multi-frequency point joint observation model; step four, performing observation compensation and correction on the multi-frequency point joint observation model through a multi-frequency residual Gram-Schmidt orthogonal correction technology; step five, adopting an improved LAMBDA ambiguity fixing algorithm to perform fixing processing on carrier phase ambiguity parameters; step six, performing receiver positioning solution; and step seven, performing positioning result packaging to obtain high-precision positioning results of the target GNSS receiver. Through the frequency point credibility dynamic calibration network and the improved LAMBDA ambiguity fixing algorithm, the application improves the reliability of GNSS receiver high-precision positioning.

Description

Technical Field

[0001] This invention relates to the field of high-precision positioning technology, and in particular to a high-precision positioning method for GNSS receivers based on multi-frequency joint calculation. Background Technology

[0002] With the continuous application of global satellite navigation systems such as BeiDou, GPS, and Galileo in scenarios such as autonomous driving, UAV mapping, engineering surveying, intelligent agricultural machinery, and precision time synchronization, the joint solution of multi-frequency satellite navigation observation data and high-precision positioning technology for GNSS receivers has attracted widespread attention. Existing GNSS high-precision positioning methods typically rely on single-frequency or multi-frequency pseudorange observations and carrier phase observations to establish observation equations, and combine conventional error correction, the LAMBDA ambiguity fixing algorithm, and least squares solution to obtain the receiver position. However, in practical applications, the following problems commonly exist: The pseudorange, carrier phase, and Doppler observations, as well as the signal-to-noise ratio, corresponding to different satellite navigation frequencies are significantly affected by obstruction, multipath effects, and receiver noise. Existing methods typically use uniform weights or empirical thresholds to process the observation quality of different frequencies, making it difficult to accurately distinguish between high-confidence and anomalous frequencies. This leads to low-quality frequencies easily lowering the overall positioning accuracy when participating in joint calculations. During multi-frequency joint observations, ionospheric delay, tropospheric delay, inter-frequency bias, differences in observation types, and common residuals are interdependent. Traditional error correction methods struggle to effectively decompose and compensate for multi-frequency residuals, resulting in structural errors remaining in the joint observation model. The standard LAMBDA ambiguity fixing algorithm typically performs a uniform integer search for carrier phase ambiguity parameters, failing to fully consider frequency quality differences and the convergence of residuals after error correction. This can easily lead to ambiguity misfixation or unstable fixing success rates, thus affecting the stability and reliability of high-precision GNSS receiver positioning results.

[0003] Therefore, how to provide a high-precision positioning method for GNSS receivers based on multi-frequency joint calculation is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0004] One objective of this invention is to propose a high-precision positioning method for GNSS receivers based on multi-frequency joint calculation. This invention fully utilizes a frequency point reliability dynamic calibration network, multi-frequency residual Gram-Schmidt orthogonal correction technology, and an improved LAMBDA ambiguity fixing algorithm. It details the high-precision positioning process from epoch synchronization of frequency point observation data, frequency point quality calibration, multi-frequency joint observation modeling, observation compensation correction to carrier phase ambiguity fixing and receiver positioning calculation. It has the advantages of accurate frequency point observation quality identification, strong joint observation error correction capability, high reliability of ambiguity fixing, and good stability of positioning results.

[0005] The high-precision positioning method for GNSS receivers based on multi-frequency joint calculation according to embodiments of the present invention includes the following steps: Step 1: Acquire frequency point observation data and positioning auxiliary data received by the target GNSS receiver, and perform epoch synchronization preprocessing on the frequency point observation data to obtain epoch-synchronized multi-frequency observation data; Step 2: Based on epoch-synchronized multi-frequency observation data and positioning auxiliary data, frequency point quality is calibrated through a frequency point reliability dynamic calibration network to obtain quality-calibrated multi-frequency observation data; Step 3: Construct pseudorange observation equations and carrier phase observation equations based on quality-calibrated multi-frequency observation data, and then perform weighted joint calculations using frequency point confidence weights to obtain a multi-frequency joint observation model; Step 4: Use the multi-frequency residual Gram-Schmidt orthogonal correction technique to perform observation compensation correction on the multi-frequency joint observation model to obtain the error-corrected joint observation model; Step 5: Based on the error correction joint observation model, the improved LAMBDA ambiguity fixing algorithm is used to fix the carrier phase ambiguity parameters to obtain the carrier phase ambiguity fixed vector; Step 6: Use the carrier phase ambiguity fixed vector and error correction joint observation model to solve the receiver positioning problem and obtain the high-precision positioning status result of the receiver; Step 7: Encapsulate the high-precision positioning status result of the receiver to obtain the high-precision positioning result of the target GNSS receiver.

[0006] Optionally, the frequency point observation data includes observation epoch, satellite number, satellite navigation frequency point identifier, and pseudorange observation values, carrier phase observation values, Doppler observation values, and signal-to-noise ratio corresponding to multiple satellite navigation frequency points; The positioning assistance data includes satellite ephemeris data, satellite clock difference data, satellite elevation angle, satellite azimuth angle, and receiver clock status data; The epoch synchronization preprocessing includes time alignment, satellite number matching, and frequency point identifier association, specifically: Time alignment of frequency point observation data according to observation epochs is performed to obtain synchronous observation data; Based on the satellite number, the frequency point observation data corresponding to different satellite navigation frequency points belonging to the same satellite are matched to obtain satellite matching observation data; Based on satellite navigation frequency identifiers, pseudorange observations, carrier phase observations, Doppler observations, and signal-to-noise ratios are associated with the corresponding satellite navigation frequencies to generate epoch-synchronized multi-frequency observation data.

[0007] Optionally, step two specifically includes: Based on epoch-synchronous multi-frequency observation data and positioning auxiliary data, the pseudorange observation value, carrier phase observation value, Doppler observation value, signal-to-noise ratio, satellite navigation frequency point identifier, and satellite elevation angle and satellite azimuth angle corresponding to the satellite number are read for each satellite navigation frequency point to obtain the multi-frequency observation quality feature vector of each satellite navigation frequency point. The multi-frequency observation quality feature vector is input into the frequency point confidence dynamic calibration network, which includes a multi-frequency observation feature input layer, a frequency point embedding coding layer, an epoch time series stability coding layer, a frequency point consistency constraint layer, and a confidence weight output layer. In the frequency point embedding coding layer, a corresponding frequency point embedding vector is configured for each satellite navigation frequency point based on the satellite navigation frequency point identifier, and the multi-frequency observation quality feature vector is linearly fused with the corresponding frequency point embedding vector to generate an initial frequency point feature vector; In the epoch-time stability coding layer, the initial frequency point feature vectors of multiple consecutive observation epochs are encoded for epoch-time stability through one-dimensional temporal convolution to generate epoch-time stability feature vectors. In the frequency point consistency constraint layer, the mean deviation distance between the epoch time series stability feature vectors corresponding to different satellite navigation frequencies under the same satellite number is calculated to obtain the frequency point consistency deviation. In the confidence weight output layer, the epochal time series stability feature vector is linearly mapped to obtain the initial confidence score; The initial confidence score and the consistency deviation between frequency points are weighted and corrected, and then normalized using the Softmax function to obtain the frequency confidence weight of each satellite navigation frequency point. Frequency point confidence weights are added to the epoch-synchronized multi-frequency observation data according to satellite number, observation epoch, and satellite navigation frequency point identifier to obtain quality calibration multi-frequency observation data.

[0008] Optionally, step three specifically includes: Based on the quality calibration multi-frequency observation data, read the pseudorange observation value, carrier phase observation value and frequency point confidence weight corresponding to each satellite navigation frequency point; Satellite ephemeris data and satellite clock bias data are read based on positioning-assisted data; Based on the satellite navigation frequency point identifier, determine the carrier wavelength corresponding to each satellite navigation frequency point; Based on satellite ephemeris data, determine the location of each satellite corresponding to its number; Construct a set of parameters to be estimated, which includes the target GNSS receiver's receiver position, receiver clock error, ionospheric delay parameter, tropospheric delay parameter, pseudo-range-frequency offset parameter, carrier phase-frequency offset parameter, and carrier phase ambiguity parameter. Calculate the spatial distance between the estimated receiver position of the target GNSS receiver and the positions of each satellite to obtain the geometric distance term; Based on pseudorange observations, geometric distance terms, receiver clock bias, satellite clock bias data, ionospheric delay parameters, tropospheric delay parameters, and pseudorange frequency offset parameters, pseudorange observation equations are established between pseudorange observations corresponding to each satellite navigation frequency point and the set of parameters to be estimated. Based on carrier phase observations, geometric distance terms, receiver clock bias, satellite clock bias data, ionospheric delay parameters, tropospheric delay parameters, carrier wavelength, carrier phase ambiguity parameters, and carrier phase inter-frequency deviation parameters, a carrier phase observation equation is established between the carrier phase observations corresponding to each satellite navigation frequency point and the set of parameters to be estimated. The frequency point confidence weights are generated into an observation weight matrix according to the observation epoch, satellite number, and satellite navigation frequency point identifier; Based on the pseudorange observation equation, generate pseudorange observation model values ​​for each satellite navigation frequency; based on the carrier phase observation equation, generate carrier phase observation model values ​​for each satellite navigation frequency. The pseudorange observation model values ​​and carrier phase observation model values ​​are vectorized according to the observation epoch, satellite number and satellite navigation frequency point identifier to generate a joint observation vector; Based on the observation weight matrix, the joint observation vectors are weighted and jointly organized to obtain a multi-frequency joint observation model.

[0009] Optionally, the multi-frequency residual Gram-Schmidt orthogonal correction technique specifically includes: Based on the multi-frequency joint observation model, the pseudorange observation model value and carrier phase observation model value corresponding to each satellite navigation frequency point are read; The difference between the pseudorange observation value and the pseudorange observation model value is calculated to obtain the pseudorange observation residual; the difference between the carrier phase observation value and the carrier phase observation model value is calculated to obtain the carrier phase observation residual. According to the observation epoch, satellite number, satellite navigation frequency point identifier, and observation type, the pseudorange observation residual and carrier phase observation residual are stacked into a multi-frequency residual vector; the observation type includes pseudorange observation type and carrier phase observation type; Based on the satellite number, satellite navigation frequency point identifier, carrier wavelength and observation type of each residual element in the multi-frequency residual vector, a common residual candidate basis matrix, a frequency dispersion residual candidate basis matrix, an inter-frequency deviation residual candidate basis matrix and an observation type difference residual candidate basis matrix are constructed. The observation weight matrix is ​​used as the Gram-Schmidt orthogonal weighted inner product metric matrix. The candidate basis matrix groups of residuals are subjected to hierarchical weighted orthogonal processing in the order of common residuals, frequency dispersion residuals, inter-frequency deviation residuals, and observation type difference residuals to obtain the common residual orthogonal basis matrix, frequency dispersion residual orthogonal basis matrix, inter-frequency deviation residual orthogonal basis matrix, and observation type difference residual orthogonal basis matrix. The multi-frequency residual vectors are projected onto the orthogonal subspaces corresponding to the common residual orthogonal basis matrix, the frequency dispersion residual orthogonal basis matrix, the inter-frequency deviation residual orthogonal basis matrix, and the observation type difference residual orthogonal basis matrix, respectively, to obtain the common residual components, the frequency dispersion residual components, the inter-frequency deviation residual components, and the observation type difference residual components; The average value of the frequency point confidence weight associated with each residual element in the common residual component, frequency dispersion residual component, inter-frequency bias residual component, and observation type difference residual component is calculated, and the average value is normalized to obtain the confidence scaling factor. The common residual component, frequency dispersion residual component, inter-frequency bias residual component and observation type difference residual component are scaled and corrected according to the confidence scaling factor, and the scaled and corrected residual components are weighted and superimposed to obtain the error correction component. Based on the error correction component, the joint observation vector in the multi-frequency joint observation model is corrected by observation compensation to obtain the error correction joint observation model.

[0010] Optionally, the improved LAMBDA ambiguity fixing algorithm specifically includes: Based on the error correction joint observation model, the least squares estimation method is used to solve the parameter set to be estimated, and the ambiguity floating-point solution and ambiguity covariance matrix corresponding to the carrier phase ambiguity parameter are extracted from the parameter solution results. The frequency point confidence weights are associated with the corresponding carrier phase ambiguity parameters according to the observation epoch, satellite number and satellite navigation frequency point identifier, and the carrier phase ambiguity parameters are divided into a high confidence ambiguity parameter group and a ambiguity parameter group to be verified based on the frequency point confidence weights. Perform decorrelation transformation on the floating-point fuzzy solution and fuzzy covariance matrix corresponding to the high-confidence fuzzy parameter set, and perform LAMBDA integer search in the integer search space after decorrelation transformation to obtain high-confidence fuzzy candidate solution and high-confidence suboptimal fuzzy candidate solution; Using the high-confidence fuzzy candidate solution as an integer constraint, the fuzzy floating-point solution and fuzzy covariance matrix corresponding to the fuzzy parameter group to be verified are updated with conditional constraints to obtain the conditional floating-point solution and conditional covariance matrix of the fuzzy parameter group to be verified. Based on the conditional floating-point solution and conditional covariance matrix of the ambiguity to be verified, a second-level LAMBDA integer search is performed on the parameter set of the ambiguity to be verified to obtain the candidate solution and the candidate suboptimal solution of the ambiguity to be verified. Based on the observation epoch, satellite number, and satellite navigation frequency identifier, the high-confidence ambiguity candidate solution and the ambiguity candidate solution to be verified are combined into the optimal ambiguity candidate solution, and the high-confidence suboptimal ambiguity candidate solution and the suboptimal ambiguity candidate solution to be verified are combined into the suboptimal ambiguity candidate solution. The objective function values ​​of the LAMBDA integer search corresponding to the suboptimal ambiguity candidate solution and the optimal ambiguity candidate solution are calculated respectively, and the ratio test value is obtained by ratio operation; Substitute the optimal ambiguity candidate solution and the suboptimal ambiguity candidate solution into the error correction joint observation model to obtain the optimal joint observation model vector and the suboptimal joint observation model vector, respectively. Calculate the vector differences between the optimal joint observation model vector and the suboptimal joint observation model vector and the joint observation vector after observation compensation correction, and calculate the L2 norm square of the vector differences to obtain the optimal observation residual evaluation value and the suboptimal observation residual evaluation value. Calculate the difference between the suboptimal observation residual evaluation value and the optimal observation residual evaluation value to obtain the residual convergence interval; If the ratio test value is greater than or equal to the dynamic ratio test threshold, and the residual convergence interval is greater than or equal to the residual convergence interval threshold, then the corresponding optimal ambiguity candidate solution is determined as the carrier phase ambiguity fixed vector.

[0011] Optionally, step six specifically includes: The carrier phase ambiguity fixed vector is replaced with the carrier phase ambiguity parameter in the error correction joint observation model according to the observation epoch, satellite number and satellite navigation frequency point identifier to obtain the fixed correction observation model. Based on the fixed correction observation model, the receiver position, receiver clock error, ionospheric delay parameter, tropospheric delay parameter, pseudo-range-frequency offset parameter and carrier phase-frequency offset parameter in the parameter set to be estimated are jointly solved to obtain the joint solution results of the parameters; The three-dimensional position coordinates and receiver clock error of the target GNSS receiver are extracted from the joint parameter solution results, and the positioning residual is calculated based on the fixed correction observation model. The three-dimensional position coordinates, receiver clock error, and positioning residual are used as the high-precision positioning status results of the receiver.

[0012] Optionally, step seven specifically includes: The root mean square value of the positioning residual is used as the evaluation value of positioning accuracy. Read the ambiguity fixing status corresponding to the carrier phase ambiguity fixing vector; the ambiguity fixing status includes carrier phase ambiguity fixing successful and carrier phase ambiguity fixing unsuccessful. Based on the mean and standard deviation of the absolute values ​​of each positioning residual element corresponding to the observation epoch, a positioning accuracy threshold is generated; A positioning status identifier is generated based on the positioning accuracy evaluation value and the ambiguity fixation status: if the positioning accuracy evaluation value is less than or equal to the positioning accuracy threshold and the carrier phase ambiguity is successfully fixed, the positioning status identifier is 1; if the positioning accuracy evaluation value is greater than the positioning accuracy threshold, or the carrier phase ambiguity is not successfully fixed, the positioning status identifier is 0. The three-dimensional position coordinates, receiver clock error, positioning residual, positioning accuracy evaluation value, ambiguity fixed state, and positioning status identifier are associated and encapsulated according to the observation epoch to obtain the high-precision positioning result of the target GNSS receiver.

[0013] The beneficial effects of this invention are: This invention employs a frequency point reliability dynamic calibration network to jointly characterize the quality of pseudorange observations, carrier phase observations, Doppler observations, signal-to-noise ratio, satellite elevation angle, and satellite azimuth angle, generating frequency point reliability weights. This enables a quantitative expression of the observation reliability of each satellite navigation frequency point at different observation epochs. Furthermore, in the multi-frequency joint observation model, low-quality frequencies are suppressed, high-reliability frequencies are enhanced, and the impact of multipath interference, obstruction attenuation, and inter-frequency observation fluctuations on subsequent positioning calculations is reduced, improving the effectiveness and stability of multi-frequency observation data in joint calculations. During the multi-frequency joint observation modeling stage, the pseudorange observation equation, carrier phase observation equation, observation weight matrix, and set of parameters to be estimated are weighted and jointly organized, enhancing the collaborative utilization capability of multi-frequency observation data in positioning calculations. In the error correction stage, the pseudorange observation residual and carrier phase observation residual are decomposed into common residual components, frequency dispersion residual components, inter-frequency deviation residual components, and observation type difference residual components using the multi-frequency residual Gram-Schmidt orthogonal correction technique. This is combined with a confidence scaling factor for observation compensation correction, reducing the impact of multi-frequency residual coupling on the positioning model. In the ambiguity fixing stage, an improved LAMBDA ambiguity fixing algorithm is used to perform hierarchical search, condition constraint update, and residual convergence interval determination for the carrier phase ambiguity parameters, improving the reliability of the carrier phase ambiguity fixing vector. Finally, the receiver positioning solution is calculated using the fixed correction observation model, generating high-precision positioning results. This improves the positioning accuracy, positioning stability, and ambiguity fixing success rate of the target GNSS receiver in complex observation environments. Attached Figure Description

[0014] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram of the high-precision positioning method for GNSS receivers based on multi-frequency joint calculation proposed in this invention; Figure 2 This is a flowchart of the network structure for dynamic calibration of frequency point reliability in the high-precision positioning method for GNSS receivers based on multi-frequency point joint calculation proposed in this invention. Figure 3 This is a flowchart of the improved LAMBDA ambiguity fixing algorithm in the high-precision positioning method for GNSS receivers based on multi-frequency joint calculation proposed in this invention. Detailed Implementation

[0015] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0016] refer to Figures 1-3 A high-precision positioning method for GNSS receivers based on multi-frequency joint calculation includes the following steps: Step 1: Acquire frequency point observation data and positioning auxiliary data received by the target GNSS receiver, and perform epoch synchronization preprocessing on the frequency point observation data to obtain epoch-synchronized multi-frequency observation data; Step 2: Based on epoch-synchronized multi-frequency observation data and positioning auxiliary data, frequency point quality is calibrated through a frequency point reliability dynamic calibration network to obtain quality-calibrated multi-frequency observation data; Step 3: Construct pseudorange observation equations and carrier phase observation equations based on quality-calibrated multi-frequency observation data, and then perform weighted joint calculations using frequency point confidence weights to obtain a multi-frequency joint observation model; Step 4: Use the multi-frequency residual Gram-Schmidt orthogonal correction technique to perform observation compensation correction on the multi-frequency joint observation model to obtain the error-corrected joint observation model; Step 5: Based on the error correction joint observation model, the improved LAMBDA ambiguity fixing algorithm is used to fix the carrier phase ambiguity parameters to obtain the carrier phase ambiguity fixed vector; Step 6: Use the carrier phase ambiguity fixed vector and error correction joint observation model to solve the receiver positioning problem and obtain the high-precision positioning status result of the receiver; Step 7: Encapsulate the high-precision positioning status result of the receiver to obtain the high-precision positioning result of the target GNSS receiver.

[0017] In this embodiment, the frequency point observation data includes the observation epoch, satellite number, satellite navigation frequency point identifier, and pseudorange observation values, carrier phase observation values, Doppler observation values, and signal-to-noise ratio corresponding to multiple satellite navigation frequency points; Positioning assistance data includes satellite ephemeris data, satellite clock bias data, satellite elevation angle, satellite azimuth angle, and receiver clock status data; Erato synchronization preprocessing includes time alignment, satellite number matching, and frequency point identifier association, specifically: Time alignment of frequency point observation data according to observation epochs is performed to obtain synchronous observation data; Based on the satellite number, the frequency point observation data corresponding to different satellite navigation frequency points belonging to the same satellite are matched to obtain satellite matching observation data; Based on satellite navigation frequency identifiers, pseudorange observations, carrier phase observations, Doppler observations, and signal-to-noise ratios are associated with the corresponding satellite navigation frequencies to generate epoch-synchronized multi-frequency observation data.

[0018] In this embodiment, step two specifically includes: Based on epoch-synchronous multi-frequency observation data and positioning auxiliary data, the pseudorange observation value, carrier phase observation value, Doppler observation value, signal-to-noise ratio, satellite navigation frequency point identifier, and satellite elevation angle and satellite azimuth angle corresponding to the satellite number are read for each satellite navigation frequency point to obtain the multi-frequency observation quality feature vector of each satellite navigation frequency point. The multi-frequency observation quality feature vector is input into the frequency point confidence dynamic calibration network; the frequency point confidence dynamic calibration network includes a multi-frequency observation feature input layer, a frequency point embedding coding layer, an epoch time series stability coding layer, a frequency point consistency constraint layer, and a confidence weight output layer; In the frequency point embedding coding layer, a corresponding frequency point embedding vector is configured for each satellite navigation frequency point based on the satellite navigation frequency point identifier, and the multi-frequency observation quality feature vector is linearly fused with the corresponding frequency point embedding vector to generate an initial frequency point feature vector; In the epoch-time stability coding layer, the initial frequency point feature vectors of multiple consecutive observation epochs are encoded for epoch-time stability through one-dimensional temporal convolution to generate epoch-time stability feature vectors. In the frequency point consistency constraint layer, the mean deviation distance between the epoch time series stability feature vectors corresponding to different satellite navigation frequencies under the same satellite number is calculated to obtain the frequency point consistency deviation. In the confidence weight output layer, the epochal time series stability feature vector is linearly mapped to obtain the initial confidence score; The initial confidence score and the consistency deviation between frequency points are weighted and corrected, and then normalized using the Softmax function to obtain the frequency confidence weight of each satellite navigation frequency point. Frequency point confidence weights are added to the epoch-synchronized multi-frequency observation data according to satellite number, observation epoch, and satellite navigation frequency point identifier to obtain quality calibration multi-frequency observation data.

[0019] In this embodiment, step three specifically includes: Based on the quality calibration multi-frequency observation data, read the pseudorange observation value, carrier phase observation value and frequency point confidence weight corresponding to each satellite navigation frequency point; Satellite ephemeris data and satellite clock bias data are read based on positioning-assisted data; Based on the satellite navigation frequency point identifier, determine the carrier wavelength corresponding to each satellite navigation frequency point; Based on satellite ephemeris data, determine the location of each satellite corresponding to its number; Construct a set of parameters to be estimated, which includes the target GNSS receiver's receiver position, receiver clock error, ionospheric delay parameters, tropospheric delay parameters, pseudo-range-frequency offset parameters, carrier phase-frequency offset parameters, and carrier phase ambiguity parameters. Calculate the spatial distance between the estimated receiver position of the target GNSS receiver and the positions of each satellite to obtain the geometric distance term; Based on pseudorange observations, geometric distance terms, receiver clock bias, satellite clock bias data, ionospheric delay parameters, tropospheric delay parameters, and pseudorange frequency offset parameters, pseudorange observation equations are established between pseudorange observations corresponding to each satellite navigation frequency point and the set of parameters to be estimated. Based on carrier phase observations, geometric distance terms, receiver clock bias, satellite clock bias data, ionospheric delay parameters, tropospheric delay parameters, carrier wavelength, carrier phase ambiguity parameters, and carrier phase inter-frequency deviation parameters, a carrier phase observation equation is established between the carrier phase observations corresponding to each satellite navigation frequency point and the set of parameters to be estimated. The frequency point confidence weights are generated into an observation weight matrix according to the observation epoch, satellite number, and satellite navigation frequency point identifier; Based on the pseudorange observation equation, generate pseudorange observation model values ​​for each satellite navigation frequency; based on the carrier phase observation equation, generate carrier phase observation model values ​​for each satellite navigation frequency. The pseudorange observation model values ​​and carrier phase observation model values ​​are vectorized according to the observation epoch, satellite number and satellite navigation frequency point identifier to generate a joint observation vector; Based on the observation weight matrix, the joint observation vectors are weighted and jointly organized to obtain a multi-frequency joint observation model.

[0020] In this embodiment, the multi-frequency residual Gram-Schmidt orthogonal correction technique specifically includes: Based on the multi-frequency joint observation model, the pseudorange observation model value and carrier phase observation model value corresponding to each satellite navigation frequency point are read; The difference between the pseudorange observation value and the pseudorange observation model value is calculated to obtain the pseudorange observation residual; the difference between the carrier phase observation value and the carrier phase observation model value is calculated to obtain the carrier phase observation residual. According to the observation epoch, satellite number, satellite navigation frequency point identifier, and observation type, the pseudorange observation residual and carrier phase observation residual are stacked into a multi-frequency residual vector; the observation type includes pseudorange observation type and carrier phase observation type; Based on the satellite number, satellite navigation frequency identifier, carrier wavelength, and observation type of each residual element in the multi-frequency residual vector, a common residual candidate basis matrix, a frequency dispersion residual candidate basis matrix, an inter-frequency deviation residual candidate basis matrix, and an observation type difference residual candidate basis matrix are constructed. Specifically: residual elements belonging to the same satellite number are assigned a value of 1 in the same column of the common residual candidate basis matrix to generate the common residual candidate basis matrix; the carrier wavelength corresponding to each residual element is converted into a frequency dispersion coefficient, and the frequency dispersion coefficient is written into the row of the corresponding residual element in the frequency dispersion residual candidate basis matrix to generate the frequency dispersion residual candidate basis matrix; residual elements belonging to the same satellite navigation frequency identifier are assigned a value of 1 in the same column of the inter-frequency deviation residual candidate basis matrix to generate the inter-frequency deviation residual candidate basis matrix; residual elements corresponding to pseudorange observation types are assigned a value of 1 in the observation type difference residual candidate basis matrix, and residual elements corresponding to carrier phase observation types are assigned a value of -1 in the observation type difference residual candidate basis matrix to generate the observation type difference residual candidate basis matrix. Among them, the common residual candidate basis matrix is ​​used to characterize the residual changes that occur together with different satellite navigation frequencies and different observation types under the same satellite number; the frequency dispersion residual candidate basis matrix is ​​used to characterize the residual changes related to the carrier wavelength; the inter-frequency deviation residual candidate basis matrix is ​​used to characterize the residual offset between different satellite navigation frequencies; and the observation type difference residual candidate basis matrix is ​​used to characterize the residual differences between pseudorange observation type and carrier phase observation type. The observation weight matrix is ​​used as the Gram-Schmidt orthogonal weighted inner product metric matrix. The candidate basis matrix groups of residuals are subjected to hierarchical weighted orthogonal processing in the order of common residuals, frequency dispersion residuals, inter-frequency deviation residuals, and observation type difference residuals to obtain the common residual orthogonal basis matrix, frequency dispersion residual orthogonal basis matrix, inter-frequency deviation residual orthogonal basis matrix, and observation type difference residual orthogonal basis matrix. Among them, the hierarchical weighted orthogonal processing refers to: for the current residual candidate basis matrix, calculating the weighted projection components of the current residual candidate basis matrix on the generated previous residual orthogonal basis matrix according to the observation weight matrix, subtracting the weighted projection components from the current residual candidate basis matrix to obtain the current deprojected candidate basis matrix, and then performing weighted normalization processing on the current deprojected candidate basis matrix to obtain the current residual orthogonal basis matrix; The multi-frequency residual vectors are projected onto the orthogonal subspaces corresponding to the common residual orthogonal basis matrix, the frequency dispersion residual orthogonal basis matrix, the inter-frequency deviation residual orthogonal basis matrix, and the observation type difference residual orthogonal basis matrix, respectively, to obtain the common residual components, the frequency dispersion residual components, the inter-frequency deviation residual components, and the observation type difference residual components; The average value of the frequency point confidence weight associated with each residual element in the common residual component, frequency dispersion residual component, inter-frequency bias residual component, and observation type difference residual component is calculated, and the average value is normalized to obtain the confidence scaling factor. The common residual component, frequency dispersion residual component, inter-frequency bias residual component and observation type difference residual component are scaled and corrected according to the confidence scaling factor, and the scaled and corrected residual components are weighted and superimposed to obtain the error correction component. Based on the error correction component, the joint observation vector in the multi-frequency joint observation model is corrected by observation compensation to obtain the error correction joint observation model.

[0021] In this embodiment, the improved LAMBDA ambiguity fixing algorithm specifically includes: Based on the error correction joint observation model, the least squares estimation method is used to solve the parameter set to be estimated, and the ambiguity floating-point solution and ambiguity covariance matrix corresponding to the carrier phase ambiguity parameter are extracted from the parameter solution results. The frequency point confidence weights are associated with the corresponding carrier phase ambiguity parameters according to the observation epoch, satellite number, and satellite navigation frequency point identifier. Based on the frequency point confidence weights, the carrier phase ambiguity parameters are divided into a high-confidence ambiguity parameter group and a ambiguity parameter group to be reviewed. Among them, the carrier phase ambiguity parameters with frequency point confidence weights greater than or equal to the confidence stratification threshold are assigned to the high-confidence ambiguity parameter group, and the carrier phase ambiguity parameters with frequency point confidence weights less than the confidence stratification threshold are assigned to the ambiguity parameter group to be reviewed. Perform decorrelation transformation on the floating-point fuzzy solution and fuzzy covariance matrix corresponding to the high-confidence fuzzy parameter set, and perform LAMBDA integer search in the integer search space after decorrelation transformation to obtain high-confidence fuzzy candidate solution and high-confidence suboptimal fuzzy candidate solution; Using the high-confidence fuzzy candidate solution as an integer constraint, the fuzzy floating-point solution and fuzzy covariance matrix corresponding to the fuzzy parameter group to be verified are updated with conditional constraints to obtain the conditional floating-point solution and conditional covariance matrix of the fuzzy parameter group to be verified. Based on the conditional floating-point solution and conditional covariance matrix of the ambiguity to be verified, a second-level LAMBDA integer search is performed on the parameter set of the ambiguity to be verified to obtain the candidate solution and the candidate suboptimal solution of the ambiguity to be verified. Based on the observation epoch, satellite number, and satellite navigation frequency identifier, the high-confidence ambiguity candidate solution and the ambiguity candidate solution to be verified are combined into the optimal ambiguity candidate solution, and the high-confidence suboptimal ambiguity candidate solution and the suboptimal ambiguity candidate solution to be verified are combined into the suboptimal ambiguity candidate solution. The objective function values ​​of the LAMBDA integer search corresponding to the suboptimal ambiguity candidate solution and the optimal ambiguity candidate solution are calculated respectively, and the ratio test value is obtained by ratio operation; Substitute the optimal ambiguity candidate solution and the suboptimal ambiguity candidate solution into the error correction joint observation model to obtain the optimal joint observation model vector and the suboptimal joint observation model vector, respectively. Calculate the vector differences between the optimal joint observation model vector and the suboptimal joint observation model vector and the joint observation vector after observation compensation correction, and calculate the L2 norm square of the vector differences to obtain the optimal observation residual evaluation value and the suboptimal observation residual evaluation value. Calculate the difference between the suboptimal observation residual evaluation value and the optimal observation residual evaluation value to obtain the residual convergence interval; If the ratio test value is greater than or equal to the dynamic ratio test threshold, and the residual convergence interval is greater than or equal to the residual convergence interval threshold, then the corresponding optimal ambiguity candidate solution is determined as the carrier phase ambiguity fixed vector.

[0022] In this invention, the mean and standard deviation of the objective function values ​​corresponding to each ambiguity candidate solution obtained by LAMBDA integer search are used to generate a dynamic ratio test threshold; the difference between the mean and minimum values ​​of the observation residual evaluation values ​​obtained by substituting each integer ambiguity candidate solution into the error correction joint observation model is used as the residual convergence interval threshold; the carrier phase ambiguity fixed vector is an integer ambiguity fixed value arranged according to the observation epoch, satellite number and satellite navigation frequency point identifier.

[0023] In this invention, the improved LAMBDA ambiguity fixing algorithm has the same basic structure as the standard LAMBDA ambiguity fixing algorithm. Their common process includes: obtaining the ambiguity floating-point solution and ambiguity covariance matrix corresponding to the carrier phase ambiguity parameters using the least squares estimation method; performing decorrelation transformation on the ambiguity floating-point solution and ambiguity covariance matrix to reduce the correlation between different carrier phase ambiguity parameters; performing LAMBDA integer search in the decorrelation-transformed integer search space to obtain integer ambiguity candidate solutions; judging the reliability of the integer ambiguity candidate solutions using the objective function value and ratio test value; and outputting the carrier phase ambiguity fixing result in integer form when the fixing conditions are met.

[0024] The improved LAMBDA ambiguity fixing algorithm does not perform a one-time unified integer search on all carrier phase ambiguity parameters. Instead, it introduces a hierarchical constraint based on frequency point confidence weights and a two-level LAMBDA integer search structure on the basis of the standard LAMBDA algorithm. Specifically, the improved LAMBDA ambiguity fixing algorithm first associates the frequency point confidence weights with the corresponding carrier phase ambiguity parameters according to the observation epoch, satellite number, and satellite navigation frequency identifier. Then, based on the frequency point confidence weights, it divides the carrier phase ambiguity parameters into a high-confidence ambiguity parameter group and a ambiguity parameter group to be verified. Subsequently, it prioritizes performing decorrelation transformation and LAMBDA integer search on the high-confidence ambiguity parameter group to obtain high-confidence ambiguity candidate solutions. These high-confidence ambiguity candidate solutions are then used as integer constraints to update the ambiguity floating-point solutions and ambiguity covariance matrix of the ambiguity parameter group to be verified, and finally, it performs a two-level LAMBDA integer search. In addition, the improved LAMBDA fuzziness fixing algorithm substitutes the optimal fuzziness candidate solution and the suboptimal fuzziness candidate solution into the error correction joint observation model, and calculates the optimal observation residual evaluation value, the suboptimal observation residual evaluation value and the residual convergence interval by combining the joint observation vector after observation compensation correction. The ratio test value and the residual convergence interval are used together as the fixing criteria.

[0025] By first fixing a high-confidence ambiguity parameter set and then constraining the secondary search of the ambiguity parameter set to be verified with high-confidence ambiguity candidate solutions, the subsequent integer search space can be narrowed, reducing the interference of low-quality satellite navigation frequencies on the overall ambiguity fixing results and improving the stability of the integer ambiguity search and the reliability of candidate solutions. Simultaneously, by substituting candidate solutions into the error-corrected joint observation model and calculating the residual convergence interval, the residual convergence advantage of the optimal ambiguity candidate solution over the second-best ambiguity candidate solution in the actual observation model can be further determined, reducing the erroneous fixing problem caused by relying solely on the standard ratio test. Therefore, this application can improve the success rate and reliability of carrier phase ambiguity fixing, thereby enhancing the stability and accuracy of the high-precision positioning results of the target GNSS receiver.

[0026] In this embodiment, step six specifically includes: The carrier phase ambiguity fixed vector is replaced with the carrier phase ambiguity parameter in the error correction joint observation model according to the observation epoch, satellite number and satellite navigation frequency point identifier to obtain the fixed correction observation model. Based on the fixed correction observation model, the receiver position, receiver clock error, ionospheric delay parameter, tropospheric delay parameter, pseudo-range-frequency offset parameter and carrier phase-frequency offset parameter in the parameter set to be estimated are jointly solved to obtain the joint solution results of the parameters; The three-dimensional position coordinates and receiver clock error of the target GNSS receiver are extracted from the joint parameter solution results, and the positioning residual is calculated based on the fixed correction observation model. The three-dimensional position coordinates, receiver clock error, and positioning residual are used as the high-precision positioning status results of the receiver.

[0027] In this embodiment, step seven specifically includes: The root mean square value of the positioning residual is used as the evaluation value of positioning accuracy. Read the ambiguity fixation status corresponding to the carrier phase ambiguity fixation vector; the ambiguity fixation status includes carrier phase ambiguity fixation successful and carrier phase ambiguity fixation unsuccessful. Based on the mean and standard deviation of the absolute values ​​of each positioning residual element corresponding to the observation epoch, a positioning accuracy threshold is generated; A positioning status identifier is generated based on the positioning accuracy evaluation value and the ambiguity fixation status: if the positioning accuracy evaluation value is less than or equal to the positioning accuracy threshold and the carrier phase ambiguity is successfully fixed, the positioning status identifier is 1; if the positioning accuracy evaluation value is greater than the positioning accuracy threshold, or the carrier phase ambiguity is not successfully fixed, the positioning status identifier is 0. The three-dimensional position coordinates, receiver clock error, positioning residual, positioning accuracy evaluation value, ambiguity fixed state, and positioning status identifier are associated and encapsulated according to the observation epoch to obtain the high-precision positioning result of the target GNSS receiver.

[0028] Example 1: To verify the feasibility of this invention in practice, the method was applied to a high-precision positioning scenario using a GNSS receiver on a vehicle-mounted section of an elevated expressway in a city. This expressway has continuous sound barriers on both sides, some tall buildings obstructing the view, and reflective surfaces under the elevated bridge. During vehicle operation, issues such as fluctuations in the number of visible satellites, enhanced multipath reflections, and inconsistent signal-to-noise ratios at different satellite navigation frequencies arise. While traditional GNSS receivers can receive observation data from multiple frequencies such as GPS L1 / L2, BeiDou B1 / B2, and Galileo E1 / E5 in this scenario, carrier phase observations at some satellite navigation frequencies are prone to jitter in areas such as bridge ramps, road curves, and building obstructions. This also significantly increases pseudorange observation residuals, leading to unstable carrier phase ambiguity and jumps in positioning results.

[0029] In this invention, a target GNSS receiver is mounted on the top of a test vehicle, and frequency point observation data and positioning assistance data are continuously collected during vehicle operation. The frequency point observation data includes observation epochs, satellite numbers, satellite navigation frequency identifiers, and corresponding pseudorange, carrier phase, Doppler, and signal-to-noise ratio (SNR) values ​​for each satellite navigation frequency. The positioning assistance data includes satellite ephemeris data, satellite clock bias data, satellite elevation angle, satellite azimuth angle, and receiver clock status data. The system first performs time alignment of the frequency point observation data according to the observation epochs and completes matching according to the satellite numbers and satellite navigation frequency identifiers to generate epoch-synchronized multi-frequency observation data. Subsequently, the epoch-synchronized multi-frequency observation data and positioning assistance data are input into a frequency point reliability dynamic calibration network. Through frequency point embedding coding, epoch timing stability coding, and frequency point consistency constraint processing, the network outputs the frequency point reliability weights corresponding to each satellite navigation frequency, thereby weakening frequencies with low SNR, strong reflections, or significant timing fluctuations.

[0030] After quality calibration, pseudorange and carrier phase observation equations are constructed based on the multi-frequency observation data from the quality calibration. An observation weight matrix is ​​generated using frequency point confidence weights, and the joint observation vectors are weighted and jointly organized to obtain a multi-frequency joint observation model. To address the common residuals, frequency dispersion residuals, inter-frequency bias residuals, and observation type difference residuals caused by overpasses and building obstructions, a multi-frequency residual Gram-Schmidt orthogonal correction technique is further employed for hierarchical weighted orthogonal processing. Observation compensation corrections are then applied to the joint observation vectors to obtain an error-corrected joint observation model. Subsequently, an improved LAMBDA ambiguity fixing algorithm is used to fix the carrier phase ambiguity parameters, obtaining a fixed carrier phase ambiguity vector. This fixed vector replaces the carrier phase ambiguity parameters in the error-corrected joint observation model, resulting in a fixed-correction observation model. Finally, the three-dimensional position coordinates of the target GNSS receiver, receiver clock error, and positioning residuals are calculated.

[0031] To further verify the practical effectiveness of this invention, comparative experiments were conducted with the single-frequency pseudorange positioning scheme, the standard multi-frequency weighted positioning scheme, and the standard LAMBDA multi-frequency positioning scheme. The single-frequency pseudorange positioning scheme uses only pseudorange observations from a single frequency point for positioning calculation; the standard multi-frequency weighted positioning scheme uses multi-frequency observations and fixed weights for joint calculation; and the standard LAMBDA multi-frequency positioning scheme uses the standard LAMBDA algorithm for ambiguity fixation based on multi-frequency joint calculation. The experimental results are shown in Table 1.

[0032] Table 1. Comparison of positioning performance of different GNSS positioning schemes in urban expressway elevated road scenarios.

[0033] As shown in Table 1, the method of this invention achieves significant improvements in multiple comparative indicators. Specifically, the root mean square error (RMS) for planar positioning is reduced to 0.072 m, and the RMS error for elevation positioning is reduced to 0.128 m. This indicates that the synergistic processing of the frequency point reliability dynamic calibration network, multi-frequency residual Gram-Schmidt orthogonal correction technology, and the improved LAMBDA ambiguity fixing algorithm effectively improves the position calculation accuracy of the target GNSS receiver. Compared to the standard LAMBDA multi-frequency positioning scheme, the ambiguity fixing success rate of the method of this invention increases from 89.2% to 96.8%, and the average first-time fixing time is shortened from 11.4 s to 6.3 s, demonstrating that the improved LAMBDA ambiguity fixing algorithm can improve the reliability and convergence speed of the carrier phase ambiguity fixing vector.

[0034] Furthermore, the method of this invention involves only two positioning jumps, reduces the average positioning residual to 0.071m, and achieves a continuous effective positioning rate of 99.2%. This demonstrates that the error-corrected joint observation model can reduce the impact of multi-frequency residual coupling and false participation of low-quality frequencies on the positioning results. The false participation rate of low-quality frequencies is reduced from 28.5% in the standard multi-frequency weighted positioning scheme to 7.4%, further proving that this invention can accurately identify and suppress low-quality satellite navigation frequencies in complex observation environments, improving the stability of multi-frequency joint solution and the reliability of high-precision positioning results.

[0035] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A high-precision positioning method for GNSS receivers based on multi-frequency joint calculation, characterized in that, Includes the following steps: Step 1: Acquire frequency point observation data and positioning auxiliary data received by the target GNSS receiver, and perform epoch synchronization preprocessing on the frequency point observation data to obtain epoch-synchronized multi-frequency observation data; Step 2: Based on epoch-synchronized multi-frequency observation data and positioning auxiliary data, frequency point quality is calibrated through a frequency point reliability dynamic calibration network to obtain quality-calibrated multi-frequency observation data; Step 3: Construct pseudorange observation equations and carrier phase observation equations based on quality-calibrated multi-frequency observation data, and then perform weighted joint calculations using frequency point confidence weights to obtain a multi-frequency joint observation model; Step 4: Use the multi-frequency residual Gram-Schmidt orthogonal correction technique to perform observation compensation correction on the multi-frequency joint observation model to obtain the error-corrected joint observation model; Step 5: Based on the error correction joint observation model, the improved LAMBDA ambiguity fixing algorithm is used to fix the carrier phase ambiguity parameters to obtain the carrier phase ambiguity fixed vector; Step 6: Use the carrier phase ambiguity fixed vector and error correction joint observation model to solve the receiver positioning problem and obtain the high-precision positioning status result of the receiver; Step 7: Encapsulate the high-precision positioning status result of the receiver to obtain the high-precision positioning result of the target GNSS receiver.

2. The high-precision positioning method for GNSS receivers based on multi-frequency joint calculation according to claim 1, characterized in that, The frequency point observation data includes observation epoch, satellite number, satellite navigation frequency point identifier, and pseudorange observation values, carrier phase observation values, Doppler observation values, and signal-to-noise ratio corresponding to multiple satellite navigation frequency points; The positioning assistance data includes satellite ephemeris data, satellite clock difference data, satellite elevation angle, satellite azimuth angle, and receiver clock status data; The epoch synchronization preprocessing includes time alignment, satellite number matching, and frequency point identifier association, specifically: Time alignment of frequency point observation data according to observation epochs is performed to obtain synchronous observation data; Based on the satellite number, the frequency point observation data corresponding to different satellite navigation frequency points belonging to the same satellite are matched to obtain satellite matching observation data; Based on satellite navigation frequency identifiers, pseudorange observations, carrier phase observations, Doppler observations, and signal-to-noise ratios are associated with the corresponding satellite navigation frequencies to generate epoch-synchronized multi-frequency observation data.

3. The high-precision positioning method for GNSS receivers based on multi-frequency joint calculation according to claim 1, characterized in that, Step two specifically includes: Based on epoch-synchronous multi-frequency observation data and positioning auxiliary data, the pseudorange observation value, carrier phase observation value, Doppler observation value, signal-to-noise ratio, satellite navigation frequency point identifier, and satellite elevation angle and satellite azimuth angle corresponding to the satellite number are read for each satellite navigation frequency point to obtain the multi-frequency observation quality feature vector of each satellite navigation frequency point. The multi-frequency observation quality feature vector is input into the frequency point confidence dynamic calibration network, which includes a multi-frequency observation feature input layer, a frequency point embedding coding layer, an epoch time series stability coding layer, a frequency point consistency constraint layer, and a confidence weight output layer. In the frequency point embedding coding layer, a corresponding frequency point embedding vector is configured for each satellite navigation frequency point based on the satellite navigation frequency point identifier, and the multi-frequency observation quality feature vector is linearly fused with the corresponding frequency point embedding vector to generate an initial frequency point feature vector; In the epoch-time stability coding layer, the initial frequency point feature vectors of multiple consecutive observation epochs are encoded for epoch-time stability through one-dimensional temporal convolution to generate epoch-time stability feature vectors. In the frequency point consistency constraint layer, the mean deviation distance between the epoch time series stability feature vectors corresponding to different satellite navigation frequencies under the same satellite number is calculated to obtain the frequency point consistency deviation. In the confidence weight output layer, the epochal time series stability feature vector is linearly mapped to obtain the initial confidence score; The initial confidence score and the consistency deviation between frequency points are weighted and corrected, and then normalized using the Softmax function to obtain the frequency confidence weight of each satellite navigation frequency point. Frequency point confidence weights are added to the epoch-synchronized multi-frequency observation data according to satellite number, observation epoch, and satellite navigation frequency point identifier to obtain quality calibration multi-frequency observation data.

4. The high-precision positioning method for GNSS receivers based on multi-frequency joint calculation according to claim 1, characterized in that, Step three specifically includes: Based on the quality calibration multi-frequency observation data, read the pseudorange observation value, carrier phase observation value and frequency point confidence weight corresponding to each satellite navigation frequency point; Satellite ephemeris data and satellite clock bias data are read based on positioning-assisted data; Based on the satellite navigation frequency point identifier, determine the carrier wavelength corresponding to each satellite navigation frequency point; Based on satellite ephemeris data, determine the location of each satellite corresponding to its number; Construct a set of parameters to be estimated, which includes the target GNSS receiver's receiver position, receiver clock error, ionospheric delay parameter, tropospheric delay parameter, pseudo-range-frequency offset parameter, carrier phase-frequency offset parameter, and carrier phase ambiguity parameter. Calculate the spatial distance between the estimated receiver position of the target GNSS receiver and the positions of each satellite to obtain the geometric distance term; Based on pseudorange observations, geometric distance terms, receiver clock bias, satellite clock bias data, ionospheric delay parameters, tropospheric delay parameters, and pseudorange frequency offset parameters, pseudorange observation equations are established between pseudorange observations corresponding to each satellite navigation frequency point and the set of parameters to be estimated. Based on carrier phase observations, geometric distance terms, receiver clock bias, satellite clock bias data, ionospheric delay parameters, tropospheric delay parameters, carrier wavelength, carrier phase ambiguity parameters, and carrier phase inter-frequency deviation parameters, a carrier phase observation equation is established between the carrier phase observations corresponding to each satellite navigation frequency point and the set of parameters to be estimated. The frequency point confidence weights are generated into an observation weight matrix according to the observation epoch, satellite number, and satellite navigation frequency point identifier; Based on the pseudorange observation equation, generate pseudorange observation model values ​​for each satellite navigation frequency; based on the carrier phase observation equation, generate carrier phase observation model values ​​for each satellite navigation frequency. The pseudorange observation model values ​​and carrier phase observation model values ​​are vectorized according to the observation epoch, satellite number and satellite navigation frequency point identifier to generate a joint observation vector; Based on the observation weight matrix, the joint observation vectors are weighted and jointly organized to obtain a multi-frequency joint observation model.

5. The high-precision positioning method for GNSS receivers based on multi-frequency joint calculation according to claim 1, characterized in that, The multi-frequency residual Gram-Schmidt orthogonal correction technique specifically includes: Based on the multi-frequency joint observation model, the pseudorange observation model value and carrier phase observation model value corresponding to each satellite navigation frequency point are read; The difference between the pseudorange observation value and the pseudorange observation model value is calculated to obtain the pseudorange observation residual; the difference between the carrier phase observation value and the carrier phase observation model value is calculated to obtain the carrier phase observation residual. According to the observation epoch, satellite number, satellite navigation frequency point identifier, and observation type, the pseudorange observation residual and carrier phase observation residual are stacked into a multi-frequency residual vector; the observation type includes pseudorange observation type and carrier phase observation type; Based on the satellite number, satellite navigation frequency point identifier, carrier wavelength and observation type of each residual element in the multi-frequency residual vector, a common residual candidate basis matrix, a frequency dispersion residual candidate basis matrix, an inter-frequency deviation residual candidate basis matrix and an observation type difference residual candidate basis matrix are constructed. The observation weight matrix is ​​used as the Gram-Schmidt orthogonal weighted inner product metric matrix. The candidate basis matrix groups of residuals are subjected to hierarchical weighted orthogonal processing in the order of common residuals, frequency dispersion residuals, inter-frequency deviation residuals, and observation type difference residuals to obtain the common residual orthogonal basis matrix, frequency dispersion residual orthogonal basis matrix, inter-frequency deviation residual orthogonal basis matrix, and observation type difference residual orthogonal basis matrix. The multi-frequency residual vectors are projected onto the orthogonal subspaces corresponding to the common residual orthogonal basis matrix, the frequency dispersion residual orthogonal basis matrix, the inter-frequency deviation residual orthogonal basis matrix, and the observation type difference residual orthogonal basis matrix, respectively, to obtain the common residual components, the frequency dispersion residual components, the inter-frequency deviation residual components, and the observation type difference residual components; The average value of the frequency point confidence weight associated with each residual element in the common residual component, frequency dispersion residual component, inter-frequency bias residual component, and observation type difference residual component is calculated, and the average value is normalized to obtain the confidence scaling factor. The common residual component, frequency dispersion residual component, inter-frequency bias residual component and observation type difference residual component are scaled and corrected according to the confidence scaling factor, and the scaled and corrected residual components are weighted and superimposed to obtain the error correction component. Based on the error correction component, the joint observation vector in the multi-frequency joint observation model is corrected by observation compensation to obtain the error correction joint observation model.

6. The high-precision positioning method for GNSS receivers based on multi-frequency joint calculation according to claim 1, characterized in that, The improved LAMBDA ambiguity fixing algorithm specifically includes: Based on the error correction joint observation model, the least squares estimation method is used to solve the parameter set to be estimated, and the ambiguity floating-point solution and ambiguity covariance matrix corresponding to the carrier phase ambiguity parameter are extracted from the parameter solution results. The frequency point confidence weights are associated with the corresponding carrier phase ambiguity parameters according to the observation epoch, satellite number and satellite navigation frequency point identifier, and the carrier phase ambiguity parameters are divided into a high confidence ambiguity parameter group and a ambiguity parameter group to be verified based on the frequency point confidence weights. Perform decorrelation transformation on the floating-point fuzzy solution and fuzzy covariance matrix corresponding to the high-confidence fuzzy parameter set, and perform LAMBDA integer search in the integer search space after decorrelation transformation to obtain high-confidence fuzzy candidate solution and high-confidence suboptimal fuzzy candidate solution; Using the high-confidence fuzzy candidate solution as an integer constraint, the fuzzy floating-point solution and fuzzy covariance matrix corresponding to the fuzzy parameter group to be verified are updated with conditional constraints to obtain the conditional floating-point solution and conditional covariance matrix of the fuzzy parameter group to be verified. Based on the conditional floating-point solution and conditional covariance matrix of the ambiguity to be verified, a second-level LAMBDA integer search is performed on the parameter set of the ambiguity to be verified to obtain the candidate solution and the candidate suboptimal solution of the ambiguity to be verified. Based on the observation epoch, satellite number, and satellite navigation frequency identifier, the high-confidence ambiguity candidate solution and the ambiguity candidate solution to be verified are combined into the optimal ambiguity candidate solution, and the high-confidence suboptimal ambiguity candidate solution and the suboptimal ambiguity candidate solution to be verified are combined into the suboptimal ambiguity candidate solution. The objective function values ​​of the LAMBDA integer search corresponding to the suboptimal ambiguity candidate solution and the optimal ambiguity candidate solution are calculated respectively, and the ratio test value is obtained by ratio operation; Substitute the optimal ambiguity candidate solution and the suboptimal ambiguity candidate solution into the error correction joint observation model to obtain the optimal joint observation model vector and the suboptimal joint observation model vector, respectively. Calculate the vector differences between the optimal joint observation model vector and the suboptimal joint observation model vector and the joint observation vector after observation compensation correction, and calculate the L2 norm square of the vector differences to obtain the optimal observation residual evaluation value and the suboptimal observation residual evaluation value. Calculate the difference between the suboptimal observation residual evaluation value and the optimal observation residual evaluation value to obtain the residual convergence interval; If the ratio test value is greater than or equal to the dynamic ratio test threshold, and the residual convergence interval is greater than or equal to the residual convergence interval threshold, then the corresponding optimal ambiguity candidate solution is determined as the carrier phase ambiguity fixed vector.

7. The high-precision positioning method for GNSS receivers based on multi-frequency joint calculation according to claim 1, characterized in that, Step six specifically includes: The carrier phase ambiguity fixed vector is replaced with the carrier phase ambiguity parameter in the error correction joint observation model according to the observation epoch, satellite number and satellite navigation frequency point identifier to obtain the fixed correction observation model. Based on the fixed correction observation model, the receiver position, receiver clock error, ionospheric delay parameter, tropospheric delay parameter, pseudo-range-frequency offset parameter and carrier phase-frequency offset parameter in the parameter set to be estimated are jointly solved to obtain the joint solution results of the parameters; The three-dimensional position coordinates and receiver clock error of the target GNSS receiver are extracted from the joint parameter solution results, and the positioning residual is calculated based on the fixed correction observation model. The three-dimensional position coordinates, receiver clock error, and positioning residual are used as the high-precision positioning status results of the receiver.

8. The high-precision positioning method for GNSS receivers based on multi-frequency joint calculation according to claim 1, characterized in that, Step seven specifically includes: The root mean square value of the positioning residual is used as the evaluation value of positioning accuracy. Read the ambiguity fixing status corresponding to the carrier phase ambiguity fixing vector; the ambiguity fixing status includes carrier phase ambiguity fixing successful and carrier phase ambiguity fixing unsuccessful. Based on the mean and standard deviation of the absolute values ​​of each positioning residual element corresponding to the observation epoch, a positioning accuracy threshold is generated; A positioning status identifier is generated based on the positioning accuracy evaluation value and the ambiguity fixation status: if the positioning accuracy evaluation value is less than or equal to the positioning accuracy threshold and the carrier phase ambiguity is successfully fixed, the positioning status identifier is 1; if the positioning accuracy evaluation value is greater than the positioning accuracy threshold, or the carrier phase ambiguity is not successfully fixed, the positioning status identifier is 0. The three-dimensional position coordinates, receiver clock error, positioning residual, positioning accuracy evaluation value, ambiguity fixed state, and positioning status identifier are associated and encapsulated according to the observation epoch to obtain the high-precision positioning result of the target GNSS receiver.

Images