PPP (Point-to-Point Protocol) observation quality control method fusing mutual verification and residual feedback between systems

Abstract

The invention discloses a PPP observation quality control method fusing inter-system mutual verification and residual feedback, and belongs to the technical field of satellite navigation high-precision positioning. The method comprises the following steps: firstly, acquiring current epoch multi-system multi-frequency original observation data and corresponding prior state information; secondly, performing gross error detection and elimination on the pseudo-range observation value by adopting an RANSAC method, and constructing a pseudo-range healthy satellite subset; then, carrier phase anomaly detection and cycle slip repair are carried out based on the subset, and an inter-system mutual verification mechanism and a dynamic weight adjustment strategy are introduced to further improve the reliability of an observation value; and finally, performing PPP calculation by using the high-quality observation data subjected to multi-level verification, and outputting a steady positioning result. According to the method, the problems that in a traditional method, due to pseudo-range gross error interference, carrier phase cycle slip misjudgment or missing detection is caused, and the anomaly detection capacity is insufficient in a multi-satellite simultaneous fault scene are effectively solved, the robustness and positioning precision of PPP in a complex environment are enhanced, and technical support is provided for high-reliability GNSS application.

Description

Technical Field

[0001] This invention relates to the field of high-precision positioning technology for Global Navigation Satellite System (GNSS), specifically to an observation quality control method for Precise Point Positioning (PPP), which is particularly suitable for robust detection and processing of pseudorange gross errors, multipath and carrier phase cycle slips in complex environments such as urban canyons and forest areas, so as to improve the accuracy, convergence speed and reliability of PPP positioning. Background Technology

[0002] Precise Point Positioning (PPP) technology utilizes carrier phase and pseudorange observations from a single receiver, combined with externally provided precise satellite orbit and clock bias products, to achieve centimeter-level high-precision positioning globally. However, in practical applications, PPP performance is highly susceptible to the impact of observation quality. Especially in complex environments, GNSS signals are often affected by multipath effects, obstruction, atmospheric disturbances, and signal interference, leading to anomalies in observations. For example, pseudorange observations are easily affected by multipath and gross errors, potentially resulting in deviations ranging from decimeters to meters; while carrier phase observations are highly accurate, cycle slips can occur, leading to integer cycle count errors. To address these issues, existing technologies typically employ data probing or simple residual checks to identify anomalous observations, but these methods suffer from drawbacks such as coupling interference, sensitivity to multi-satellite failures, and reliance on prior information. Therefore, a robust observation data quality control method is urgently needed that can decouple pseudorange and carrier phase anomaly detection. Summary of the Invention

[0003] In view of this, the present invention provides a PPP observation quality control method that integrates inter-system mutual verification and residual feedback. This method controls the observation quality of Precise Point Positioning (PPP) based on a multi-level collaborative detection and adaptive feedback mechanism. It aims to solve the problems of existing PPP technology in complex environments, such as carrier phase cycle slip misjudgment caused by pseudorange gross error interference, insufficient anomaly detection capability under multi-satellite faults, and difficulty in identifying systematic deviations.

[0004] The technical solution adopted in this invention is as follows:

[0005] A PPP observation quality control method integrating inter-system cross-verification and residual feedback includes the following steps:

[0006] Step 1: Obtain the raw GNSS observation data and precise orbit clock error products of the current epoch multi-system multi-frequency GNSS, and use the state estimation results of the previous epoch of the PPP filter as prior information;

[0007] Step 2: Remove satellites with a carrier-to-noise ratio below the threshold, an elevation angle below the cutoff elevation angle, or unhealthy ephemeris. Then, use the RANSAC algorithm to detect anomalies in the pseudorange observations of the remaining satellites without ionospheric combination to obtain a set of pseudorange healthy satellites. At the same time, form a suspicious satellite set by arranging the satellites that have been excluded multiple times in the RANSAC algorithm iteration.

[0008] Step 3: Perform preliminary PPP calculation based on the pseudorange healthy satellite set and calculate the carrier phase pre-verification residual of each satellite; take the satellites of the Beidou satellite navigation system as the benchmark, estimate the deviation trend of the carrier phase residual of other navigation system satellites relative to the satellites in the same elevation angle region of the Beidou satellite navigation system according to the elevation angle interval; if the deviation trend exceeds the preset threshold for multiple consecutive epochs, dynamically reduce the observation weight of the corresponding system and perform cycle slip detection and repair on the suspicious satellite set.

[0009] Step 4: Input the observation data of all available satellites, including the repaired suspicious satellites, processed in Step 3 into the main Kalman filter for state update to obtain the state estimate and post-hoc residual for the current epoch.

[0010] Step 5: Perform a consistency check based on the post-test residuals. If the check passes, output the state estimation result of the current epoch as the PPP positioning result. If it fails, retain the PPP positioning result of the previous epoch, or use only the BeiDou satellite navigation system to perform PPP calculation and output the PPP positioning result.

[0011] Step 6: Based on the statistical characteristics of the post-hoc residuals, adaptively adjust the prior information for the next epoch.

[0012] Furthermore, in step 2 of the RANSAC algorithm, four satellites are randomly selected as the minimum configuration sample in each iteration. The least squares method is used to solve for the receiver position and clock error, generating a candidate positioning solution. If the ionospheric-free combined pseudorange residual of a certain satellite satisfies... If so, then in the current iteration, the satellite is determined to be an interior point; where, , This represents the prior error of the pseudorange observations. The threshold is determined based on the significance level. For the first The satellite in the pseudo-range residuals in the next iteration For the first Ionospheric pseudorange observations of the satellites. For the first The geometric distance between the satellite and the receiver For the first The state estimation vector in the next iteration contains estimates of the receiver position and clock bias. The speed of light in a vacuum. For the first The estimated value of the receiver clock bias in the next iteration. For the first The tropospheric delay corresponding to each satellite For the first The observation noise of each satellite and the number of iterations can be set according to the number of satellites and the preset inlier rate.

[0013] Furthermore, in step 2, after the RANSAC algorithm completes its iterations, the model with the minimum sum of squared errors is selected as the optimal localization solution, where the first... Minimum sum of squared errors in the next iteration The calculation method is as follows:

[0014]

[0015] in, Indicates the first The set of satellite indices identified as interior points in the next iteration;

[0016] Count the number of times each satellite is excluded as an outlier in all iterations. If a satellite is excluded more than 70% of the total number of iterations, it is marked as a suspected faulty satellite and added to the suspicious satellite set.

[0017] Furthermore, in step 3, the deviation trend is estimated as follows:

[0018] For each preset elevation angle interval, the theoretical values ​​of the carrier phase of each satellite in the current epoch are predicted using the receiver position, clock error and atmospheric parameters estimated in the previous epoch.

[0019] The prediction bias for each satellite is obtained by subtracting the theoretical carrier phase value from the actual observed carrier phase value.

[0020] The average deviation of the predictions for BeiDou system satellites and non-BeiDou system satellites within the current elevation angle range is calculated separately to obtain the average deviation of the two systems.

[0021] The difference between the average deviations of the two systems is the inter-system deviation trend in the current elevation angle range, which is used to reflect the observation consistency of different navigation systems under the same observation conditions.

[0022] Furthermore, step 6 is specifically implemented as follows:

[0023] Step 601: Standardize the carrier phase post-hoc residuals of each satellite in the current epoch to obtain the standardized residuals. :

[0024]

[0025] in, For the first Post-test residuals of each satellite The covariance matrix of the filter state estimation is related to the first... The diagonal elements of the state components corresponding to the observations of each satellite;

[0026] Step 602, if 3, then determine the first The observation data from the first satellite showed anomalies, and the second... The satellite is excluded from the set of satellites participating in the positioning solution at the current epoch, and a filtering update is performed again based on the remaining health observations to obtain a corrected state estimate;

[0027] Step 603: Calculate the root mean square (RMS) of the carrier phase post-verification residuals for each navigation satellite system, including BDS, GPS, Galileo, and GLONASS, at each operating frequency. If the RMS of a navigation satellite system at a certain frequency exceeds 0.05m for two consecutive epochs, then in the next epoch, increase the process noise variance of the corresponding state component of that navigation satellite system using the following formula:

[0028]

[0029] in: Let V be the initial process noise variance of this state component. The root mean square of the carrier phase residual after verification for this navigation satellite system at the current frequency. This is an empirical proportionality coefficient.

[0030] The advantages of this invention compared to the prior art are:

[0031] (1) This invention uses a layered design to first process pseudorange anomalies, which effectively avoids the contamination of carrier phase residuals by large pseudorange deviations, so that the cycle slip detection algorithm can still maintain a high accuracy in complex environments.

[0032] (2) Based on pseudorange “maximum consistency” detection, even if there are multiple abnormal satellites, as long as there is a subset consisting of healthy satellites, the abnormal satellites can be correctly identified and removed, which has strong robustness against multi-satellite failures.

[0033] (3) This invention provides high-quality observation input for the PPP master filter, which can significantly improve positioning accuracy and effectively shorten convergence time, thereby enhancing the reliability of PPP in challenging environments such as urban canyons.

[0034] (4) Compared with existing satellite navigation observation quality control methods, the present invention has a clear process and adopts a modular design, which can be integrated into various high-precision GNSS receiver firmware or data processing software, and is suitable for various application scenarios such as autonomous driving, drones, and surveying.

[0035] In summary, this invention significantly improves the robustness, convergence speed, and positioning reliability of PPP in complex scenarios such as urban canyons, ionospheric disturbances, and severe multipath conditions through a four-level collaborative processing architecture consisting of pseudorange gross error screening, carrier anomaly repair, inter-system consistency verification, and residual feedback covariance adjustment. In particular, it enhances the ability to suppress multi-satellite concurrent failures and systematic biases, providing effective technical support for high integrity and high precision GNSS applications. Attached Figure Description

[0036] Figure 1 This is a principle block diagram of the method according to an embodiment of the present invention;

[0037] Figure 2 This is a block diagram illustrating the principle of pseudorange initial screening and suspicious satellite marking in this embodiment of the invention. Detailed Implementation

[0038] The present invention will now be described in further detail with reference to the accompanying drawings.

[0039] A PPP observation quality control method integrating inter-system cross-verification and residual feedback is proposed. This method first acquires multi-system (GPS, BDS, Galileo, GLONASS) multi-frequency (at least dual-frequency) raw GNSS observation data at the current epoch, including pseudorange and carrier phase observations, as well as externally provided precise satellite orbit and clock bias products. The state estimation results of the previous epoch from the precise point positioning (PPP) filter (including receiver position, clock offset, tropospheric delay, etc.) are used as prior information. Then, satellites with carrier-to-noise ratios below a threshold, elevation angles below a set cutoff elevation angle, or unhealthy ephemeris are removed. The RANSAC algorithm is used to detect anomalies in the ionospheric-free combination pseudorange observations of the remaining satellites, obtaining a set of pseudorange-healthy satellites. Simultaneously, satellites repeatedly excluded during the RANSAC iteration process are designated as a suspicious satellite set. Preliminary PPP calculations are performed based on the aforementioned pseudorange healthy satellite set. The pre-hoc residuals of the carrier phase of each satellite are calculated. Then, using BDS system satellites as a benchmark, the deviation trend of the carrier phase residuals of other navigation system satellites relative to BDS satellites in the same elevation angle region is evaluated. If the deviation trend exceeds a preset threshold for multiple consecutive epochs, the observation weight of the corresponding system is dynamically reduced. At the same time, cycle slip detection and repair are performed on suspicious satellite sets. The processed observation data is input into the master Kalman filter for state update to obtain the state estimate and post-hoc residuals for the current epoch. The post-hoc residuals are subjected to a consistency check. If the check passes, the state estimate for the current epoch is output as the final PPP positioning result. Simultaneously, based on the statistical characteristics of the post-hoc residuals, the process noise of the state components affected by anomalies in the prior covariance matrix of the state in the next epoch is adaptively adjusted to enhance the robustness of the filter to potential observation anomalies.

[0040] The method specifically includes the following steps:

[0041] (1) Obtain raw GNSS observation data from multiple systems (GPS, BDS, Galileo, GLONASS) at the current epoch, including pseudorange and carrier phase observations, as well as externally provided precision satellite orbit and clock bias products. Use the state estimation results of the previous epoch (including receiver position, clock offset, tropospheric delay, etc.) from the precision point positioning (PPP) filter as prior information. The specific steps are as follows:

[0042] (101) Read the raw observation data (including pseudorange) of the current epoch from the GNSS data acquisition equipment in real time. carrier phase Carrier-to-noise ratio (etc.) and broadcast ephemeris information, and obtains precise satellite orbit (SP3 format) and clock bias (CLK format) products for the corresponding epoch from external data sources (such as IGS, WHU, CNES, and other analysis centers). Among them: For the receiver, For satellites, For frequencies (at least including dual frequencies such as L1 / L2 or B1 / B2). For satellite systems ( The value ranges from 1 to 4, with the four values ​​representing BDS, GPS, GLONASS, and Galileo respectively.

[0043] (102) Obtain the state vector calculated from the Kalman filter of the previous epoch. and its covariance matrix .in: The state vector represents the current observation epoch number and includes at least the receiver position. Receiver clock bias Tropospheric wet delay Ambiguity parameters of each system at each frequency ;

[0044] (103) Based on the above broadcast ephemeris, under a unified time reference (GPST) and coordinate frame (ITRF), calculate the satellite positions and elevation angles of each system. and azimuth .

[0045] (2) Satellites with a carrier-to-noise ratio below the threshold, an elevation angle below the set cutoff elevation angle, or unhealthy ephemeris are removed. The RANSAC algorithm is then used to detect anomalies in the pseudorange observations of the remaining satellites without ionospheric combinations, resulting in a set of pseudorange-healthy satellites. Simultaneously, satellites repeatedly excluded during the RANSAC iteration process are designated as a suspicious satellite set. The specific steps are as follows:

[0046] (201) If the satellite elevation angle <10° or If a dB-Hz or satellite broadcast ephemeris is marked as "unhealthy", the satellite will be removed from the subsequent processing of the current epoch.

[0047] (202) Using dual-frequency observations, construct the ionosphere-free combination equation:

[0048]

[0049] (203) Perform RANSAC iterative processing (assuming iteration is required) (This is repeated 10 times). In each iteration, the minimum configuration sample (usually 4 satellites) is randomly selected, and the receiver position and clock error are calculated using the least squares method to form a candidate positioning model.

[0050] Assuming the current is the th In the next iteration, the pseudorange residuals of each satellite without ionospheric combination are... Represented as:

[0051]

[0052] in, For the first The satellite in the pseudo-range residuals in the next iteration For the first Ionospheric pseudorange observations of the satellites. For the first The geometric distance between the satellite and the receiver For the first The state estimation vector in the next iteration (containing estimates of parameters such as receiver position and clock error). The speed of light in a vacuum. The estimated value of the receiver clock bias (the first value) (nth iteration) For the first The tropospheric delay corresponding to each satellite For the first The observation noise of each satellite.

[0053] like If the condition is met, the satellite is classified as an interior point in the current iteration; otherwise, it is classified as an exterior point. The prior mean square error of the pseudorange observations. Based on significance level (e.g.) The determined threshold is usually taken as... .

[0054] Then calculate the sum of squared residuals for the interior satellite set in this iteration. :

[0055]

[0056] (204) Completed After each iteration, the model with the smallest SSE value is selected from all valid localization solutions (i.e., the number of inliers ≥ 4) as the optimal localization solution, and its corresponding inlier satellite set is determined as the pseudorange healthy satellite set. Simultaneously, the number of times each satellite is identified as an outlier during the iteration process is counted. If a satellite is excluded from the inlier set more than 70% of the total number of iterations, it is marked as a suspected faulty satellite and added to the suspicious satellite set.

[0057] (3) Based on the pseudorange healthy satellite set, perform preliminary PPP calculation, calculate the pre-approval residuals of each satellite's carrier phase, and then, using BDS system satellites as a benchmark, evaluate the deviation trend of the carrier phase residuals of other navigation system satellites relative to BDS satellites in the same elevation angle region. If the deviation trend exceeds the preset threshold for multiple consecutive epochs, dynamically reduce the observation weight of the corresponding system, and focus on cycle slip detection and repair for suspicious satellite sets; the specific method is as follows:

[0058] (301) Use the pseudorange healthy satellite set determined in step (2) to perform PPP filtering solution, and synchronously calculate and output the pre-hoc residuals of each satellite carrier phase;

[0059] (302) Divide the visible satellites into multiple intervals according to their elevation angles (e.g., with 10° intervals). Then, within each elevation angle interval, first calculate the mean value of the carrier phase prior residuals of the BDS satellites. Then, calculate the mean residual of each non-BDS system (such as GPS, Galileo, GLONASS) within this interval, and find its mean relative to the mean residual. Deviation;

[0060] (303) The weighted average of the systematic biases obtained from all effective elevation angle intervals (i.e., intervals with no less than 2 BDS satellites) is used to obtain the comprehensive residual bias of the system. ,in:

[0061] ,

[0062] Subsequently, based on a length of A sliding window is used to determine whether the system is continuous. Each epoch satisfies If the condition is met, the system is determined to have a persistent bias, and the observation noise variance of all its frequency observations is amplified by a factor of 3, thereby dynamically reducing its weight in the Kalman filter.

[0063] (304) For the suspicious satellite set, carrier cycle slip detection is performed using the Melbourne-Wübbena (MW) combination and the geometrically independent (GF) combination. If the change in the residual of the MW combination is greater than 0.1 cycles and the change in the residual of the GF combination is greater than 0.1 m, the satellite is determined to have experienced a cycle slip. At this time, the carrier phase ambiguity is repaired using a polynomial fitting method, or the corresponding ambiguity parameters are reset if the repair is unreliable, and the re-initialization process is initiated.

[0064] (4) Based on the pseudorange healthy satellite set obtained in step (2) and the suspected satellites confirmed to be valid after cycle slip repair in step (304), a set of available satellites for the current epoch is formed; the observation data of this set is input into the master Kalman filter for state update, wherein the observation noise covariance of each navigation system has been adaptively amplified according to the deviation evaluation results in step (303); the state vector of the current epoch and the post-hoc residuals of each satellite are output. ;

[0065] (5) Perform a consistency test on the post-test residuals obtained in step (4). If the test passes, output the state estimate of the current epoch as the final PPP positioning result.

[0066] (6) Based on the statistical characteristics of the post-hoc residuals, the process noise of the state components affected by anomalies in the prior covariance matrix of the next epoch is adaptively adjusted to enhance the robustness of the filter to potential observation anomalies; the specific method is as follows:

[0067] (601) Standardize the post-hoc residuals of each satellite calculated in step (4), and denote the standardized residuals as . :

[0068]

[0069] in, For the first Post-test residuals of each satellite This represents the diagonal element of the state component in the filter state estimation covariance matrix that corresponds to the satellite observation.

[0070] like 3. Exclude the satellite from the set of satellites participating in the positioning solution at the current epoch, and re-execute a filtering update based on the remaining health observations. If the overall chi-square test after re-resolution passes, the state estimation result of the current epoch is output as the final precise point positioning (PPP) solution; if it still fails, the valid solution of the previous epoch is retained or downgraded to solving using only a more reliable subset such as BDS.

[0071] (602) For all navigation systems (including BDS) involved in the positioning calculation and their respective frequency signals, calculate the root mean square (RMS) of the carrier phase post-hoc residuals at each frequency; if the RMS of a certain system at a certain frequency exceeds 0.05m for two consecutive epochs, then in the next epoch, use the following formula to specifically increase the process noise variance of its corresponding state components (such as receiver position, ambiguity parameters, etc.):

[0072] ,

[0073] (603) Update the state prior covariance matrix using the following formula for filtering in subsequent epochs:

[0074]

[0075] in: Here is the state transition matrix. Let be the posterior covariance of the previous epoch.

[0076] This method significantly improves the robustness, convergence speed, and positioning reliability of PPP in complex scenarios such as urban canyons, ionospheric disturbances, and severe multipath conditions through a four-level collaborative processing architecture consisting of pseudorange gross error screening, carrier anomaly repair, inter-system consistency verification, and residual feedback covariance adjustment. In particular, it enhances the ability to suppress multi-satellite concurrent failures and systematic biases, providing effective technical support for high integrity and high precision GNSS applications.

[0077] Here is a more specific example:

[0078] A PPP observation quality control method integrating inter-system cross-verification and residual feedback, with implementation process referenced. Figure 1 This includes the following steps:

[0079] (1) Read the current epoch multi-system raw observation data (including pseudorange) from the GNSS data acquisition equipment in real time. carrier phase Carrier-to-noise ratio (etc.) and broadcast ephemeris information, and obtains precise satellite orbit (SP3 format) and clock bias (CLK format) products for the corresponding epoch from external data sources (such as IGS, WHU, CNES, and other analysis centers). Among them: For the receiver, For satellites, For frequencies (at least including dual frequencies such as L1 / L2 or B1 / B2). For satellite systems ( The value ranges from 1 to 4, with the four values ​​representing BDS, GPS, GLONASS, and Galileo respectively.

[0080] Obtain the state vector calculated from the Kalman filter of the previous epoch. and its covariance matrix .in: The state vector represents the current observation epoch number and includes at least the receiver position. Receiver clock bias Tropospheric wet delay Ambiguity parameters of each system at each frequency .

[0081] Based on the acquired broadcast ephemeris, the positions and elevation angles of satellites in each system are calculated under a unified time reference (GPST) and coordinate frame (ITRF). and azimuth .

[0082] (2) For example Figure 2 As shown, if the satellite elevation angle <10° or If a satellite is marked as "unhealthy" in dB-Hz or satellite broadcast ephemeris, it is removed from the subsequent processing of the current epoch. An ionospherically unaffected combination equation is constructed using dual-frequency observations:

[0083]

[0084] RANSAC iterative processing is performed. In each iteration, the minimum configuration sample (usually 4 satellites) is randomly selected, and the receiver position and clock error are calculated using the least squares method to form a candidate positioning model.

[0085] Number of iterations Adjustments can be made based on the actual scenario, or estimates can be made using the following formula:

[0086]

[0087] in, This represents the expected confidence level, typically taken as 0.99. This represents the proportion of outliers in the dataset (i.e., the proportion of gross errors). This represents the number of samples required for each sampling.

[0088] The table below shows reference values ​​for the number of RANSAC iterations in some typical scenarios:

[0089] Table 1. Reference values ​​for the number of RANSAC iterations in typical scenarios.

[0090]

[0091] Assuming the current is the th In the next iteration, the pseudorange residuals of each satellite without ionospheric combination are... Represented as:

[0092]

[0093] in, The geometric distance from the station to the satellite. At the speed of light, For tropospheric delay, It is noise.

[0094] like If the condition is met, the satellite is classified as an interior point in the current iteration; otherwise, it is classified as an exterior point. The prior mean square error of the pseudorange observations. Based on significance level (e.g.) The determined threshold is usually taken as... .

[0095] Then calculate the sum of squared residuals for the interior satellite set in this iteration. :

[0096]

[0097] Finish After each iteration, the model with the smallest SSE value is selected from all valid localization solutions (i.e., the number of inliers ≥ 4) as the optimal localization solution, and its corresponding inlier satellite set is determined as the pseudorange healthy satellite set. Simultaneously, the number of times each satellite is identified as an outlier during the iteration process is counted. If a satellite is excluded from the inlier set more than 70% of the total number of iterations, it is marked as a suspected faulty satellite and added to the suspicious satellite set.

[0098] (3) Using the pseudorange healthy satellite set determined in the above steps, perform PPP filtering and simultaneously calculate and output the pre-approval residuals of the carrier phase of each satellite. First, divide the visible satellites into multiple intervals according to their elevation angles (e.g., with 10° intervals). Then, within each elevation angle interval, first calculate the mean value of the pre-approval residuals of the carrier phase of the BDS satellites. Then, calculate the mean residual of each non-BDS system (such as GPS, Galileo, GLONASS) within this interval, and find its mean relative to the mean residual. The deviation is calculated by weighting the systematic deviations obtained from all effective elevation angle intervals (i.e., intervals with at least 2 BDS satellites) to obtain the comprehensive residual deviation of the system. ,in:

[0099] ,

[0100] Subsequently, based on a length of (for example Using a sliding window (=3), determine whether the system is continuous. Each epoch satisfies If the condition is met, the system is determined to have a persistent bias, and the observation noise variance of all its frequency observations is amplified by a factor of 3, thereby dynamically reducing its weight in the Kalman filter.

[0101] For suspicious satellite sets, carrier cycle slip detection is performed using the Melbourne-Wübbena (MW) combination and the geometrically independent (GF) combination. If the change in the MW combination residual is greater than 0.1 cycles and the change in the GF combination residual is greater than 0.1 m, the satellite is determined to have experienced a cycle slip. In this case, a polynomial fitting method is used to repair the carrier phase ambiguity, or if the repair is unreliable, the corresponding ambiguity parameters are reset, and a re-initialization process is initiated.

[0102] (4) Based on the set of pseudorange healthy satellites obtained in step (2) and the suspected satellites confirmed to be valid after cycle slip repair in step (3), a set of available satellites for the current epoch is formed. The observation data of this set is input into the master Kalman filter for state update, wherein the observation noise covariance of each navigation system has been adaptively amplified according to the deviation evaluation results in step (3), and the state vector of the current epoch and the post-hoc residuals of each satellite are output. .

[0103] (5) Perform a consistency test on the post-test residuals obtained in step (4). If the test passes, output the state estimate of the current epoch as the final PPP positioning result.

[0104] (6) Standardize the post-test residuals for each satellite, and denot the standardized residuals as follows: :

[0105]

[0106] in, For the first Post-test residuals of each satellite This represents the diagonal element of the state component in the filter state estimation covariance matrix that corresponds to the satellite observation.

[0107] like 3. Exclude the satellite from the set of satellites participating in the positioning solution at the current epoch, and re-execute a filtering update based on the remaining health observations. If the overall chi-square test after re-resolution passes, the state estimation result of the current epoch is output as the final precise point positioning (PPP) solution; if it still fails, the valid solution of the previous epoch is retained or downgraded to solving using only a more reliable subset such as BDS.

[0108] For all navigation systems (including BDS) involved in positioning calculation and their respective frequency signals, the root mean square (RMS) of the carrier phase post-hoc residuals at each frequency is calculated. If the RMS of a certain system at a certain frequency exceeds 0.05m for two consecutive epochs, then in the next epoch, the process noise variance of its corresponding state components (such as receiver position, ambiguity parameters, etc.) is increased in a targeted manner using the following formula.

[0109] ,

[0110] Then, the prior state covariance matrix is ​​updated using the following formula for filtering in subsequent epochs:

[0111]

[0112] in: Here is the state transition matrix. Let be the posterior covariance of the previous epoch.

[0113] In summary, this invention first employs the RANSAC algorithm to eliminate pseudorange gross errors and construct a healthy satellite set. Then, using BDS as a benchmark, it estimates the carrier phase residual deviations of other systems through elevation angle partitioning, dynamically downweights abnormal systems, and performs cycle slip detection and repair on suspicious satellites. Based on this, it integrates standardized post-hoc residuals and chi-square tests to further eliminate abnormal observations and recalculate the model. Finally, based on the RMS trend of the residuals at each system-frequency point, it adaptively increases the process noise of the corresponding state and updates the prior covariance matrix. This method can significantly improve the robustness, integrity, and positioning accuracy of PPP in complex environments.

[0114] This invention proposes a hierarchical, progressive, and multi-system collaborative quality control framework, effectively solving the problems of misjudgment or missed detection of carrier phase cycle slips caused by pseudorange gross interference in traditional methods, as well as insufficient anomaly detection capability in scenarios with simultaneous multi-satellite failures. The method of this invention significantly enhances the robustness and positioning accuracy of PPP in complex environments (such as urban canyons and ionospheric disturbances), providing key technical support for highly reliable GNSS applications.

Claims

1. A method for PPP observation quality control by fusing inter-system mutual checking and residual feedback, characterized in that, Includes the following steps: Step 1: Obtain the raw GNSS observation data and precise orbit clock error products of the current epoch multi-system multi-frequency GNSS, and use the state estimation results of the previous epoch of the PPP filter as prior information; Step 2: Remove satellites with a carrier-to-noise ratio below the threshold, an elevation angle below the cutoff elevation angle, or unhealthy ephemeris. Then, use the RANSAC algorithm to detect anomalies in the pseudorange observations of the remaining satellites without ionospheric combination to obtain a set of pseudorange healthy satellites. At the same time, form a suspicious satellite set by arranging the satellites that have been excluded multiple times in the RANSAC algorithm iteration. Step 3: Perform preliminary PPP calculation based on the pseudorange healthy satellite set, and calculate the pre-approval residual of each satellite carrier phase; Based on the satellites of the BeiDou Navigation Satellite System, the deviation trend of the carrier phase residual of other navigation system satellites relative to the satellites in the same elevation region of the BeiDou Navigation Satellite System is estimated according to the elevation angle interval. If the deviation trend exceeds the preset threshold for multiple consecutive epochs, the observation weight of the corresponding system is dynamically reduced, and cycle slip detection and repair are performed on the suspicious satellite set. Step 4: Input the observation data of all available satellites, including the repaired suspicious satellites, processed in Step 3 into the main Kalman filter for state update to obtain the state estimate and post-hoc residual for the current epoch. Step 5: Perform a consistency check based on the post-test residuals. If the check passes, output the state estimation result of the current epoch as the PPP positioning result. If it fails, retain the PPP positioning result of the previous epoch, or use only the BeiDou satellite navigation system to perform PPP calculation and output the PPP positioning result. Step 6: Based on the statistical characteristics of the post-hoc residuals, adaptively adjust the prior information for the next epoch.

2. The PPP observation quality control method based on inter-system mutual verification and residual feedback according to claim 1, characterized in that, In step 2 of the RANSAC algorithm, four satellites are randomly selected as the minimum configuration sample in each iteration. The least squares method is used to solve the receiver position and clock error to generate a candidate positioning solution. If the pseudorange residual of a satellite with ionospheric combination satisfies If so, then in the current iteration, the satellite is determined to be an interior point; where, , This represents the prior error of the pseudorange observations. The threshold is determined based on the significance level. For the first The satellite in the pseudo-range residuals in the next iteration For the first Ionospheric pseudorange observations of the satellites. For the first The geometric distance between the satellite and the receiver For the first The state estimation vector in the next iteration contains estimates of the receiver position and clock bias. The speed of light in a vacuum. For the first The estimated value of the receiver clock bias in the next iteration. For the first The tropospheric delay corresponding to each satellite For the first The observation noise of each satellite.

3. The PPP observation quality control method based on inter-system mutual verification and residual feedback according to claim 2, characterized in that, In step 2, after the RANSAC algorithm completes its iterations, the model with the minimum sum of squared errors is selected as the optimal localization solution. The model with the minimum sum of squared errors is then chosen as the optimal localization solution. Minimum sum of squared errors in the next iteration The calculation method is as follows: in, Indicates the first The set of satellite indices identified as interior points in the next iteration; Count the number of times each satellite is excluded as an outlier in all iterations. If a satellite is excluded more than 70% of the total number of iterations, it is marked as a suspected faulty satellite and added to the suspicious satellite set.

4. The PPP observation quality control method based on inter-system mutual verification and residual feedback according to claim 3, characterized in that, In step 3, the deviation trend is estimated as follows: For each preset elevation angle interval, the theoretical values ​​of the carrier phase of each satellite in the current epoch are predicted using the receiver position, clock error and atmospheric parameters estimated in the previous epoch. The prediction bias for each satellite is obtained by subtracting the theoretical carrier phase value from the actual observed carrier phase value. The average deviation of the predictions for BeiDou system satellites and non-BeiDou system satellites within the current elevation angle range is calculated separately to obtain the average deviation of the two systems. The difference between the average deviations of the two systems is the inter-system deviation trend in the current elevation angle range, which is used to reflect the observation consistency of different navigation systems under the same observation conditions.

5. The PPP observation quality control method based on inter-system mutual verification and residual feedback according to claim 4, characterized in that, The specific method for step 6 is as follows: Step 601: Standardize the carrier phase post-hoc residuals of each satellite in the current epoch to obtain the standardized residuals. : in, For the first Post-test residuals of each satellite The covariance matrix of the filter state estimation is related to the first... The diagonal elements of the state components corresponding to the observations of each satellite; Step 602, if 3, then determine the first The observation data from the first satellite showed anomalies, and the second... The satellite is excluded from the set of satellites participating in the positioning solution at the current epoch, and a filtering update is performed again based on the remaining health observations to obtain a corrected state estimate; Step 603: Calculate the root mean square (RMS) of the carrier phase post-verification residuals for each navigation satellite system, including BDS, GPS, Galileo, and GLONASS, at each operating frequency. If the RMS of a navigation satellite system at a certain frequency exceeds 0.05m for two consecutive epochs, then in the next epoch, increase the process noise variance of the corresponding state component of that navigation satellite system using the following formula: in: Let V be the initial process noise variance of this state component. The root mean square of the carrier phase residual after verification for this navigation satellite system at the current frequency. This is an empirical proportionality coefficient.

Images