GNSS troposphere delay error real-time correction method, device and medium

Abstract

The invention relates to the technical field of GNSS high-precision positioning calculation, and discloses a GNSS troposphere delay error real-time correction method and device and a medium, and the method comprises the steps: presetting a first process noise model and a second process noise model corresponding to different constraint values; acquiring motion state parameters of a GNSS receiver, atmospheric disturbance characteristic measurement of bypass information and a model mismatch index of a Kalman filter; when a preset dynamic motion condition is met, selecting a first process noise model; when the preset dynamic motion condition is not met, the first process noise model or the second process noise model is selected according to the atmospheric disturbance characteristic measurement and the model mismatch index, multi-source information is used for cooperative judgment, an algorithm closed loop capable of achieving autonomous recognition and avoiding carrier motion pollution and bypass information failure is constructed, and the algorithm closed loop is optimized. Mathematical coupling of elevation and zenith delay is avoided, and high-reliability rapid convergence can be achieved for independent users without network assistance.

Description

Technical Field

[0001] This invention relates to a method, device, and medium for real-time correction of GNSS tropospheric delay error, belonging to the field of GNSS high-precision positioning and calculation technology. Background Technology

[0002] In current high-precision positioning calculations for Global Navigation Satellite Systems (GNSS), particularly in Precise Point Positioning (PPP) applications, the industry standard practice is to construct a state vector. This vector simultaneously includes the station's position parameters, clock error parameters, integer ambiguity, and tropospheric zenith total delay (ZTD) parameters to be estimated. Statistical estimation algorithms, such as Kalman filtering, are then used to perform real-time optimal estimation of this state vector. Within this framework, the station elevation parameters to be estimated in the observation equations and the tropospheric zenith total delay parameters exhibit a high degree of similarity in mathematical form, showing a strong correlation. This strong coupling between parameters makes it difficult for statistical filters to clearly distinguish, in the early stages of calculation, whether the vertical variation in the observation residuals originates from the station's actual elevation movement or from changes in atmospheric water vapor content. The filter must rely on the slow changes in satellite geometry to gradually separate these two parameters, a process that often takes tens of minutes. This results in excessively long convergence times for high-precision positioning, hindering its application in scenarios requiring real-time, rapid acquisition of high-precision results, such as autonomous driving and precision exploration.

[0003] To alleviate this problem, one approach to improving existing technologies is to introduce external information, such as using data from ground-based augmentation systems (CORS) and other ground-based reference station networks to spatially model ZTD and provide the model results as strong external constraints to independent users. However, this approach introduces more general limitations: 1. It relies on a large and expensive ground infrastructure network, making it unable to provide services to independent users in vast areas without network coverage, such as oceans, aviation, and remote areas; 2. The effectiveness of this approach is limited by the density and geometric distribution of reference stations. When the network is sparse or the terrain is complex, the accuracy and reliability of the spatial model, especially in the vertical direction, will decrease; 3. Shifting the solution problem to a complex external system that requires real-time maintenance does not fundamentally improve the solution capability of the independent receiver itself. Besides the aforementioned approach that relies on external physical network stations, other research attempts to shift... Complex algorithmic models, such as those employing artificial intelligence or deep learning, are used to predict tropospheric delay. However, these methods also have limitations. For example, Chinese invention patent CN118504616A discloses a GNSS tropospheric delay error processing method and system. The technical idea of ​​this scheme is to use deep learning models, such as encoder-decoder models, to train and predict based on grid point data to obtain accurate tropospheric delay prediction values. This approach is essentially still an external correction or prediction mode, which not only relies on acquiring external grid data but also places extremely high demands on the algorithm's computing power and the preprocessing of massive training data. This poses a new burden for independent receivers that require real-time low-power computation. More importantly, this prediction method does not address the mathematical coupling problem of the elevation-zenith delay parameter within the Kalman filter itself.

[0004] Therefore, the technical problem to be solved by this invention is to break away from the existing approach of relying on external space modeling and instead start from the estimation algorithm itself to find a simple and efficient mechanism that can fundamentally break the mathematical coupling of elevation-zenith delay at the solution level by using only the information of the GNSS receiver itself or easily accessible auxiliary information. Summary of the Invention

[0005] To address the problems mentioned in the background art, the technical solution of this invention is as follows: A real-time correction method for GNSS tropospheric delay error, applied to a GNSS receiver containing a Kalman filter, wherein the Kalman filter uses the station elevation parameter and the tropospheric zenith total delay parameter (ZTD) as state parameters to be estimated, and the method includes:

[0006] A first process noise model with a first constraint value and a second process noise model with a second constraint value are preset for the ZTD parameters, wherein the first constraint value is less than the second constraint value;

[0007] Obtain motion state parameters of the GNSS receiver from the Kalman filter; acquire bypass information and calculate atmospheric disturbance characteristic measures of the bypass information;

[0008] Obtain the posterior observation residuals of the Kalman filter and calculate the model mismatch index; when the motion state parameters meet the preset dynamic motion conditions, select the first process noise model to constrain the ZTD parameters in the current epoch; when the motion state parameters do not meet the preset dynamic motion conditions:

[0009] If the atmospheric disturbance characteristic measure is greater than the preset disturbance threshold or the model mismatch index is greater than the preset mismatch threshold, then the second process noise model is used to constrain the ZTD parameters; otherwise, the first process noise model is used to constrain the ZTD parameters.

[0010] Preferably, the bypass information comes from the air pressure readings collected by the air pressure sensor integrated with the GNSS receiver, and the atmospheric disturbance characteristic measure is the standard deviation of the air pressure readings within the sliding time window; the motion state parameter is the vertical velocity of the GNSS receiver, and the preset dynamic motion condition is that the absolute value of the vertical velocity is greater than a preset velocity threshold; the model mismatch index is the root mean square value of the posterior observation residual within another sliding time window, and the preset mismatch threshold is a preset residual threshold.

[0011] Preferably, the method further includes: obtaining the GNSS signal quality characteristics of the GNSS receiver; determining whether multipath pollution exists based on the GNSS signal quality characteristics; when multipath pollution is determined to exist, preventing the model mismatch index from exceeding the preset mismatch threshold in the step of executing when the motion state parameters do not meet the preset dynamic motion conditions, and then selecting the logic execution of the second process noise model.

[0012] Preferably, the GNSS signal quality characteristic is the carrier-to-noise ratio of the GNSS signal; the step of determining whether multipath pollution exists includes calculating the statistical index of the carrier-to-noise ratio and comparing it with a preset pollution threshold.

[0013] Preferably, the first constraint value corresponds to the process noise value that keeps the ZTD parameter stable in the time domain; the second constraint value corresponds to the process noise value that allows the ZTD parameter to track real atmospheric changes. The GNSS receiver is an independent receiver that operates without relying on external differential network assistance.

[0014] A real-time GNSS tropospheric delay error correction device, comprising: a storage module, a Kalman filter, a bypass information module, a processing module, and a constraint control module.

[0015] The storage module is used to preset a first process noise model corresponding to a first constraint value and a second process noise model corresponding to a second constraint value, wherein the first constraint value is less than the second constraint value;

[0016] The Kalman filter is used to take the station elevation parameters and the tropospheric zenith total delay (ZTD) parameters as state parameters to be estimated, and outputs the motion state parameters of the GNSS receiver and the a posteriori observation residuals.

[0017] The bypass information module is used to obtain bypass information;

[0018] The processing module connects the Kalman filter and the bypass information module, and is used to calculate the atmospheric disturbance characteristic measure of the bypass information, as well as the model mismatch index of the posterior observation residuals.

[0019] The constraint control module, connected to the Kalman filter, processing module, and storage module, is used to acquire motion state parameters, atmospheric disturbance characteristic measurements, and model mismatch indices. When the motion state parameters meet the preset dynamic motion conditions, the Kalman filter is controlled to select the first process noise model to constrain the ZTD parameters in the current epoch. When the motion state parameters do not meet the preset dynamic motion conditions: if the atmospheric disturbance characteristic measurement is greater than the preset disturbance threshold or the model mismatch index is greater than the preset mismatch threshold, the Kalman filter is controlled to select the second process noise model to constrain the ZTD parameters; otherwise, the Kalman filter is controlled to select the first process noise model to constrain the ZTD parameters.

[0020] A real-time correction medium for GNSS tropospheric delay error is disclosed. The medium is a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, it is used to implement a real-time correction method for GNSS tropospheric delay error.

[0021] Compared with the prior art, the beneficial effects of the present invention are:

[0022] 1. By utilizing bypass information indicating the physical stability of the atmosphere, statistical disturbance characteristics are used as the basis for logical judgment to dynamically adjust the process noise model of the tropospheric zenith total delay parameter in the Kalman filter of the GNSS receiver. This method avoids the traditional path of using the absolute measurement value of bypass information to calculate the delay parameter. Instead, it acts as a logical switch to control the internal constraints of the algorithm. When the atmosphere is stable, it applies strong constraints of the first process noise model, forcing the elevation parameter and the estimated zenith total delay parameter to be separated in the time domain. This avoids the mathematical coupling between the two in the observation equation, enabling independent user receivers to achieve positioning results, especially rapid convergence of the elevation component, without relying on external differential network assistance.

[0023] 2. The GNSS receiver motion state parameters calculated by the Kalman filter itself are used to arbitrate and cover the judgment logic of the bypass information. When the receiver is in dynamic motion, the bypass information will be distorted by the interference of the carrier motion. At this time, the method no longer relies on the bypass information, but forces the first process noise model to be selected based on the motion state parameters. This not only avoids false decisions introduced by the carrier motion, but also forces the filter to reduce all vertical observation changes to the elevation parameters by strongly constraining the total zenith delay parameter at this time, thus accelerating the tracking of the elevation parameters to the real physical motion and ensuring the correctness and applicability of the core decoupling mechanism in dynamic application scenarios.

[0024] 3. Construct a cross-validation and diagnostic mechanism based on GNSS solution master information such as posterior observation residuals and signal quality characteristics, and bypass auxiliary information such as disturbance characteristics; by comparing the statistical characteristics of posterior observation residuals with the atmospheric conditions indicated by bypass information, autonomously identify whether there is a logical contradiction between the two, and then determine whether the bypass information has a silent failure fault; use signal quality characteristics to determine whether the current high residual is caused by multipath contamination, and then decide whether to suppress the above silent failure diagnosis; this multi-level information verification and logic gating work together enables the system to autonomously identify and avoid interference under real working conditions such as motion contamination, sensor failure, and signal contamination, and can automatically execute the correct constraint strategy under different conditions, improving the convergence speed while ensuring the full-cycle reliability of high-precision positioning results. Attached Figure Description

[0025] Figure 1 This is a system architecture flowchart of the modified method of the present invention;

[0026] Figure 2 This is a performance comparison chart of the smooth transition and hard switching mechanisms of this invention;

[0027] Figure 3 This is a schematic diagram illustrating the application deployment of the modified device of the present invention. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the invention and are not intended to limit the scope of protection of the invention.

[0029] This invention discloses a method, device, and medium for real-time correction of tropospheric delay error in GNSS receiver systems. The system incorporates a Kalman filter, and its core lies in reconstructing the Kalman filter's solution logic to address the long-standing problem of strong mathematical coupling between station elevation parameters and the tropospheric zenith total delay (ZTD) parameter in high-precision positioning. A constraint mechanism for multi-source information collaborative decision-making is established. By acquiring the GNSS receiver's own motion state parameters, bypass information indicating atmospheric physical stability, and the Kalman filter's own model mismatch index in real time, and based on a set of preset logical rules, the system automatically selects either a first-process noise model or a second-process noise model under different operating conditions to apply dynamic time-domain constraints to the ZTD parameter. This breaks the coupling between elevation and ZTD at the algorithmic level, enabling high-reliability, rapid convergence for independent receivers that do not rely on external differential networks. In a specific implementation, the GNSS receiver system, such as an independent receiver for Precise Point Positioning (PPP), incorporates the station elevation parameters in the state vector of the Kalman filter. Using the tropospheric zenith total delay parameter (ZTD) as the state parameter to be estimated, the system presets at least two process noise models for the ZTD parameter to achieve dynamic constraints on the ZTD parameter. One of these models is the first process noise model, corresponding to the first constraint value. Set as a minimum value; taking engineering practice in the field of geodesy as an example, this value can be set to... The first model is used to strongly constrain the ZTD parameters under atmospheric stability or specific operating conditions, ensuring stability in the time domain; the second is the second process noise model, corresponding to the second constraint value. The first constraint value is set to a relatively large value, where the first constraint value is less than the second constraint value. For example, the second constraint value can be set to... This model is designed to allow the ZTD parameters to change rapidly in the time domain to track the real atmospheric conditions in the context of atmospheric dynamics or filter mismatch.

[0030] The core logic of this method lies in establishing a hierarchical, highly adaptable switching mechanism to determine which process noise model to use at each epoch of the filter. This mechanism operates based on the acquisition and processing of the following key information: The system must cope with interference caused by the motion of the GNSS receiver itself. In high-dynamic application scenarios, the vertical motion of the carrier can directly cause drastic changes in bypass sensor readings. The changes introduced by motion are difficult to distinguish from real atmospheric disturbances. To avoid false judgments caused by this motion contamination, the first layer of logic in this method is motion state judgment. The system obtains the motion state parameters of the GNSS receiver from the state output of the Kalman filter itself. In the specific implementation, these motion state parameters are the vertical velocity of the GNSS receiver. The system sets a preset speed threshold. This is used to determine whether the carrier is in a high dynamic state; for example, it can be set to 0.5 m / s, when the vertical velocity... The absolute value is greater than that When the preset dynamic motion conditions are met, the system determines that the bypass information is unreliable, unconditionally rejects subsequent judgments, and forces the use of the first process noise model in the current epoch. Imposing constraints serves a dual purpose: firstly, it avoids interference from motion contamination; secondly, at this point, the vertical variation in the observation equations primarily originates from elevation. The actual motion, and the strong constraints on the ZTD parameters, actually force the filter to reduce all observed changes to elevation parameters. This accelerates the tracking of elevation parameters to real physical motion; secondly, when the motion state parameters do not meet the preset dynamic motion conditions, i.e., the absolute value is not greater than... When the system determines that the carrier is in a quasi-static state, the bypass information is considered reliable, and the logic proceeds to the second level of judgment: determining the true atmospheric state. The system then acquires the bypass information, which originates from the barometric pressure sensor integrated into the GNSS receiver. This method utilizes only the disturbance characteristics of this sensor, rather than its absolute reading. The system collects the barometric pressure readings in real time and calculates the atmospheric disturbance characteristic metric of the bypass information. Specifically, this metric can be the standard deviation of the barometric pressure reading within the sliding time window. This window is, for example, 60 seconds, and the system sets a preset disturbance threshold. This threshold is used to define the stability of the atmospheric physical state; for example, it can be set to 0.1 hPa. If the calculated... Greater than that If the atmosphere is in a dynamic state, the system will select the second process noise model. This allows ZTD parameters to track real atmospheric changes.

[0031] To address the risk of bypass sensor quiescent failure, such as physical blockage causing reading freezes, this method introduces a third-layer judgment based on the health diagnosis of the filter's internal state. This type of failure would cause the sensor to erroneously output a constant low value. The value caused the system to misjudge the atmosphere as stable and become permanently locked. Ultimately, this leads to filter divergence under real atmospheric changes. To avoid this risk, the system obtains the posterior observation residuals of the Kalman filter. The model mismatch index is calculated, which can be the root mean square value of the posterior observation residuals within another sliding time window. The system sets a preset mismatch threshold. This threshold represents the upper tolerance limit for filter model mismatch. Greater than that This indicates that the filter model is severely mismatched. In this case, even if Even if the system indicates stability, it will still determine that the logical contradiction is caused by sensor failure and force the use of the second process noise model. By loosening the constraints on ZTD, high reliability is ensured. In summary, when the dynamic motion conditions are not met, the logical induction of this method is as follows: if the atmospheric disturbance characteristic measurement... Greater than the preset disturbance threshold or model mismatch index Greater than the preset mismatch threshold Then the second process noise model is selected. Constrain ZTD parameters; otherwise, in Not greater than and Not greater than Under the condition that the atmosphere is stable and the sensor is healthy, the first process noise model is selected. Constraining the ZTD parameters, in this case the ZTD parameters are time-domain locked, and high-frequency vertical variations are forced to be reduced to elevation parameters. Thus achieving Decoupling the ZTD parameters in the time domain and accelerating convergence; to further improve the adaptability of this method in typical complex environments such as urban canyons, this invention may also include suppression logic for multipath contamination. In real-world applications, multipath effects can also lead to posterior observation residuals. The surge caused the health diagnostic mechanism to make false judgments, incorrectly interpreting signal contamination as sensor failure, and thus switching to... The GNSS receiver loses its fast convergence capability. To resolve this logical confusion, the GNSS signal quality characteristics of the GNSS receiver are further obtained. Specifically, this involves obtaining the carrier-to-noise ratio of the GNSS signal. It calculates statistical indicators, such as high-frequency jitter or deviation from the nominal model, to determine whether multipath pollution exists. When multipath pollution is determined to exist, the statistical indicator exceeds a preset pollution threshold. At that time, the system determines It is no longer reliable and will prevent the model mismatch index. Greater than the preset mismatch threshold The logic execution of the second process noise model is selected to ensure the effectiveness of the core decoupling mechanism in a multi-path environment.

[0032] Furthermore, to ensure the stability of the highly reliable positioning solution during the initialization phase, this invention may also include two pre-emptive verification steps: First, convergence protection for cold start of the filter. During initialization, the initial estimates of the ZTD parameters have significant uncertainty; if forced... This can lead to convergence to an error value; therefore, this method can also obtain the posterior state variance of the ZTD parameters in the Kalman filter. And determine whether it is less than the preset convergence variance threshold. ,Should The ZTD parameters have converged only if... Only when the condition is met will the execution of the aforementioned complete hierarchical logical judgment be authorized; otherwise, i.e. If the process is determined to be non-convergent, then the second process noise model is unconditionally selected. First, it forces the ZTD parameters into a free search state until they converge; second, it performs integrity checks on gross errors in the initial elevation, such as if the user-input initial elevation is incorrect. If errors of tens of meters or more exist, the entire solution model will fail. Therefore, this method can also obtain the initial elevation of the filter during the initialization phase of the Kalman filter. It also reuses bypass information to obtain the mean absolute pressure indicated by the bypass information. Calculate the initial elevation based on the standard atmospheric model. Corresponding expected air pressure When the mean absolute pressure With desired air pressure The deviation is greater than the preset elevation gross error threshold. At that time, For example, a 50-meter elevation deviation would be considered... If gross errors exist, the execution of the aforementioned complete hierarchical logic judgment will be suppressed. Furthermore, the method can also increase the value of the Kalman filter relative to the station elevation parameter. The corresponding initial variance forces the filter to re-estimate reasonable elevation values ​​from GNSS observation data.

[0033] In another alternative implementation, to eliminate hard switching logic (in To mitigate the potential for location result jitter caused by nearby oscillations, and in steps where motion state parameters do not meet preset dynamic motion conditions, the judgment logic regarding atmospheric disturbance characteristic measurements is replaced with a nonlinear smooth transition mechanism. The system is based on atmospheric disturbance characteristic measurements. Smooth weights between 0 and 1 are calculated using nonlinear mapping functions such as the Sigmoid function or piecewise linear functions. ,Should Within the preset transition zone ( to The ZTD parameter smoothly increases from 0 to 1 in the current epoch of process noise. The calculation formula is: ,in, This is the first constraint value. This is the second constraint value. The value can respond The continuous change between two constraint values ​​is smoothed, thus eliminating algorithm jitter and improving the smoothness of high-precision positioning results. This invention also provides a real-time GNSS tropospheric delay error correction device for implementing any of the aforementioned methods. This device can be integrated into a GNSS receiver and includes: a storage module, a Kalman filter, a bypass information module, a processing module, and a constraint control module. The storage module is used to preset the first process noise model and the second process noise model and their corresponding constraint values. The Kalman filter, as the core solution unit, uses the station elevation parameters and ZTD as the state parameters to be estimated, and outputs internal motion state parameters such as vertical velocity and posterior observation residuals in real time. The bypass information module is used to acquire bypass information, such as integrating a barometric pressure sensor to collect barometric pressure readings. The processing module connects the Kalman filter and the bypass information module to calculate atmospheric disturbance characteristic measures such as... And calculation model mismatch indicators such as The constraint control module connects all other modules and serves as the decision-making core. It is used to acquire motion state parameters, atmospheric disturbance characteristic measurements, and model mismatch indices. It strictly executes the aforementioned hierarchical logic judgment, prioritizing the judgment of dynamic motion conditions. If these conditions are not met, it comprehensively judges the disturbance characteristics and mismatch indices, controlling the Kalman filter to select either the first process noise model or the second process noise model at the current epoch. This invention also provides a real-time correction medium for GNSS tropospheric delay errors. The medium is a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it is used to implement any of the aforementioned real-time correction methods for GNSS tropospheric delay errors.

[0034] Example 1: In a high-precision positioning application scenario, a GNSS receiver employing a correction method, integrating a standard barometric sensor, and mounted on an unmanned aerial vehicle (UAV) for high-precision aerial mapping, performs stop-and-go precision point positioning (PPP) measurement tasks in areas without ground differential network coverage. The UAV is required to briefly hover over multiple mapping points at different altitudes to collect high-precision position data, and then quickly maneuver to the next mapping point. When the UAV hovers over the first mapping point, it is in a quasi-static state, and the motion state parameter output by the Kalman filter, i.e., the vertical velocity... The absolute value is less than the preset speed threshold. At this point, the system determines that the preset dynamic motion conditions are not met, and the logic control is transferred to subsequent steps for judgment. Under this condition, the atmospheric physical state is stable, and the atmospheric disturbance characteristic measure calculated by the bypass information module, namely the standard deviation of air pressure, is... Less than the preset disturbance threshold Meanwhile, the sensor is working normally, and the model mismatch index calculated by the filter is the root mean square value of the posterior observation residuals. Less than the preset mismatch threshold Based on the specific implementation method's judgment logic, the system selects the first process noise model at this time and applies the first constraint value. Strong time-domain constraints are applied to the ZTD parameters, forcing the elevation parameters to... The ZTD parameters are separated from the Kalman filter solution, enabling the independent receiver to obtain the required high-precision converged solution in a short time.

[0035] When the unmanned aerial vehicle completes the data collection at the first point and rapidly climbs to the second surveying point at a higher altitude, a specific working condition occurs in this field: the vehicle's high-speed vertical movement. Absolute value greater than This caused a drastic change in the integrated barometric pressure sensor reading, affecting the readings calculated by the bypass information module. The value then surged, exceeding ,Should The signal morphology is highly similar to that of real severe atmospheric disturbances; in solution methods lacking the multi-source collaborative mechanism of this invention, the system will rely solely on this... The system indicates that the atmosphere has entered a dynamic state, and then switches to the second process noise model, applying a second constraint value. Loosening the constraints on ZTD led to The re-coupling of ZTD parameters during the solution process suppresses the convergence capability of high-precision positioning during maneuvers; based on the multi-source cooperative mechanism, the Kalman filter outputs high... While providing the value, it also outputs high-precision values ​​that meet the preset dynamic motion conditions. At this point, the preset judgment logic regarding motion state parameters is executed first, the system determines that the carrier is in a high-dynamic maneuvering state, and identifies that at this time... The indicated atmospheric conditions are unreliable; the system then determines, based on this motion state, the first process noise model is selected, and applied... Constraining ZTD parameters avoids erroneous constraint switching caused by carrier motion contamination, by determining the carrier is in a high dynamic state through the filter. Absolute value greater than When the ZTD parameters are locked, the filter is forced to reduce all observed vertical signal variations to elevation parameters. This accelerates the Kalman filter's tracking of the carrier's physical height.

[0036] Example 2: This example provides a set of comparative experimental data to objectively verify the technical effectiveness of the multi-source information collaborative constraint method in the specific implementation under typical high-precision positioning conditions, compared with the standard solution method of the prior art, in terms of convergence speed and high dynamic adaptability. The experimental platform is built in a semi-physical simulation environment composed of a GNSS signal simulator and a high-precision receiver hardware with an integrated barometric sensor. This simulator is used to simulate the satellite constellation, orbit, clock bias, and controllable tropospheric delay time series required for high-precision positioning. Three different solution algorithms are deployed on the receiver hardware, and the convergence criterion for high-precision positioning is set as the elevation component. The root mean square value of the three-dimensional position error is consistently less than 10 cm, and the time required to reach this standard is recorded as the convergence time.

[0037] The experiment set up three solution schemes: 1. Control group A (fixed noise): The standard PPP solution method was used, and the Kalman filter was set to a fixed, moderate process noise value for the ZTD parameter; 2. Control group B (bypass information only): A simplified adaptive method was used, based on the characteristic measurement of atmospheric disturbance. The judgment logic, that is and To switch between comparisons and However, this group does not have motion state parameters. Arbitration logic; 3. Sample of this invention: adopts a complete multi-source information collaborative judgment method, and simultaneously integrates motion state parameters. Atmospheric disturbance characteristic measurement and model mismatch index The three-level hierarchical judgment logic; three typical working conditions were designed for the experiment, and the three solution schemes were run under the same working conditions: Working condition 1 (static stability): The GNSS receiver remains absolutely stationary, and the simulator simulates a stable, low-disturbance tropospheric environment. consistently less than Condition 2 (Static Disturbance): The GNSS receiver remains absolutely stationary, but 5 minutes after the calculation begins, the simulator simulates a rapid frontal event, causing a 60mm step change in ZTD. Under this condition... It will be greater than briefly Condition 3 (Dynamically Stable): The simulator simulates a stable, low-disturbance tropospheric environment, i.e. Always less than However, two minutes after the calculation began, the GNSS receiver started to climb vertically at a speed of 2.0 m / s. Under this condition, the receiver's own movement caused... The test data for the three solution schemes under the above three working conditions are shown in Table 1.

[0038] Table 1: Performance Comparison of Three Solution Schemes under Different Operating Conditions

[0039]

[0040] The data analysis is as follows: In operating condition one (static stability), the fixed noise of control group A leads to... The decoupling from the ZTD parameters was slow, with a convergence time of 21.5 minutes; both the control group B and the sample group of this invention were able to correctly identify stable bypass information. The first process noise model is selected. Applying strong temporal constraints shortened the convergence time to approximately 5 minutes, verifying the effectiveness of the bypass information-assisted mechanism in static fast convergence. In condition two (static disturbance), the fixed noise of control group A could not adapt to the step change of ZTD, leading to model mismatch, affecting convergence, and degrading the accuracy to 12.6 cm after stabilization. Both control group B and the sample group of this invention passed the test. Greater than Correctly identify atmospheric disturbances and switch to the second process noise model. To track ZTD changes, the convergence time was extended to 7.6-7.8 minutes, but ultimately maintained a high accuracy on the order of 4.cm, verifying... The role of monitoring in maintaining high reliability; in operating condition three (dynamic stability), key differences in high-dynamic applications are presented: control group A still converges slowly; control group B, due to the lack of motion state judgment, moves the receiver vertically ( Bypass information caused by ) Pollution was incorrectly identified as atmospheric disturbance. ), incorrectly switched to This leads to the removal of the ZTD constraint. Recoupling from ZTD resulted in a convergence time degrading to 23.8 minutes, losing its fast convergence capability; at this point, the sample group of this invention, based on cooperative logic, had motion state parameters... Judgment ( (Highest priority, determined by the system) The signal was contaminated by motion, therefore arbitration and rejection were carried out. The judgment forces the use of the first process noise model. This forces the filter to reduce all vertical variations to elevation. It maintains rapid convergence, with a convergence time of only 5.4 minutes.

[0041] Example 3: This example combines Figures 1 to 3 This document describes a method, equipment, and medium for real-time correction of GNSS tropospheric delay errors, such as... Figure 1As shown, the core solution closed loop is illustrated. The GNSS satellite signal is input to the Kalman filter solution unit. This unit outputs high-precision positioning solution results to the user / system on one hand, and outputs the posterior observation residuals, motion state parameters, GNSS signal quality characteristics, and posterior state variance to the key decision flow on the other hand. Bypass information modules, such as barometric pressure sensors, collect barometric pressure readings and send them to the processing module. The processing module also receives the posterior observation residuals from the Kalman filter and calculates atmospheric disturbance characteristic measures and model mismatch indices. These indices are sent to the constraint control module, which obtains preset thresholds and constraint values ​​from the B1 preset model and threshold storage module. It then performs collaborative decision-making by combining multi-source information from the Kalman filter. Finally, the constraint control module outputs the selected process noise model and feeds it back to the Kalman filter solution unit to adjust its constraint conditions.

[0042] like Figure 2 As shown in the figure, the horizontal axis represents time (in minutes), and the vertical axis on the left represents the ZTD process noise. / epoch, with the right vertical axis representing the standard deviation of air pressure. (hPa), as shown in the figure, when the standard deviation of air pressure As time gradually increases, the process noise using a hard-switching mechanism will experience a step change at a specific threshold point, around the 9th minute, while the process noise using a smooth transition mechanism will decrease as time progresses. The change smoothly increases from the first constraint value to the second constraint value within the transition zone; such as... Figure 3 As shown, GNSS satellites serve as external signal sources, and the signals are received by the GNSS receiver hardware on the application carrier, such as an unmanned aerial vehicle. Within this hardware, an integrated barometric pressure sensor bypass information module sends the barometric pressure readings to the processing module, while the Kalman filter core calculation unit also sends its motion state parameters and posterior observation residuals to the processing module. After calculating the corresponding indicators, the processing module submits them to the constraint control module for multi-source information collaborative decision-making. The constraint control module reads the preset model and threshold from the storage module and feeds back the final constraint decision to the Kalman filter. This entire logic runs as a software system deployed within the receiver as a real-time GNSS tropospheric delay error correction device, ultimately outputting highly reliable positioning results to the user terminal, such as a flight control system or mapping software.

[0043] Example 4: This example provides a standardized engineering calibration procedure for determining multiple preset thresholds. This procedure is used in high-precision positioning applications to quantify and calibrate the core parameters of the multi-source information collaborative judgment logic before deploying a GNSS receiver on a specific carrier or operating environment. This eliminates reliance on empirical settings, ensuring the algorithm's adaptability and high-reliability positioning performance, specifically targeting preset speed thresholds. To determine the optimal method, the GNSS receiver system, including the integrated barometric sensor, will be mounted on the final operational platform. This platform will be positioned at a high-precision positioning reference point. Under conditions of no multipath interference and stable atmospheric conditions, at least two hours of purely static observation will be conducted. During this period, the system will acquire the vertical velocity output from the Kalman filter. The time series, which should ideally be zero, but includes measurement noise from the solution and the sensor itself; the system calculates this... The statistical properties of a sequence, specifically its standard deviation. To avoid false dynamic decisions caused by computational noise in high-reliability positioning applications, a preset velocity threshold is used. Set as a predetermined multiple of the standard deviation, such as Taking a specific calibration as an example, if the value measured in this static test is... If it is 0.08 m / s, then It can be determined to be 0.4 m / s. Any vertical velocity greater than this value will be judged as the actual movement of the carrier by the highly reliable positioning system.

[0044] Secondly, calibrate the preset disturbance threshold. This calibration utilizes historical meteorological observation data of the application area. From this historical data, multiple periods known as atmospherically stable, such as prolonged clear weather, and multiple periods known as periods of severe atmospheric disturbance, such as clear frontal passages or thunderstorm events, are selected. For the pressure reading sequences of these two types of periods, the same algorithm as the bypass information module is used to calculate the pressure standard deviation within the same sliding time window, such as 60 seconds. Time series; analysis of these two types The statistical distribution of the sequence will Set as a boundary value that can effectively distinguish between these two distributions; taking a specific calibration as an example, if the analysis shows that the application area is in a stable period... There is a 99% probability that the value will be lower than 0.05 hPa, while the disturbance period... There is a 95% probability that the value is higher than 0.2 hPa. It can be set between the two, such as 0.1 hPa, as a basis for distinguishing between atmospheric stability and dynamics under high-reliability positioning conditions.

[0045] Secondly, regarding the preset mismatch threshold Compared with the preset pollution threshold Correlation calibration was performed by deploying the receiver in an open environment with good signal quality (i.e., the baseline operating condition), ensuring the bypass information module was functioning properly, and collecting and calculating the root mean square value of the posterior observation residuals of the Kalman filter under high-reliability positioning health conditions. The baseline distribution will The statistical high points of this baseline distribution are set as the mean plus three times the standard deviation, and are used in high-reliability positioning systems to determine whether the model is mismatched; for calibration. The receiver is placed in a high-precision positioning environment with strong multipath propagation, such as an urban canyon or near a reflector, to collect the GNSS signal quality characteristics, i.e., the carrier-to-noise ratio. And calculate its statistical indicators. ,Should The reproducible calculation method is as follows: The standard deviation of the first difference of the time series within a one-second window is recorded. The value is the value observed by the high-reliability positioning system at this time. The low point of the statistical distribution is set as Used in high-reliability positioning systems to determine high Is it caused by multi-pathway contamination?

[0046] Example 5: This example discloses a preset convergence variance threshold. The standardized calibration procedure is performed before the deployment of a high-precision positioning system. First, the GNSS receiver is placed on a reference station with known precise coordinates. The Kalman filter is then run continuously for 24 hours under stable atmospheric conditions to allow for full convergence. Then, the system collects the posterior state variance of the output ZTD parameters during the latter half of the convergence phase, such as the last 12 hours. For the time series, calculate the upper quantile of the series, such as the 99th percentile or its statistical mean, and determine the calculated statistical value as the preset convergence variance threshold. This is used in high-reliability positioning systems to subsequently determine whether ZTD parameters have entered a convergence state; this embodiment discloses a preset elevation gross error threshold. An engineering determination method is used to quantify the uncertainty of the standard atmospheric model employed in the application. The steps include collecting a sample set of geodetic control points, containing multiple (e.g., more than 100) points with known precise elevations. Average absolute pressure measured in the field under different meteorological conditions For each data pair in the sample set, its value is calculated based on the standard atmospheric model. Corresponding expected air pressure And obtain the deviation between the two: The system analyzes all deviations. The statistical distribution of the values ​​is used, and the high-order statistical value of this distribution, such as the mean plus three standard deviations or the 99.9th percentile, is set as the preset elevation gross error threshold. Used for effective identification during the high-reliability positioning initialization phase. To identify gross errors and avoid false judgments caused by model errors.

[0047] Example 6: This example discloses a method for determining the first constraint value. With the second constraint value The optimal values ​​are determined, and a systematic procedure for further calibrating the parameters of the nonlinear mapping function is established. This procedure is executed based on measured data before the deployment of the high-precision positioning system to complete the tuning of the core parameters of the algorithm; to determine The system collects at least 48 hours of static observation data from the base station, covering known periods of atmospheric stability. High-precision post-processing software is used to calculate the ZTD parameter sequence for this period, and the random walk characteristics of the sequence are analyzed. The process noise value corresponding to the random walk K-walk coefficient is set as the first constraint value. This ensures that the constraint value can both strongly constrain the ZTD parameters to accelerate the convergence of high-precision positioning and reflect the physical fluctuation characteristics of the stable atmosphere itself in the environment; to determine The system analyzes historical data containing severe atmospheric disturbances, such as ZTD sequences of frontal passages, to obtain the maximum change within a single epoch. , the second constraint value Set as the process noise value corresponding to the square of the maximum change.

[0048] For the nonlinear mapping function in the specific implementation, this embodiment discloses a parameter calibration procedure, which reuses the calibration procedure in embodiment 4. The two categories used are stable atmospheric conditions and violent atmospheric disturbances. Statistical distribution data; lower boundary of the transition zone Set as the 99.9% quantile of the atmospheric stable distribution to ensure that when At that time, smoothing weight Set the upper boundary of the transition zone to 0; Set as the 0.1% quantile of the distribution of severe atmospheric disturbances to ensure that when At that time, smoothing weight =1; in and Between, nonlinear mapping function A piecewise linear function is used to achieve a balance between computational resources and solution smoothness in high-precision positioning systems.

[0049] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0050] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A real-time correction method for tropospheric delay error in GNSS receivers, applied to a GNSS receiver including a Kalman filter, wherein the Kalman filter uses the station elevation parameter and the tropospheric zenith total delay parameter (ZTD) as state parameters to be estimated, characterized in that the method... include: A first process noise model with a first constraint value and a second process noise model with a second constraint value are preset for the ZTD parameters, wherein the first constraint value is less than the second constraint value; Obtain motion state parameters of the GNSS receiver from the Kalman filter; Obtain bypass information and calculate the atmospheric disturbance characteristic measure of the bypass information; Obtain the posterior observation residuals of the Kalman filter and calculate the model mismatch index; when the motion state parameters meet the preset dynamic motion conditions, select the first process noise model to constrain the ZTD parameters in the current epoch; when the motion state parameters do not meet the preset dynamic motion conditions: If the atmospheric disturbance characteristic measure is greater than the preset disturbance threshold or the model mismatch index is greater than the preset mismatch threshold, then the second process noise model is used to constrain the ZTD parameters; otherwise, the first process noise model is used to constrain the ZTD parameters.

2. The method for real-time correction of GNSS tropospheric delay error according to claim 1, characterized in that, The bypass information comes from the barometric pressure readings collected by the barometric pressure sensor integrated with the GNSS receiver, and the atmospheric disturbance characteristic is measured by the standard deviation of the barometric pressure readings within the sliding time window. The motion state parameter is the vertical velocity of the GNSS receiver, and the preset dynamic motion condition is that the absolute value of the vertical velocity is greater than the preset velocity threshold; the model mismatch index is the root mean square value of the posterior observation residual in another sliding time window, and the preset mismatch threshold is the preset residual threshold.

3. The method for real-time correction of GNSS tropospheric delay error according to claim 1, characterized in that, The method also includes: obtaining the GNSS signal quality characteristics of the GNSS receiver; determining whether multipath pollution exists based on the GNSS signal quality characteristics; and when multipath pollution is determined to exist, preventing the model mismatch index from exceeding the preset mismatch threshold in the step of executing when the motion state parameters do not meet the preset dynamic motion conditions, and then selecting the logic execution of the second process noise model.

4. A real-time correction method for GNSS tropospheric delay error according to claim 3, characterized in that, The quality characteristic of GNSS signals is the carrier-to-noise ratio (CNR) of the GNSS signal; the steps to determine whether multipath pollution exists include calculating the statistical index of the CNR and comparing it with a preset pollution threshold.

5. A real-time correction method for GNSS tropospheric delay error according to claim 1, characterized in that, The first constraint value corresponds to the process noise value that keeps the ZTD parameter stable in the time domain; the second constraint value corresponds to the process noise value that allows the ZTD parameter to track real atmospheric changes.

6. A GNSS tropospheric delay error real-time correction device, used to implement the GNSS tropospheric delay error real-time correction method according to claim 1, characterized in that, The equipment includes: a storage module, a Kalman filter, a bypass information module, a processing module, and a constraint control module. The storage module is used to preset a first process noise model corresponding to a first constraint value and a second process noise model corresponding to a second constraint value, wherein the first constraint value is less than the second constraint value; The Kalman filter is used to take the station elevation parameters and the tropospheric zenith total delay (ZTD) parameters as state parameters to be estimated, and outputs the motion state parameters of the GNSS receiver and the a posteriori observation residuals. The bypass information module is used to obtain bypass information; The processing module connects the Kalman filter and the bypass information module, and is used to calculate the atmospheric disturbance characteristic measure of the bypass information, as well as the model mismatch index of the posterior observation residuals. The constraint control module, connected to the Kalman filter, processing module, and storage module, is used to acquire motion state parameters, atmospheric disturbance characteristic measurements, and model mismatch indices. When the motion state parameters meet the preset dynamic motion conditions, the Kalman filter is controlled to select the first process noise model to constrain the ZTD parameters in the current epoch. When the motion state parameters do not meet the preset dynamic motion conditions: if the atmospheric disturbance characteristic measurement is greater than the preset disturbance threshold or the model mismatch index is greater than the preset mismatch threshold, the Kalman filter is controlled to select the second process noise model to constrain the ZTD parameters; otherwise, the Kalman filter is controlled to select the first process noise model to constrain the ZTD parameters.

7. A GNSS tropospheric delay error real-time correction medium, characterized in that, The medium is a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it is used to implement the GNSS tropospheric delay error real-time correction method as described in claim 1.

Images