A method, equipment, medium, and product for satellite observation data quality control

Abstract

This application discloses a method, device, medium, and product for satellite observation data quality control, relating to the field of satellite navigation and positioning and observation data processing technology. The method includes: acquiring multi-band navigation satellite observation data; extracting signal features using a sliding window based on the multi-band navigation satellite observation data to construct a multivariate vectorized feature sequence; performing unsupervised clustering of the multivariate vectorized feature sequence using a variational Bayesian-Gaussian mixture model to obtain line-of-sight signal distribution parameter features; calculating the Mahalanobis distance to dispersion ratio and performing a multi-frequency dispersion consistency test based on the line-of-sight signal distribution parameter features to identify abnormal observation components; constructing an adaptive weighting function based on the consistency test results; performing weight reduction or removal processing on abnormal observation components based on the adaptive weighting function; and updating the observation noise variance of the positioning filter to obtain the navigation and positioning solution. This application can improve the accuracy and stability of satellite positioning in complex environments.

Description

Technical Field

[0001] This application relates to the field of satellite navigation positioning and observation data processing technology, and in particular to a satellite observation data quality control method, equipment, medium and product. Background Technology

[0002] In complex scenarios such as urban canyons and obstructed environments, satellite navigation and positioning systems are susceptible to the effects of non-line-of-sight (NLOS) signals, multipath effects, and radio frequency interference on observation data. This results in observation errors exhibiting a non-Gaussian, multimodal distribution, which severely reduces positioning accuracy and reliability.

[0003] Traditional satellite observation data quality control methods mostly employ single-threshold judgment methods or supervised learning classification methods. Single-threshold judgment methods struggle to adapt to complex noise distributions, while supervised learning classification methods rely on a large number of labeled samples, limiting their generalization ability. Probabilistic methods based on Gaussian Mixture Models (GMMs) can characterize multimodal error distributions, but traditional expectation-maximization (EM) algorithms suffer from initial value sensitivity, susceptibility to local optima, and the need for manual setting of the number of components. Existing variational Bayesian (VB) methods are mostly based on single-feature modeling, lacking joint constraints from multiple features, resulting in insufficient robustness in anomaly detection and ultimately leading to low satellite positioning accuracy and poor stability.

[0004] Therefore, there is an urgent need for a satellite observation data quality control method that does not require labeled data, can adaptively model, fuse multiple features, and conducts joint verification with multiple frequencies, in order to improve the accuracy and stability of satellite positioning in complex environments. Summary of the Invention

[0005] The purpose of this application is to provide a satellite observation data quality control method, equipment, medium, and product that can improve the accuracy and stability of satellite positioning in complex environments.

[0006] To achieve the above objectives, this application provides the following solution.

[0007] In a first aspect, this application provides a satellite observation data quality control method, which includes the following steps.

[0008] Acquire observation data from multi-band navigation satellites.

[0009] Based on the observation data of the multi-band navigation satellites, a sliding window is used to extract signal features and construct a multi-vectorized feature sequence.

[0010] The variational Bayesian-Gaussian mixture model is used to perform unsupervised clustering on the multivariate vectorized feature sequence to obtain the line-of-sight signal distribution parameter features.

[0011] Based on the characteristics of the line-of-sight signal distribution parameters, the Mahalanobis distance to dispersion ratio is calculated and a multi-frequency dispersion consistency test is performed. Abnormal observation components are then identified based on the consistency test results.

[0012] An adaptive weighting function is constructed based on the consistency test results. The abnormal observation components are then weighted or removed based on the adaptive weighting function, and the observation noise variance of the positioning filter is updated to obtain the navigation and positioning solution.

[0013] In a second aspect, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the satellite observation data quality control method described in any one of the above.

[0014] Thirdly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the satellite observation data quality control method described in any one of the above.

[0015] Fourthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the satellite observation data quality control method described above.

[0016] According to the specific embodiments provided in this application, this application has the following technical effects.

[0017] This application provides a satellite observation data quality control method, device, medium, and product. It utilizes multi-band navigation satellite observation data combined with a sliding window to construct a multi-dimensional vectorized feature sequence, which can comprehensively characterize satellite observation signal features, improve the completeness and reliability of feature expression, and ensure the accuracy and effectiveness of satellite observation data quality control. Unsupervised clustering is achieved through a variational Bayesian-Gaussian mixture model, which can adaptively acquire line-of-sight signal distribution parameters without requiring manual model parameter setting, thus improving model adaptability. Abnormal observation components are identified based on Mahalanobis distance, dispersion ratio, and multi-frequency dispersion consistency test, which improves the accuracy of anomaly identification and reduces the probability of single-frequency misjudgment, thereby improving the accuracy and stability of satellite positioning in complex environments. An adaptive weighting function is used to reduce or eliminate abnormal observation components and dynamically update the observation noise variance, effectively weakening the interference of abnormal observations on positioning solutions, thereby improving the accuracy and stability of navigation and positioning results. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 An application environment diagram of a satellite observation data quality control method provided in an embodiment of this application; Figure 2 A flowchart illustrating a satellite observation data quality control method provided in an embodiment of this application; Figure 3 A schematic diagram illustrating the quality control principle of a satellite observation data quality control method provided in one embodiment of this application; Figure 4 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0020] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] The purpose of this application is to provide a satellite observation data quality control method, device, medium, and product that can achieve unsupervised, adaptive, and multi-feature fusion anomaly identification and suppression, improve the accuracy and stability of satellite positioning in complex environments, and solve the problems of poor robustness, insufficient adaptive ability, and reliance on prior information in traditional methods for observation data quality control under complex environmental interference, resulting in low satellite positioning accuracy and poor stability.

[0022] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0023] The satellite observation data quality control method provided in this application embodiment can be applied to, for example... Figure 1In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be set up independently, integrated into server 104, or placed in the cloud or on another server. Terminal 102 can send multi-band navigation satellite observation data to server 104. After receiving the multi-band navigation satellite observation data, server 104 extracts signal features using a sliding window based on the data, constructing a multi-vectorized feature sequence. It then performs unsupervised clustering on the multi-vectorized feature sequence using a Variational Bayesian Gaussian Mixture Model (VB-GMM) to obtain line-of-sight (LOS) signal distribution parameter features. Based on these LOS signal distribution parameter features, it calculates the Mahalanobis distance to dispersion ratio and performs a multi-frequency dispersion consistency check to identify anomalous observation components. An adaptive weighting function is constructed based on the consistency check results. Anomalous observation components are then weighted or removed using this function, and the observation noise variance of the positioning filter is updated to obtain the navigation and positioning solution. Server 104 can then feed back the obtained navigation and positioning solution to terminal 102. In addition, in some embodiments, the satellite observation data quality control method can also be implemented by the server 104 or the terminal 102 separately. For example, the terminal 102 can directly perform quality control processing on the multi-band navigation satellite observation data, or the server 104 can obtain the multi-band navigation satellite observation data from the data storage system and perform quality control processing on the multi-band navigation satellite observation data.

[0024] The terminal 102 can be, but is not limited to, various desktop computers, laptops, smartphones, tablets, and IoT devices. The server 104 can be implemented using a standalone server or a server cluster consisting of multiple servers, or it can be a cloud server.

[0025] In one exemplary embodiment, such as Figure 2 As shown, a satellite observation data quality control method is provided. This method is executed by computer equipment, specifically by a terminal or server alone, or by both a terminal and a server. In this embodiment, the method is applied to... Figure 1 Taking server 104 as an example, the following steps are included.

[0026] S1: Acquire observation data from multi-band navigation satellites.

[0027] S2: Based on the observation data of the multi-band navigation satellites, a sliding window is used to extract signal features and construct a multi-vectorized feature sequence.

[0028] S3: The variational Bayesian-Gaussian mixture model is used to perform unsupervised clustering on the multivariate vectorized feature sequence to obtain the line-of-sight signal distribution parameter features.

[0029] S4: Based on the characteristics of the line-of-sight signal distribution parameters, calculate the Mahalanobis distance to dispersion ratio and perform a multi-frequency dispersion consistency test, and identify abnormal observation components based on the consistency test results.

[0030] S5: Construct an adaptive weighting function based on the consistency test results, reduce or eliminate the abnormal observation components based on the adaptive weighting function, and update the observation noise variance of the positioning filter to obtain the navigation and positioning solution results.

[0031] By implementing steps S1 to S5 above, multivariate feature fusion and unsupervised variational Bayesian modeling are adopted, which can adaptively represent complex error distributions without the need for labeled data. The accuracy of anomaly identification is improved by using Mahalanobis distance, dispersion ratio and multi-frequency dispersion consistency test. By dynamically updating adaptive weights and observation noise, non-line-of-sight and multipath interference are effectively suppressed, and the continuity, accuracy and stability of positioning in complex scenarios are improved. This solves the problems of poor robustness of observation data quality control, insufficient adaptive ability and reliance on prior information in traditional methods under complex environmental interference, which leads to low satellite positioning accuracy and poor stability.

[0032] As an optional implementation, in step S1, the multi-band navigation satellite observation data includes observations of pseudorange, carrier phase, and signal-to-noise ratio (SNR) for each frequency band of each navigation satellite.

[0033] As an optional implementation, in step S2, based on the multi-band navigation satellite observation data, a sliding window is used to extract signal features and construct a multi-vectorized feature sequence, which specifically includes the following steps.

[0034] S21: Based on the multi-band navigation satellite observation data, extract the satellite pseudorange observation residual, satellite elevation angle, mean signal-to-noise ratio, and standard deviation of signal-to-noise ratio within the sliding window of each frequency band.

[0035] S22: Construct a multivariate vectorized feature sequence based on the satellite pseudorange observation residuals, satellite elevation angles, mean signal-to-noise ratios, and standard deviations of signal-to-noise ratios within the sliding window of each frequency band.

[0036] By acquiring observation data from multi-band navigation satellites and combining it with sliding window extraction of observation residuals, satellite elevation angles, mean signal-to-noise ratio (SNR), and standard deviation of SNR, a multi-dimensional vectorized feature sequence is constructed to comprehensively characterize the signal's geometry, strength, and stability information.

[0037] As an optional implementation, in step S3, an unsupervised clustering of the multivariate vectorized feature sequence is performed using a variational Bayesian-Gaussian mixture model to obtain the line-of-sight signal distribution parameter features, specifically including the following steps.

[0038] S31: Input the multivariate vectorized feature sequence into the variational Bayesian Gaussian mixture model, perform variational inference using the Evidence Lower Bound (ELBO), and adaptively determine the model complexity, i.e., adaptively determine the number of Gaussian mixture components, to obtain the approximate posterior distribution of the model parameters.

[0039] S32: Based on the approximate posterior distribution of the model parameters, select the Gaussian component with the largest mixing weight as the line-of-sight signal cluster.

[0040] S33: Obtain the mean vector and covariance matrix corresponding to the line-of-sight signal cluster, and use them as the distribution parameter features of the line-of-sight signal.

[0041] As an optional implementation, in step S4, based on the characteristics of the line-of-sight signal distribution parameters, the Mahalanobis distance to dispersion ratio is calculated and a multi-frequency dispersion consistency test is performed. Abnormal observation components are identified based on the consistency test results. This specifically includes the following steps.

[0042] S41: Based on the line-of-sight signal distribution parameter characteristics, calculate the square of the Mahalanobis distance by which the current multivariate vectorized feature deviates from the LOS distribution.

[0043] S42: Calculate the dispersion ratio of each frequency band based on the square of the Mahalanobis distance and the preset chi-square threshold.

[0044] S43: Perform a multi-frequency dispersion consistency test based on the dispersion ratio of each frequency band to obtain the consistency test result.

[0045] S44: The observation component in the consistency test result that is determined to be abnormal in multi-band consistency is taken as the abnormal observation component.

[0046] Here, "observation component" refers to satellite observation data from a single satellite, at a single epoch, and in a single frequency band. "Abnormal observation component" refers to observation data from a single satellite or in a single frequency band that, in satellite navigation and positioning, is significantly affected by non-line-of-sight signals, multipath effects, noise interference, etc., causing the observed values ​​to deviate significantly from the normal line-of-sight signal distribution and severely reducing positioning accuracy. After calculating the dispersion ratio of each frequency band, the average dispersion ratio is further calculated. If the average dispersion ratio of an observation component is less than or equal to 1, the observation component is determined to be a normal observation component; if the average dispersion ratio of an observation component is greater than 1, the observation component is determined to be an abnormal observation component.

[0047] Furthermore, after distinguishing between normal and abnormal observation components, normal observation components retain their full weight and are not removed or downweighted. For abnormal observation components, if the average dispersion ratio of an abnormal observation component is greater than 1 and less than or equal to the threshold for rejecting severe outliers, the abnormal observation component is determined to be a mild outlier and its weight is downweighted; if the average dispersion ratio of an abnormal observation component is greater than the threshold for rejecting severe outliers, the abnormal observation component is determined to be a severe outlier and it is directly removed.

[0048] By calculating the square of the Mahalanobis distance of the current observed feature relative to the LOS distribution, it is converted into a dispersion ratio. Single-frequency anomaly detection is then performed based on a preset chi-square threshold. Multi-frequency consistency verification enhances the reliability of anomaly identification and reduces the false positive rate. Through Mahalanobis distance calculation, dispersion ratio determination, and multi-frequency dispersion consistency verification, anomalous observation components deviating from the normal line-of-sight signal distribution or affected by non-line-of-sight or multipath interference can be identified. These components are then further eliminated or downweighted based on the weighting factor of their adaptive weighting function, achieving satellite observation data quality control and ensuring the accuracy and stability of satellite positioning in complex environments.

[0049] As an optional implementation, in step S5, the adaptive weighting function is a piecewise parameterized function, which assigns full weight to normal observation components based on the average dispersion ratio, and reduces or eliminates the weight of abnormal observation components (i.e., removes them); the updated observation noise variance is inversely proportional to the weight.

[0050] As an optional implementation, in step S5, the abnormal observation components are deweighted or removed based on the adaptive weighting function, specifically including the following situations.

[0051] (1) When the weighting factor in the adaptive weighting function is equal to 0 (i.e. When abnormal observation components are detected, they are removed.

[0052] (2) When the weighting factor in the adaptive weighting function is greater than 0 and less than 1 (i.e. When an anomaly observation component is detected, its weight is reduced.

[0053] (3) When the weighting factor in the adaptive weighting function is equal to 1 (i.e. When the observed component is normal, it is not removed or downweighted.

[0054] By constructing a piecewise adaptive weighting function based on the average dispersion ratio, the normal observation components ( Maintain full weighting, and treat mildly anomalous components ( Reduce the weight of severely anomalous components () Directly remove; update the observation noise variance inversely proportional to the weight to achieve dynamic quality control in the filtering process.

[0055] As an optional implementation, in step S5, after obtaining the updated observation noise variance of the updated positioning filter, the observation data after weight reduction or elimination processing and the updated observation noise variance are input into the navigation positioning filter module, and the navigation positioning filter algorithm is used to perform filter state estimation and solution to obtain the navigation positioning solution result.

[0056] The navigation and positioning filtering module refers to the filtering processing unit that completes state estimation based on satellite observation data. It usually uses Kalman filtering (or extended Kalman filtering) to calculate the navigation and positioning solution based on the quality-controlled observation values ​​and the updated observation noise variance.

[0057] The navigation and positioning solution results include the receiver's three-dimensional position coordinates, receiver clock bias, and carrier phase ambiguity. These results can be directly used for high-precision satellite navigation positioning services, providing continuous, reliable, and high-precision position and time services for scenarios such as vehicle navigation, handheld terminals, UAV positioning, mechanical control, and surveying, ensuring the stability and accuracy of satellite navigation and positioning in complex environments.

[0058] In an exemplary embodiment, to illustrate the technical solution provided by the embodiments of this application in detail, a satellite observation data quality control method is provided, the quality control principle of which is as follows: Figure 3 As shown, the specific steps include:

[0059] Step 1: Acquire multi-frequency navigation satellite observation data, extract the signal features of the multi-frequency navigation satellite observation data, and construct multi-vectorized feature sequences of each satellite using a sliding window.

[0060] In this embodiment of the application, when acquiring multi-frequency navigation satellite observation data, the acquisition... Time Frequency The satellite observations, combined with preprocessing and filtering, yielded the epochs. The state vector prior estimate is used for multi-source feature extraction: on the one hand, the pseudorange residuals of multiple frequency bands are calculated; on the other hand, the signal-to-noise ratio observations of multiple frequency bands are extracted, and the local mean and standard deviation of the signal-to-noise ratio in a short period of time are calculated; at the same time, the satellite elevation angle is extracted.

[0061] In this embodiment, satellite observation data (such as pseudorange, carrier phase, and signal-to-noise ratio) from multiple frequency bands of various GNSS systems (e.g., GPS: L1, L2, and L5 bands; Galileo: E1, E5a, E5b, and E6 bands; BDS: B1I, B1C, B2a, B2b, and B3I bands) are acquired. This data is used to construct multi-dimensional vectorized features for each satellite, including: determining the satellite pseudorange observation residual, satellite elevation angle, mean signal-to-noise ratio (the local mean of the signal-to-noise ratio over a short time), and standard deviation of the signal-to-noise ratio within the sliding window of each frequency band; and constructing frequency... , calendar Next satellite Multi-vectorization features , expressed as the following formula.

[0062] (1).

[0063] in, Represents pseudorange observations. This represents the difference between the observed and predicted pseudorange values, i.e., the pseudorange residual. Indicates the satellite's elevation angle; Indicates the mean signal-to-noise ratio; The standard deviation of the signal-to-noise ratio is represented by T, and T represents the transpose.

[0064] Step 2: Perform unsupervised probabilistic clustering on the multivariate vectorized feature sequence obtained in Step 1 based on the variational Bayesian Gaussian mixture model, automatically determine the model complexity, and obtain the line-of-sight signal distribution parameter features and non-line-of-sight signal distribution parameter features.

[0065] In this embodiment, multivariate vectorized features are input into a variational Bayesian Gaussian mixture model. Automatic inference is performed based on maximizing the lower bound of evidence. Through adaptive clustering, the LOS signal cluster with the largest mixture weight is identified, and the mean vector of this LOS distribution is obtained. With the precision matrix (i.e., the inverse of the covariance matrix) ).

[0066] Specifically, the first step is to set the characteristics of the observed epochs. The distribution model is a mixture model of multiple Gaussian components, expressed as follows.

[0067] (2).

[0068] in, Represents the total number of observed epochs. This represents the total number of Gaussian components. Indicates the first The mixing coefficients of each component satisfy the following conditions: , and These represent the mean vector and precision matrix of the component, respectively. Represents the set of features of the observed epoch. In the mixing coefficient Mean vector And precision matrix The joint likelihood function under this Gaussian mixture model is the total probability density of all observed data. Indicates the first eigenvectors Belongs to the The conditional probability density function of Gaussian components.

[0069] Automatic inference is performed by maximizing the lower bound of evidence to balance the model's fit with the data and its complexity, resulting in an approximate posterior distribution of the model parameters. The Gaussian component with the largest mixture weight is selected as the LOS signal cluster, and its corresponding mean vector and covariance matrix are obtained. The LOS signal cluster is expressed as follows.

[0070] (3).

[0071] in, This indicates that locking the independent variable will maximize the function value. This represents the index of the identified LOS component, and its corresponding mean vector is... The covariance matrix is The inverse of the covariance matrix is .

[0072] Step 3: Using the LOS signal distribution parameter characteristics obtained in Step 2, calculate the deviation distance to dispersion ratio of the current observation value, and perform a multi-frequency dispersion consistency test to identify abnormal observation components.

[0073] In this embodiment, the Mahalanobis distance to dispersion ratio of the current observation is calculated using the LOS signal distribution parameter characteristics, and a multi-frequency dispersion consistency test is performed to identify abnormal observations. This includes: using the mean vector and precision matrix obtained in step 2, calculating the Mahalanobis distance of the current observation vectorized feature from the LOS signal distribution. square The calculation formula is as follows.

[0074] (4).

[0075] in, Represents frequency , calendar Next satellite Multi-vectorization features, Represents frequency , calendar Next satellite The squared Mahalanobis distance between the multivariate vectorized features and the LOS distribution. This represents the mean vector of the LOS distribution. Denotes the inverse of the covariance matrix. This indicates transpose.

[0076] In this embodiment, a chi-square threshold at a specific confidence level is introduced. For example, we introduce a four-degree-of-freedom chi-square threshold corresponding to a 95% confidence level (corresponding to the four features in this paper). =9.5, converting Mahalanobis distance into a dispersion ratio under multi-band conditions. And based on the dispersion ratio of each frequency band obtained by the conversion Perform a multi-frequency dispersion consistency check. If all multi-frequency observations indicate anomalies, it is determined that the satellite has a non-line-of-sight anomaly, and the average dispersion ratio is calculated. The calculation formula is as follows.

[0077] (5).

[0078] (6).

[0079] (7).

[0080] in, Represents frequency , calendar Next satellite binary indicator, Represents a binary indicator function, when the condition is... The value is 1 if the condition is met, otherwise it is 0; the satellite is only confirmed as a non-line-of-sight anomaly when multiple frequencies consistently indicate an anomaly.

[0081] Step 4: Based on the consistency test results obtained in Step 3, construct an adaptive weighting function to reduce or eliminate the abnormal observation components obtained in Step 3, and update the variance of the filtered observation noise for use in satellite navigation and positioning processing, thereby obtaining the navigation and positioning solution and realizing the navigation and positioning solution.

[0082] In this embodiment, a parameterized adaptive weighting function is constructed based on the consistency check results (the adaptive weighting function includes a scaling factor that controls the sensitivity of weight decay). and the power exponent that determines the order of decay rate Abnormal observations are downweighted or removed, and the observation noise variance in the GNSS positioning solution filtering is updated, using the average dispersion ratio. Calculate weighting factors The calculation formula is as follows.

[0083] (8).

[0084] in, Indicates the weighting factor. Represents the epoch Next satellite The average dispersion ratio, To establish a cutoff threshold for rejecting severe outliers, in this embodiment of the application... Set to 4.8 (corresponding to a conservative cutoff point at a 99% confidence level, which can effectively eliminate severe outliers while retaining sufficient geometric strength). To control the scaling factor of the weight decay sensitivity, The exponent is used to determine the order of the decay rate.

[0085] According to formula (8), the weighting factor The value depends on the average dispersion ratio. and the threshold for rejecting severe outliers Weighting factor Ratio to mean dispersion There is a clear correlation, based on the average dispersion ratio. or weighting factor It can identify normal observation components, slightly anomalous components, and severely anomalous components, thereby further confirming whether to remove or reduce their weight. An adaptive weighting function is used to calculate a weighting factor between 0 and 1, achieving weight reduction for anomalous observation components; the smaller the weighting factor, the greater the degree of weight reduction.

[0086] For example, in order to facilitate calculation, the embodiments of this application will... Set to 4.8, and All are set to 1. Assume that when the calculated average dispersion ratio... When the value is 1.5, the observed component is determined to be a slightly anomalous component. , , and Substituting the value into formula (8), the weighting factor is calculated. If the average dispersion ratio is 0.67, then the weight of the slightly outlier component is reduced from 1 to 0.67. When the value is 4.0, the observed component is determined to be a slightly anomalous component. , , and Substituting the value into formula (8), the weighting factor is calculated. If the average dispersion ratio is 0.25, then the weight of the slightly outlier component is reduced from 1 to 0.25. When it is 5.0, because Greater than Therefore, this observed component is determined to be a severely outlier and is directly removed. When the calculated average dispersion ratio... When it is 0.8, because Since the value is less than 1.0, the observed component is determined to be a normal observed component, and its weight is kept at 1.0. It is not removed or downweighted.

[0087] Furthermore, using weighting factors Update the observation noise variance in the filtering process to obtain the adjusted variance after filtering. , expressed as the following formula.

[0088] (9).

[0089] in, This represents the initial variance of the observation noise. This represents the variance after filtering and adjustment.

[0090] According to formula (8), when When, corresponding At this point, the abnormal observation component is directly removed. When, corresponding At this point, the abnormal observation component is weighted down to complete the precise position calculation for the current epoch and output a navigation and positioning solution with controlled quality.

[0091] like Figure 3As shown, this embodiment first extracts pseudorange residuals, elevation angles, and signal-to-noise ratio multivariate features within the sliding window to achieve unsupervised probabilistic clustering based on a variational Bayesian-Gaussian mixture model, thereby obtaining prior statistical parameters of the LOS distribution. Secondly, it identifies whether the signal is non-line-of-sight (LOS). If not, navigation calculations are performed directly; if so, the Mahalanobis distance of the current observation vector relative to the LOS distribution is calculated, and a consistency check is performed using a chi-square threshold to identify abnormal satellite observations affected by non-LOS or multipath interference. Finally, a parameterized adaptive weighting function is used to determine whether the dispersion of the observation values ​​exceeds a threshold. If so, the observation values ​​are discarded; otherwise, the observation values ​​are downweighted, and the variance of the filtered observation noise is updated. Abnormal observation data are assigned smaller weights (or directly discarded). This allows for precise calculations of three-dimensional positioning coordinates, receiver clock errors, and ambiguities in complex environments, effectively suppressing the impact of non-LOS signals and multipath interference on observation quality. It achieves adaptive identification and processing of observation anomalies without relying on external prior information, ensuring the reliability of navigation and positioning.

[0092] This application proposes a satellite observation data quality control method. First, a sliding window multivariate vectorized feature is constructed, incorporating pseudorange residuals, satellite elevation angle, and signal-to-noise ratio statistical characteristics. Then, the sliding window multivariate vectorized feature is input into a variational Bayesian-Gaussian mixture model. Variational inference is performed by maximizing the lower bound of evidence to identify line-of-sight (LOS) signal clusters and obtain their distribution parameters. Next, the Mahalanobis distance and dispersion ratio of the current observation vectorized feature deviating from the LOS distribution are calculated, and non-line-of-sight anomalies are identified through a multi-frequency dispersion consistency test. Finally, a parameterized adaptive weighting function is constructed to calculate weighting factors, and the variance of the filtered observation noise is updated accordingly to obtain the processed observation information or navigation and positioning processing results. Specifically, by combining the statistical characteristics of sliding windows to construct multivariate vectorized features including pseudorange residuals, satellite elevation angles, mean SNR, and standard deviation of SNR, and through unsupervised probabilistic clustering based on variational Bayesian-Gaussian mixture models, the limitations of traditional single-feature fixed threshold methods and standard expectation-maximization algorithms, which are sensitive to initialization and susceptible to local optima, are effectively overcome. Adaptive parameterized weighting functions are constructed using Mahalanobis distance and multi-frequency dispersion consistency tests, which can effectively isolate or attenuate non-line-of-sight observations and multipath interference. Without relying on a large amount of labeled training data or 3D city map models, it can improve the continuity, reliability, and 3D positioning accuracy of precision positioning processing in complex urban environments with tall buildings and tree obstructions.

[0093] This application addresses the quality control requirements of GNSS satellite observation data in complex environments. It leverages multi-dimensional features such as observation geometry, signal strength, and signal stability under unsupervised conditions to adaptively identify and suppress non-line-of-sight (Line-of-Sight) and multipath interference signals, thereby providing reliable observation conditions for high-precision satellite positioning services. By constructing a multi-dimensional feature model combined with VB-GMM for non-line-of-sight classification and adaptive processing, the robustness and adaptability of observation data quality control are improved. Compared to existing single-feature combined with VB-GMM methods, the method in this application has strong practicality.

[0094] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 4 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media. The database stores multi-band navigation satellite observation data. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a satellite observation data quality control method.

[0095] Figure 4 The structures shown are merely block diagrams of some structures related to the present application and do not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than shown in the figures, or combine certain components, or have different component arrangements. In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0096] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0097] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0098] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of related data must comply with relevant regulations and be authorized by the owner of the corresponding device.

[0099] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0100] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0101] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0102] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method of satellite observation data quality control, characterized by, The satellite observation data quality control method includes: Acquire multi-band navigation satellite observation data; Based on the multi-band navigation satellite observation data, a sliding window is used to extract signal features and construct a multi-vectorized feature sequence. The variational Bayesian-Gaussian mixture model is used to perform unsupervised clustering on the multivariate vectorized feature sequence to obtain the line-of-sight signal distribution parameter features. Based on the characteristics of the line-of-sight signal distribution parameters, the ratio of Mahalanobis distance to dispersion is calculated and a multi-frequency dispersion consistency test is performed. Abnormal observation components are then identified based on the consistency test results. An adaptive weighting function is constructed based on the consistency test results. The abnormal observation components are then weighted or removed based on the adaptive weighting function, and the observation noise variance of the positioning filter is updated to obtain the navigation and positioning solution.

2. The satellite observation data quality control method according to claim 1, characterized by, Based on the multi-band navigation satellite observation data, a sliding window method is used to extract signal features and construct a multi-vectorized feature sequence, specifically including: Based on the multi-band navigation satellite observation data, the satellite pseudorange observation residual, satellite elevation angle, mean signal-to-noise ratio and standard deviation of signal-to-noise ratio are extracted within the sliding window of each frequency band. Based on the satellite pseudorange observation residuals, satellite elevation angles, mean signal-to-noise ratios, and standard deviations of signal-to-noise ratios within the sliding window of each frequency band, a multivariate vectorized feature sequence is constructed.

3. The satellite observation data quality control method according to claim 1, characterized by, Unsupervised clustering of the multivariate vectorized feature sequences is performed using a variational Bayesian-Gaussian mixture model to obtain the line-of-sight signal distribution parameter features, specifically including: The multivariate vectorized feature sequence is input into the variational Bayesian-Gaussian mixture model, variational inference is performed by maximizing the lower bound of evidence, and the model complexity is adaptively determined to obtain the approximate posterior distribution of the model parameters. Based on the approximate posterior distribution of the model parameters, the Gaussian component with the largest mixing weight is selected as the line-of-sight signal cluster; Obtain the mean vector and covariance matrix corresponding to the line-of-sight signal cluster, and use them as the distribution parameter features of the line-of-sight signal.

4. The satellite observation data quality control method according to claim 1, characterized in that, Based on the characteristics of the line-of-sight signal distribution parameters, the Mahalanobis distance to dispersion ratio is calculated, and a multi-frequency dispersion consistency test is performed. Abnormal observation components are then identified based on the consistency test results. Specifically, this includes: Based on the line-of-sight signal distribution parameter characteristics, calculate the square of the Mahalanobis distance by which the current multivariate vectorized feature deviates from the LOS distribution; The dispersion ratio of each frequency band is calculated based on the square of the Mahalanobis distance and a preset chi-square threshold. A multi-frequency dispersion consistency test is performed based on the dispersion ratio of each frequency band to obtain the consistency test result; The observation component that is determined to be abnormal in the multi-band consistency test results is taken as the abnormal observation component.

5. The satellite observation data quality control method according to claim 4, characterized in that, The square of the Mahalanobis distance is calculated using the following formula: ； in, Represents frequency , calendar Next satellite Multi-vectorization features, Represents frequency , calendar Next satellite The squared Mahalanobis distance between the multivariate vectorized features and the LOS distribution. Indicates the index of the identified LOS component. This represents the mean vector of the LOS distribution. Denotes the inverse of the covariance matrix. Indicates transpose; The dispersion ratio of each frequency band is calculated using the following formula: ； wherein representing frequency , epoch lower satellite of the dispersion ratio, representing chi-square threshold.

6. The satellite observation data quality control method according to claim 1, characterized by, The expression for the adaptive weighting function is: ； ； ； in, Indicates the weighting factor. Represents the epoch Next satellite The average dispersion ratio, Indicates frequency, Represents frequency , calendar Next satellite binary indicator, This represents a binary indicator function. Represents frequency , calendar Next satellite The dispersion ratio, To set a cutoff threshold for rejecting severe outliers, To control the scaling factor of the weight decay sensitivity, The exponent that determines the order of the decay rate; The observation noise variance for updating the positioning filter is expressed by the following formula: ； wherein, represents an initial variance of the observation noise, represents a variance adjusted by the filtering process.

7. The satellite observation data quality control method according to claim 6, characterized by, The abnormal observation components are weighted or removed based on the adaptive weighting function, specifically including: When the weighting factor in the adaptive weighting function is equal to 0, the abnormal observation component is removed. When the weighting factor in the adaptive weighting function is greater than 0 and less than 1, the abnormal observation component is weighted down.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that the processor executes the computer program to implement the satellite observation data quality control method according to any one of claims 1-7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the satellite observation data quality control method according to any one of claims 1-7.

10. A computer program product comprising a computer program, characterized in that, When executed by a processor, the computer program implements the satellite observation data quality control method according to any one of claims 1-7.

Images