Server satellite precision product anomaly detection method, system and terminal

Abstract

The invention relates to a server-side satellite precision product anomaly detection method and system and a terminal, specifically, a previous time window of a current epoch is obtained, the time window comprises a plurality of epochs, the residual mean value and the root-mean-square of each satellite in each epoch in the time window are calculated, and the residual mean value and the root-mean-square of each satellite in each epoch in the time window are calculated; normalizing the residual mean value and the root-mean-square to obtain the characteristics of the satellite in epochs; acquiring features of all the satellites in all epochs in the previous time window to form a feature set, performing outlier detection on the feature set by adopting an LOF method, and determining the LOF score of the satellite at the current epoch moment by utilizing the features of the current epoch of the satellite and the feature set after removing the satellite of which the LOF score is greater than a preset value from the feature set; the LOF scores of the satellites are sent to the PPP user side, when the PPP user side carries out ambiguity fixing, all the satellites are tried to be fixed firstly, if fixing fails, the satellite with the maximum LOF value is removed from the ambiguity set, fixing is tried again, and continuous circulation is carried out till the ambiguity is fixed successfully.

Description

Technical Field

[0001] The present invention relates to the field of satellite navigation technology, and in particular to a method, system and terminal for detecting abnormalities of satellite precision products on a server side. Background Art

[0002] Global Navigation Satellite Systems (GNSS), or Global Navigation Satellite Systems, are satellite systems that use satellite signals to provide precise positioning, navigation, and timing services to users worldwide. These systems include the US GPS, China's Beidou Navigation Satellite System, Russia's GLONASS, and Europe's Galileo Navigation Satellite System. As an important positioning and navigation technology, GNSS is widely used in urban transportation, precision agriculture, surveying and mapping, and other fields. Precise Point Positioning (PPP) is a high-precision positioning technology based on GNSS. Its key difference from conventional meter-level positioning technology lies in the use of satellite precision products to correct for most errors, achieving centimeter-level positioning accuracy. These precision products refer to error correction products such as satellite orbit error correction, satellite clock error correction, satellite signal code deviation correction, and satellite signal carrier deviation correction. These products are generally provided by commercial companies, but some satellite navigation systems, such as Beidou and Galileo, also include free satellite precision products. Therefore, the accuracy and stability of these products directly impact the positioning accuracy of PPP users. Current providers of precision satellite products primarily focus on improving product accuracy, overlooking the equally important role of product stability for users. This is because user-side algorithms often assume that precision satellite products are absolutely correct when performing high-precision position calculations. However, no system can guarantee 100% error-free reliability. Faulty satellite products can lead to unreliable positioning results for users, often difficult to detect. Therefore, timely and accurate detection of GNSS service anomalies is crucial to ensuring the reliability and security of positioning and navigation.

[0003] The reliability of GNSS satellite precision products is crucial. Current methods for detecting product anomalies rely primarily on comparing products from different organizations, which lacks real-time performance and relies on third-party satellite precision products, which are also not 100% reliable. Traditional anomaly detection algorithms are mostly based on statistical methods, such as the 3σ principle, or on clustering algorithms. These methods have limitations: statistical methods often assume that the data follows a specific probability distribution, while clustering methods typically only provide a 0 or 1 judgment and cannot quantify the degree of anomaly for each data point. Summary of the Invention

[0004] In view of the deficiencies in the prior art, in a first aspect of the present invention, a method for detecting abnormalities in satellite precision products on a server side is provided, the method comprising the following steps:

[0005] Obtain a time window before the current epoch, where the time window includes multiple epochs, calculate the residual mean and root mean square of each satellite in each epoch in the time window, and normalize the residual mean and root mean square to obtain the characteristics of the satellite in the epoch.

[0006] The features of all satellites in all epochs in the previous time window are obtained to form a feature set. The LOF method is used to detect outliers on the feature set. After satellites with LOF scores greater than a preset value are removed from the feature set, the LOF score of the satellite at the current epoch is determined using the features of the satellite's current epoch and the feature set.

[0007] The LOF score of the satellite is sent to the PPP user end. When the PPP user end performs ambiguity fixation, it first tries to fix all satellites. If the fixation fails, it removes the satellite with the largest LOF value from the ambiguity set and tries to fix it again. This cycle continues until the ambiguity is successfully fixed.

[0008] Preferably, the calculating of the residual mean and the root mean square of each satellite in each epoch in the time window is specifically as follows:

[0009] Calculate the posterior residuals of the satellite in one epoch Here, i represents an observation station and s represents a satellite.

[0010] Construct the residual vector V of n observation stations for satellite s s ,in,

[0011] Calculate the residual mean μ of satellite s s ,in,

[0012] Calculate the root mean square σ of satellite s s ,in, P is the weight matrix composed of the weights of the observation quantities when solving the precision product.

[0013] Preferably, the determining of the LOF score of the satellite at the current epoch by using the characteristics and characteristic set of the current epoch of the satellite is specifically:

[0014] The characteristics of the satellite at the current epoch and the characteristics of the retained satellites are taken as a whole, and the LOF score is calculated using the LOF method.

[0015] Preferably, the determining of the LOF score of the satellite at the current epoch by using the characteristics and characteristic set of the current epoch of the satellite is specifically:

[0016] Based on the characteristics of the satellite at the current epoch, find the k neighboring satellites with the most similar characteristics among the retained satellite characteristics.

[0017] Calculate the ratio of the average value of the distances between k neighbor satellites to the average value of the distances from the satellite to its k neighbor satellites at the current epoch, calculate the average local density of the k neighbor satellites, obtain the local density of the satellite at the current epoch based on the ratio and the average local density, and obtain the LOF score based on the local density of the satellite at the current epoch.

[0018] In a second aspect of the present invention, a service-side satellite precision product anomaly detection system is provided, the system comprising the following modules:

[0019] The satellite feature acquisition module is used to obtain the posterior residual of the satellite relative to the observation station in a time window before the current epoch through at least one precision product, calculate the residual mean and root mean square of the satellite based on the posterior residual, and use the normalized residual mean and root mean square as the satellite feature.

[0020] The satellite LOF score calculation module is used to perform outlier detection on multiple satellites using the LOF method based on satellite characteristics, eliminate satellites with LOF scores greater than a preset value, and determine the LOF score of the satellite at the current epoch.

[0021] The anomaly detection module is used to send the satellite's LOF score to the PPP user end. When the PPP user end performs ambiguity fixation, it first tries to fix all satellites. If the fixation fails, it removes the satellite with the largest LOF value from the ambiguity set and tries to fix it again. This cycle continues until the ambiguity is successfully fixed.

[0022] Preferably, the calculating of the residual mean and the root mean square of each satellite in each epoch in the time window is specifically as follows:

[0023] Calculate the posterior residuals of the satellite in one epoch Here, i represents an observation station and s represents a satellite.

[0024] Construct the residual vector V of n observation stations for satellite s s ,in,

[0025] Calculate the residual mean μ of satellite s s ,in,

[0026] Calculate the root mean square σ of satellite s s ,in, P is the weight matrix composed of the weights of the observation quantities when solving the precision product.

[0027] Preferably, the determining of the LOF score of the satellite at the current epoch by using the characteristics and characteristic set of the current epoch of the satellite is specifically:

[0028] The characteristics of the satellite at the current epoch and the characteristics of the retained satellites are taken as a whole, and the LOF score is calculated using the LOF method.

[0029] Preferably, the determining of the LOF score of the satellite at the current epoch by using the characteristics and characteristic set of the current epoch of the satellite is specifically:

[0030] Based on the characteristics of the satellite at the current epoch, find the k neighboring satellites with the most similar characteristics among the retained satellite characteristics.

[0031] Calculate the ratio of the average value of the distances between k neighbor satellites to the average value of the distances from the satellite to its k neighbor satellites at the current epoch, calculate the average local density of the k neighbor satellites, obtain the local density of the satellite at the current epoch based on the ratio and the average local density, and obtain the LOF score based on the local density of the satellite at the current epoch.

[0032] In a third aspect of the present invention, a service-side satellite precision product anomaly detection terminal is provided, the terminal comprising at least a computer-readable storage medium and a processor, the computer-readable storage medium storing a computer program, and the computer program, when executed by the processor, implements the method described in the first aspect.

[0033] The present invention uses the mean and root mean square of the residuals from the server-side precision product estimation process as characteristic values for anomaly detection. A sliding window is used to update the historical LOF model in real time to adapt to changes in the characteristics of the detection samples, improve the detection success rate, and reduce the false positive rate. The satellite LOF score is broadcast to the PPP user terminal along with the satellite precision product. The PPP user terminal then performs corresponding processing based on the LOF score of each satellite to avoid the impact of poor satellite products. Furthermore, each time window in the present invention includes multiple epochs, and the characteristics of the satellite change with each epoch. If the change is relatively large, points early in the time window are more likely to become outliers, which will not affect the judgment of the current epoch. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 Schematic diagram of LOF anomaly detection principle;

[0035] Figure 2 This is a flow chart of Example 1;

[0036] Figure 3 A flowchart of the method for using LOF to detect abnormal satellites and user terminals for satellite precision products;

[0037] Figure 4Schematic diagram of GPS satellite anomaly detection results. DETAILED DESCRIPTION

[0038] In the embodiments of the present invention, words such as "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of this application should not be construed as being preferred or advantageous over other embodiments or designs. Rather, the use of words such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner to facilitate understanding.

[0039] It will be understood that the “embodiment” mentioned throughout the specification means that the specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present application. Therefore, the various embodiments throughout the specification do not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It will be understood that in the various embodiments of the present application, the size of the sequence number of each process does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present application.

[0040] In the present invention, unless otherwise specified, the same or similar parts between the various embodiments can refer to each other. In the various embodiments of the present invention, and the various implementation methods / implementation methods / implementation methods in each embodiment, if there is no special explanation and logical conflict, the terms and / or descriptions between different embodiments and the various implementation methods / implementation methods / implementation methods in each embodiment are consistent and can be referenced to each other. The technical features in different embodiments and the various implementation methods / implementation methods / implementation methods in each embodiment can be combined to form new embodiments, implementation methods, implementation methods, or implementation methods according to their inherent logical relationships. The implementation methods of the present application described below do not constitute a limitation on the scope of protection of the present application.

[0041] LOF (Local Outlier Factor) is an unsupervised anomaly detection method in machine learning that uses local point density to identify data with outliers. This method is based on the assumption that the density around non-outlier objects is similar to the density around their neighbors, while the density around outlier objects is significantly different from the density of the points around their neighbors. Figure 1 The figure shows a schematic diagram of LOF anomaly detection. LOF calculates the local density of each feature point based on the number of points near it. Normal features are densely distributed, resulting in a high local density, while outliers have a low local density due to their outlier nature.

[0042] The present invention does not need to rely on third-party product comparisons, and according to the characteristics of the LOF method, it can perform adaptive anomaly detection based on the characteristic values of the data, without making too many assumptions about the distribution of abnormal data, and can also quantify the degree of abnormality of each data.

[0043] Example 1

[0044] like Figure 2 As shown, a method for detecting abnormalities of satellite precision products on a server side is provided, comprising the following steps:

[0045] S1. Obtain the time window before the current epoch, where the time window includes multiple epochs. Calculate the residual mean and root mean square (RMS) of each satellite in each epoch within the time window. Normalize the residual mean and RMS to obtain the characteristics of the satellite in the epoch.

[0046] Observation stations are used to track and observe GNSS (Global Navigation Satellite System) satellites. Precision products are obtained by processing satellite data collected by observation stations to obtain precise orbit products, precise clock products, signal deviations, etc. Each precision product may obtain data from multiple satellites through multiple observation stations. When judging whether a satellite is abnormal, the posterior residual of the satellite relative to at least one observation station is obtained through at least one precision product. In one embodiment, the posterior residual is when the data of the observation station is used to estimate the orbit, clock error and other parameters of the satellite. The observation value is predicted based on certain parameters, and the actual observed value is compared with the predicted value of the model. The difference between the two is the posterior residual. For the same satellite, the mean residual and root mean square of the satellite can be obtained through at least one precision product and at least one observation station.

[0047] Assume that there are two satellites and a time window has 4 epochs. As shown in Table 1, the characteristics of the 4 epochs of Satellite 1 in a time window are: (μ11, σ11), (μ12, σ12), (μ13, σ13), (μ14, σ14); the characteristics of Satellite 2 in a time window are: (μ21, σ21), (μ22, σ22), (μ23, σ23), (μ24, σ24).

[0048] Table 1

[0049]

[0050] Then 8 features are obtained, and LOF calculation is performed on these 8 features.

[0051] In one embodiment, the residual mean and root mean square of the satellite are calculated based on the posterior residual, specifically:

[0052] Calculate the posterior residual for each satellite in the current epoch Here, i represents an observation station and s represents a satellite.

[0053] Construct the residual vector V of n observation stations for satellite s s ,in,

[0054] Calculate the residual mean μ of satellite s s ,in,

[0055] Calculate the root mean square σ of satellite s s ,in, P is the weight matrix composed of the weights of the observation quantities when solving the precision product.

[0056] S2, obtaining the features of all satellites in all epochs in the previous time window to form a feature set, using the LOF method to detect outliers on the feature set, removing satellites with LOF scores greater than a preset value from the feature set, and then using the features of the satellite's current epoch and the feature set to determine the LOF score of the satellite at the current epoch.

[0057] Using the above example, the feature set is {(μ11, σ11), (μ12, σ12), (μ13, σ13), (μ14, σ14), (μ21, σ21), (μ22, σ22), (μ23, σ23), (μ24, σ24)}. For example, after performing LOF outlier detection on the eight features above and removing satellites with LOF scores greater than a preset value, the feature set consisting of the features of the remaining satellites becomes {(μ11, σ11), (μ12, σ12), (μ21, σ21), (μ23, σ23)}.

[0058] Within the time window, all satellites are checked for outliers using the LOF method. If the LOF score is greater than a preset value, it indicates that the satellite is abnormal and is removed.

[0059] Then, the LOF score of the satellite at the current epoch is determined using the characteristics of the remaining satellites and the satellite at the current epoch. In one embodiment, the characteristics of the satellite at the current epoch and the characteristics of the retained satellites are taken as a whole and the LOF score is calculated using the LOF method.

[0060] In yet another embodiment, based on the characteristics of the satellite at the current epoch, k neighbor satellites with characteristics most similar to the satellite's characteristics are found from the retained satellite characteristics.

[0061] Calculate the ratio of the average distance between k neighbor satellites to the average distance from the satellite to its k neighbor satellites at the current epoch, calculate the average local density of the k neighbor satellites, calculate the local density of the satellite at the current epoch based on the ratio and the average local density, and calculate the LOF score based on the local density of the satellite at the current epoch. k is a positive integer.

[0062] Select the target satellite and find its nearest k satellites as neighbors, for example, k = 3. For each neighbor satellite (S1, S2, S3), calculate the average distance between k neighbors / the average distance from the target satellite to the neighbor to obtain a ratio. If the ratio is <1, it means that the distance between the target satellite and the neighbor is farther than the distance between the neighbors, and it may be in a sparse area. Calculate their own local densities separately and take the average. Combine the ratio with the average local density, for example, multiply them. The density of the target satellite is affected by the distance relationship between itself and the neighbors and the environment in which the neighbors are located. If the distance itself is far and the neighbor density is low, the local density is even lower. Compare the local density of the target satellite with the average local density of the neighbors to obtain the LOF score. Preferably, the LOF score is the ratio of the average local density of the neighbors / the target local density.

[0063] S3 sends the satellite's LOF score to the PPP client. When the PPP client performs ambiguity fixation, it first attempts to fix all satellites. If the fixation fails, it removes the satellite with the largest LOF value from the ambiguity set and tries to fix it again. This cycle continues until the ambiguity is successfully fixed.

[0064] By transmitting the satellite's LOF score to the user end, a dynamic elimination strategy is adopted during the ambiguity fixation process. This means that all satellites are first used for the solution. If the fixation fails, the abnormal satellite with the highest LOF value in the current satellite set is automatically removed. Unreliable satellites affected by multipath interference, orbital deviation, or signal loss are gradually eliminated through iterative screening until the observation data of the remaining satellites meet the conditions for integer ambiguity solution. Furthermore, by quantifying the degree of anomaly in the spatial distribution of satellites, intelligent satellite selection is achieved, which not only maintains the satellite geometry required for positioning, but also effectively improves the solution success rate and positioning accuracy.

[0065] Example 2

[0066] Figure 3 The flowchart of the method of using LOF to detect abnormal satellites and user terminals of satellite precision products is shown as follows: Figure 3As shown in the figure, reference station network data is observation data from satellites such as the Global Navigation Satellite System (GNSS) collected by a network of ground-based observation stations. The server first performs a floating-point PPP solution and floating-point ambiguity calculation, followed by real-time UPD estimation and UPD corrections. After calculating the mean and root mean square (RMS) of the UPD, the LOF score is calculated to identify outliers in the data, specifically to detect anomalous satellites in satellite precision products. The LOF score is then transmitted to the satellite precision product, resulting in a satellite precision product containing the LOF score.

[0067] The user terminal first performs a floating-point PPP calculation based on the receiver observations and the broadcast ephemeris. It then attempts to fix the ambiguities. If successful, it outputs a fixed solution. If not, it removes the satellite with the highest LOF score and attempts ambiguity fixation again. This process repeats until the ambiguities are successfully fixed or the number of remaining satellites falls below four. If the number of remaining satellites falls below four, reliable ambiguity fixation is unsuccessful and a floating-point solution is output.

[0068] More specifically, the method for determining abnormal values of satellite positioning precision products of this embodiment includes the following steps:

[0069] (1) Eigenvalue selection. Assume that a precision product estimation server is connected to n reference observation stations and tracks m GNSS satellites. The subscript i represents the station and the superscript s represents the satellite. After the server completes the estimation of precise orbit products, precise clock products, signal deviations and other products based on the observation data of the reference observation station, it will simultaneously obtain the posterior residuals of each observation quantity. Residuals are referred to as residuals below. Residuals follow a normal distribution. According to the properties of the normal distribution, the mean and root mean square are the characteristic values of the normal distribution. That is, a set of fixed mean and root mean square (μ, σ) corresponds to a unique normal distribution.

[0070] (2) Calculation of residual eigenvalues. Taking satellite as the dimension, the residual of a certain satellite with respect to each station is counted. The residual mean with respect to satellite s can be calculated by formula (1):

[0071]

[0072] The residual vector V is formed by the residual of satellite s to n stations s , as shown in formula (2), and then the root mean square of satellite s is obtained according to formula (3):

[0073]

[0074] P in formula (3) is the weight matrix composed of the weights of the observations when solving the precision product.

[0075] (3) Eigenvalue normalization. In order for the LOF method to better detect residual outliers, considering the differences in the magnitude of the mean and root mean square, the residual mean and root mean square are normalized before anomaly detection. The purpose of this is to unify the mean and root mean square scales to ensure uniform gradients, and on the other hand, to reduce human intervention in parameter adjustment when training the LOF model, and to speed up the convergence of the LOF training model, which is conducive to real-time processing. The mean and root mean square magnitudes are unified by the Min-Max Normalization method, and the value range of each feature is unified to [0,1] by scaling. For each eigenvalue x i ,i=1,2,…,p, the eigenvalue x of the kth sample ik ,k=1,2,…,n, where the sample is the object to be detected, one sample corresponds to one satellite, and the eigenvalue is two-dimensional. After normalization, it becomes:

[0076]

[0077] min and max are the minimum and maximum value functions, respectively. From formula (4), we can see that the normalized residual and root mean square will be mapped to the interval [0,1].

[0078] (4) Draw a distribution diagram of satellite residual eigenvalues. With the mean μ as the horizontal axis and the root mean square σ as the vertical axis, the (μ, σ) of each satellite corresponds to a point in a plane rectangular coordinate system. The residuals of normal satellites follow a normal distribution, with similar μ and σ. The corresponding points are clustered in the coordinate system. However, the residuals of abnormal satellites no longer follow a normal distribution. Their μ and σ values are different from those of normal satellites, and the corresponding points are distributed outliers in the coordinate system.

[0079] (5) Training the LOF model, that is, using historical data as a model, calculating the LOF point density of historical data, and using the point density model of historical data to detect whether the local density of new data is consistent with the historical data, thereby determining outliers. Most variables in GNSS data processing are time series. In order to fully utilize the value of historical data, based on the LOF method, the normalized UPD residual mean and root mean square are used to detect new points in a sliding window manner. The role of the sliding window is to use the data in a time window before the current epoch to train a historical LOF model, improve the detection accuracy and reduce the error rate. The reason for using a sliding window is that it has been found experimentally that the characteristic values of satellites will change slowly as a whole over time. Therefore, the LOF model is trained with a sliding window that is updated in real time so that it can always be kept up to date. The characteristic value of each satellite's current epoch is used as a point detection.

[0080] The specific approach is to assume the current epoch is t, train the LOF model using historical data from the period w minutes prior to t, and then use the trained model to detect residual anomalies at epoch t. Satellites marked as anomalies must be excluded from the training data, as this will affect the convergence of the LOF model. The purpose of training is to determine the local density of feature points within the time window closest to the current epoch. LOF is an unsupervised anomaly detection method, and points with a LOF greater than 1.5 are already excluded during training, eliminating the need for further labeling.

[0081] To prevent overfitting of the LOF training model, the sliding window length w is set to 30 minutes, and 90% of the number of training data is used as the number of k nearest neighbors of the LOF model. The specific training parameters are shown in Table 1:

[0082] Table 1 LOF model training strategy

[0083] Training parameters Training strategy Training window duration 30 minutes Abnormal satellite rejection threshold LOF score > 1.5 Number of k nearest neighbors 90% of the training data volume

[0084] (6) Detection results, taking GPS as an example, Figure 4 This is a schematic diagram of the results after using LOF detection. Other systems such as China's BeiDou, Russia's GLONASS, and Europe's Galileo are similar to GPS and will not be described in detail.

[0085] Figure 4 Each dot represents the mean and root mean square (RMS) residual for that satellite at that epoch. White dots represent model training data during the sliding window period, yellow dots represent normal satellite data for the current detection epoch, and red dots represent outlier data detected at the current epoch. The color in the figure represents the LOF score for that region, with red indicating a higher degree of outlier detection.

[0086] from Figure 4 As can be seen in the figure, the LOF value gradient is relatively uniform, with contour lines approaching circles. The mean and RMS values of the outlier satellites G07, G10, G18, and G30 exhibit clear outlier characteristics, and the LOF value increases with the distance from the outlier. This is because the LOF value is inversely proportional to the local point density surrounding the feature point: a higher local point density indicates a lower LOF value, and vice versa. Therefore, the LOF value can, to a certain extent, quantify the degree of outlier status of a feature point. The farther the feature point is from the group, the lower its local density, resulting in a higher LOF value.

[0087] (7) Satellite LOF score broadcasting and PPP user-side processing.

[0088] The LOF scores for each satellite obtained above are broadcasted to the PPP client along with the satellite precision product. When the PPP client performs ambiguity fixation, it first attempts to fix all satellites. If fixation fails, it removes the satellite with the largest LOF value from the ambiguity set and tries again. This process repeats until the ambiguity is successfully fixed.

[0089] The present invention uses a sliding window to update the historical LOF model in real time to adapt to the characteristic changes of the detection samples, improve the detection success rate and reduce the false positive rate.

[0090] Example 3

[0091] This embodiment provides a server-side satellite precision product anomaly detection system, which includes the following modules:

[0092] The satellite feature acquisition module is used to obtain the time window before the current epoch, where the time window includes multiple epochs, calculate the residual mean and root mean square of each satellite in each epoch within the time window, and normalize the residual mean and root mean square to obtain the satellite's features in the epoch.

[0093] A satellite LOF score calculation module is used to obtain the features of all satellites in all epochs in the previous time window to form a feature set, use the LOF method to perform outlier detection on the feature set, remove satellites with LOF scores greater than a preset value from the feature set, and use the features of the satellite's current epoch and the feature set to determine the LOF score of the satellite at the current epoch;

[0094] The anomaly detection module is used to send the satellite's LOF score to the PPP user end. When the PPP user end performs ambiguity fixation, it first tries to fix all satellites. If the fixation fails, it removes the satellite with the largest LOF value from the ambiguity set and tries to fix it again. This cycle continues until the ambiguity is successfully fixed.

[0095] Preferably, the calculating of the residual mean and the root mean square of each satellite in each epoch in the time window is specifically as follows:

[0096] Calculate the posterior residuals of the satellite in one epoch Here, i represents an observation station and s represents a satellite.

[0097] Construct the residual vector V of n observation stations for satellite s s ,in,

[0098] Calculate the residual mean μ of satellite s s ,in,

[0099] Calculate the root mean square σ of satellite s s ,in, P is the weight matrix composed of the weights of the observation quantities when solving the precision product.

[0100] Preferably, the determining of the LOF score of the satellite at the current epoch is specifically:

[0101] The characteristics of the satellite at the current epoch and the characteristics of the retained satellites are taken as a whole, and the LOF score is calculated using the LOF method.

[0102] Preferably, the determining of the LOF score of the satellite at the current epoch is specifically:

[0103] Based on the characteristics of the satellite at the current epoch, find the k neighboring satellites with the most similar characteristics among the retained satellite characteristics.

[0104] Calculate the ratio of the average value of the distances between k neighbor satellites to the average value of the distances from the satellite to its k neighbor satellites at the current epoch, calculate the average local density of the k neighbor satellites, obtain the local density of the satellite at the current epoch based on the ratio and the average local density, and obtain the LOF score based on the local density of the satellite at the current epoch.

[0105] Example 4

[0106] This embodiment provides a service-side satellite precision product anomaly detection terminal, which includes at least a computer-readable storage medium and a processor. The computer-readable storage medium stores a computer program, and the computer program implements the method described in Example 1 when executed by the processor.

[0107] The above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website, computer, server or data center to another website, computer, server or data center via a wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) method. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrations. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive (SSD)).

[0108] The steps of the methods or algorithms described in the embodiments of the present application can be directly embedded in hardware, software units executed by a processor, or a combination of the two. The software units can be stored in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, removable disk, CD-ROM, or other storage media in any form known in the art. Exemplarily, the storage medium can be connected to the processor so that the processor can read information from the storage medium and write information to the storage medium. Alternatively, the storage medium can also be integrated into the processor. The processor and storage medium can be arranged in an ASIC.

[0109] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0110] Although the present application has been described with reference to specific features and embodiments thereof, it is apparent that various modifications and combinations may be made thereto without departing from the spirit and scope of the present application. Accordingly, this specification and the drawings are merely illustrative of the present application as defined by the appended claims and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, those skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, the present application is intended to include such modifications and variations if they fall within the scope of the claims of the present application and their equivalents.

Claims

1. A method for detecting abnormalities of satellite precision products on a server side, characterized in that: The method comprises the following steps: Obtain a time window before the current epoch, where the time window includes multiple epochs, calculate the residual mean and root mean square of each satellite in each epoch within the time window, and normalize the residual mean and root mean square to obtain the characteristics of the satellite in the epoch; Obtain a feature set consisting of features of all satellites in all epochs in the previous time window, perform outlier detection on the feature set using the LOF method, remove satellites with LOF scores greater than a preset value from the feature set, and then determine the LOF score of the satellite at the current epoch using the features of the satellite's current epoch and the feature set; The LOF score of the satellite is sent to the PPP user end. When the PPP user end performs ambiguity fixation, it first tries to fix all satellites. If the fixation fails, it removes the satellite with the largest LOF value from the ambiguity set and tries to fix it again. This cycle continues until the ambiguity is successfully fixed.

2. The method according to claim 1, wherein The residual mean and root mean square of each satellite in each epoch within the time window are calculated as follows: Calculate the posterior residuals of the satellite in one epoch Among them, i represents the observation station and s represents the satellite; Construct the residual vector V of n observation stations for satellite s s ,in, Calculate the residual mean μ of satellite s s ,in, Calculate the root mean square σ of satellite s s ,in, P is the weight matrix composed of the weights of the observation quantities when solving the precision product.

3. The method according to claim 1, wherein The LOF score of the satellite at the current epoch is determined by using the characteristics and characteristic set of the satellite at the current epoch, specifically: The characteristics of the satellite at the current epoch and the characteristics of the retained satellites are taken as a whole, and the LOF score is calculated using the LOF method.

4. The method according to claim 1, wherein The LOF score of the satellite at the current epoch is determined by using the characteristics and characteristic set of the satellite at the current epoch, specifically: Based on the characteristics of the satellite at the current epoch, find the k neighboring satellites with the most similar characteristics among the retained satellite characteristics; Calculate the ratio of the average value of the distances between k neighbor satellites to the average value of the distances from the satellite to its k neighbor satellites at the current epoch, calculate the average local density of the k neighbor satellites, obtain the local density of the satellite at the current epoch based on the ratio and the average local density, and obtain the LOF score based on the local density of the satellite at the current epoch.

5. A server-side satellite precision product anomaly detection system, characterized in that: The system includes the following modules: A satellite feature acquisition module is used to obtain a time window before the current epoch, where the time window includes multiple epochs, calculate the residual mean and root mean square of each satellite in each epoch within the time window, and normalize the residual mean and root mean square to obtain the satellite feature in the epoch; A satellite LOF score calculation module is used to obtain the features of all satellites in all epochs in the previous time window to form a feature set, use the LOF method to perform outlier detection on the feature set, remove satellites with LOF scores greater than a preset value from the feature set, and use the features of the satellite's current epoch and the feature set to determine the LOF score of the satellite at the current epoch; The anomaly detection module is used to send the satellite's LOF score to the PPP user end. When the PPP user end performs ambiguity fixation, it first tries to fix all satellites. If the fixation fails, it removes the satellite with the largest LOF value from the ambiguity set and tries to fix it again. This cycle continues until the ambiguity is successfully fixed.

6. The system according to claim 5, wherein: The residual mean and root mean square of each satellite in each epoch within the time window are calculated as follows: Calculate the posterior residuals of the satellite in one epoch Among them, i represents the observation station and s represents the satellite; Construct the residual vector V of n observation stations for satellite s s ,in, Calculate the residual mean μ of satellite s s ,in, Calculate the root mean square σ of satellite s s ,in, P is the weight matrix composed of the weights of the observation quantities when solving the precision product.

7. The system according to claim 5, wherein: The LOF score of the satellite at the current epoch is determined by using the characteristics and characteristic set of the satellite at the current epoch, specifically: The characteristics of the satellite at the current epoch and the characteristics of the retained satellites are taken as a whole, and the LOF score is calculated using the LOF method.

8. The system according to claim 5, wherein: The LOF score of the satellite at the current epoch is determined by using the characteristics and characteristic set of the satellite at the current epoch, specifically: Based on the characteristics of the satellite at the current epoch, find the k neighboring satellites with the most similar characteristics among the retained satellite characteristics; Calculate the ratio of the average value of the distances between k neighbor satellites to the average value of the distances from the satellite to its k neighbor satellites at the current epoch, calculate the average local density of the k neighbor satellites, obtain the local density of the satellite at the current epoch based on the ratio and the average local density, and obtain the LOF score based on the local density of the satellite at the current epoch.

9. A service-side satellite precision product anomaly detection terminal, the terminal comprising at least a computer-readable storage medium and a processor, characterized in that: The computer-readable storage medium stores a computer program, which implements the method according to any one of claims 1 to 4 when executed by a processor.

Images