Satellite abnormal orbit change identification method and system

By performing feature filtering and clustering on satellite orbit data, abnormal feature vectors are identified, and the problem of large error in satellite orbit abnormality detection in the prior art is solved, and the accuracy and universality of recognition are improved.

CN118708997BActive Publication Date: 2024-11-15INST OF AUTOMATION CHINESE ACAD OF SCI
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411181151.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-27
Publication Date
2024-11-15
Estimated Expiration
2044-08-27

AI Technical Summary

Technical Problem

The prior art has problems with large errors in satellite orbit abnormal detection, especially due to insufficient or deviations in dependence on historical data, errors in astrok kinematic model and artificial selection of cluster clusters, which affect the accuracy and generality of the detection. sex.

Method used

By determining candidate feature data from the target dataset, data filtering based on variance values ​​and correlations, orbital feature vectors are generated, and clustering is performed using density-based clustering methods (such as DBSCAN), the anomaly feature vectors are identified to determine satellites Abnormal track change phenomenon.

Benefits of technology

It improves the accuracy of satellite anomaly orbit recognition, reduces detection errors, enhances the universality and scalability of the algorithm, and reduces the dependence on manual annotation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118708997B_ABST
    Figure CN118708997B_ABST
Patent Text Reader

Abstract

The present invention provides a method and system for identifying abnormal orbit changes of satellites, which can be applied to the field of aerospace measurement and control technology. The method includes: determining multiple candidate feature data based on a target data set; filtering multiple candidate feature data based on variance values ​​corresponding to each of the multiple candidate feature data and correlation values ​​between each of the multiple candidate feature data to obtain target feature data; determining feature change data based on target feature sub-data corresponding to each of the multiple time periods included in the target feature data; generating orbit feature vectors corresponding to each of the multiple time periods based on the target feature data and the feature change data; clustering the orbit feature vectors corresponding to each of the multiple time periods to obtain clustering results; and determining that the satellite has an abnormal orbit change phenomenon when it is determined that the clustering results indicate that there are abnormal feature vectors in the orbit feature vectors corresponding to each of the multiple time periods.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of aerospace measurement and control technology, and in particular to a method and system for identifying abnormal satellite orbit changes. Background Art

[0002] As the number of satellites increases, limited orbital resources become more valuable, the relative motion between satellites becomes more complex, and the potential risk of collision increases. Therefore, for non-cooperative targets, that is, satellites that may deviate from the intended orbit due to failure, loss of control or other reasons, it is of great practical significance to promptly detect orbital anomalies.

[0003] At present, the methods for detecting orbital anomalies based on the change characteristics of the selected orbital elements generally include the following: First, the threshold detection technology based on statistical analysis uses the historical data of space targets with similar orbital heights to train and generate detection thresholds, and uses the n-σ criterion to realize orbit change behavior detection. However, if it relies too much on the historical data of space targets with similar orbital heights, if the historical data is insufficient or biased, it will affect the accuracy of the detection threshold, and thus make the orbital anomaly detection error larger; second, the data fitting technology based on orbital propagation usually uses the celestial kinematic model to predict backwards, analyzes the difference between the predicted orbit and the actual orbit to realize orbital maneuver identification, but the celestial kinematic model has certain errors, and it is impossible to distinguish between model errors, data errors or maneuvering behaviors, and the error will be larger as time goes by; third, the machine learning clustering technology is used, but the relevant technology usually selects the characteristic variables and the number of clustering clusters according to the clustering silhouette coefficient, ignoring the correlation between variables and the characteristics of time series data, and requires manual specification of the number of clustering clusters or data labels, which limits the versatility and scalability of the algorithm and makes the orbital anomaly detection error larger. Summary of the invention

[0004] In view of the above problems, the present invention provides a method and system for identifying abnormal satellite orbit changes.

[0005] According to one aspect of the present invention, a method for identifying abnormal orbit changes of satellites is provided, comprising: determining a plurality of candidate feature data based on a target data set, wherein the candidate feature data is feature data related to the satellite orbit; filtering the plurality of candidate feature data based on variance values ​​corresponding to each of the plurality of candidate feature data and correlation values ​​between each of the plurality of candidate feature data to obtain target feature data; determining feature change data based on target feature sub-data corresponding to each of the plurality of time periods included in the target feature data, wherein the feature change data characterizes changes in the target feature sub-data between adjacent time periods; generating orbit feature vectors corresponding to each of the plurality of time periods based on the target feature data and the feature change data; clustering the orbit feature vectors corresponding to each of the plurality of time periods to obtain clustering results; and determining that the satellite has an abnormal orbit change phenomenon when it is determined that the clustering results characterize the existence of abnormal feature vectors in the orbit feature vectors corresponding to each of the plurality of time periods.

[0006] Another aspect of the present invention provides a satellite abnormal orbit change identification system, including: a first determination module, used to determine multiple candidate feature data based on a target data set, wherein the candidate feature data is feature data related to the satellite orbit; a filtering module, used to filter multiple candidate feature data based on variance values ​​corresponding to each of the multiple candidate feature data and correlation values ​​between each of the multiple candidate feature data to obtain target feature data; a second determination module, used to determine feature change data based on target feature sub-data corresponding to each of the multiple time periods included in the target feature data, wherein the feature change data represents the change of the target feature sub-data between adjacent time periods; a generation module, used to generate orbit feature vectors corresponding to each of the multiple time periods based on the target feature data and the feature change data; a processing module, used to cluster the orbit feature vectors corresponding to each of the multiple time periods to obtain a clustering result; a phenomenon determination module, used to determine that the satellite has an abnormal orbit change phenomenon when it is determined that the clustering result represents that there are abnormal feature vectors in the orbit feature vectors corresponding to each of the multiple time periods.

[0007] According to the satellite abnormal orbit change identification method of the present invention, multiple candidate feature data related to the satellite orbit are obtained from the target data set, and multiple candidate feature data are filtered from the perspective of the change of a single candidate feature data itself, that is, the variance value corresponding to each of the multiple candidate feature data and the correlation between the multiple candidate feature data, that is, the correlation value, so as to achieve the screening of redundant information, that is, to achieve the feature dimension reduction for subsequent processing, and determine the feature change data through the target feature sub-data corresponding to each of the multiple time periods included in the target feature data, so as to pay attention to the time information of the target feature data, and cluster the orbit feature vectors corresponding to each of the multiple time periods generated by the target feature data and the feature change data, and determine whether there is an abnormal orbit change phenomenon according to the clustering result, so as to at least partially solve the problem of large error in the identification of satellite abnormal orbit changes existing in the related art, and achieve the technical effect of improving the accuracy of satellite abnormal orbit change identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] The above contents and other objects, features and advantages of the present invention will become more apparent through the following description of the embodiments of the present invention with reference to the accompanying drawings.

[0009] Figure 1 The application scenario diagram of the satellite abnormal orbit change identification method and system according to an embodiment of the present invention is schematically shown.

[0010] Figure 2 The flowchart of the method for identifying abnormal satellite orbit changes according to an embodiment of the present invention is schematically shown.

[0011] Figure 3 A schematic diagram schematically shows correlation values ​​between multiple intermediate feature data according to an embodiment of the present invention.

[0012] Figure 4 The schematic diagram schematically shows a clustering result according to an embodiment of the present invention.

[0013] FIG5( a ) schematically shows a semi-major axis variation curve constructed based on an abnormal feature vector according to an embodiment of the present invention.

[0014] FIG5( b ) schematically shows a first schematic diagram of a first motion trajectory and a second motion trajectory according to an embodiment of the present invention.

[0015] FIG6( a ) schematically shows a tilt angle variation curve diagram constructed based on an abnormal feature vector according to an embodiment of the present invention.

[0016] FIG6( b ) schematically shows a second schematic diagram of the first motion trajectory and the second motion trajectory according to an embodiment of the present invention.

[0017] FIG. 7( a ) schematically shows an eccentricity variation curve constructed based on an abnormal feature vector according to an embodiment of the present invention.

[0018] FIG. 7( b ) schematically shows a third schematic diagram of the first motion trajectory and the second motion trajectory according to an embodiment of the present invention.

[0019] Figure 8 A flowchart of a method for identifying abnormal satellite orbit changes according to another embodiment of the present invention is schematically shown.

[0020] Fig. 9 The structure block diagram of the satellite abnormal orbit change identification system according to an embodiment of the present invention is schematically shown. DETAILED DESCRIPTION

[0021] Below, embodiments of the present invention will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present invention. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of embodiments of the present invention. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of concepts of the present invention.

[0022] The terms used herein are only for describing specific embodiments and are not intended to limit the present invention. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0023] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0024] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0025] It should be noted that the method and system for identifying abnormal satellite orbit changes of the present invention can be used in the field of aerospace measurement and control technology, such as determining whether an abnormal orbit change occurs in a satellite orbit. It can also be used in any field other than the field of aerospace measurement and control technology, such as the field of artificial intelligence technology. The present invention does not limit the application field of the method and system for identifying abnormal satellite orbit changes.

[0026] During the research, it was found that by May 2024, there will be 9,900 satellites in orbit around the world, of which 84% are low-orbit satellites. However, the surge in the number of satellites has made limited orbital resources more valuable, the relative motion between satellites has become more complex, and the potential risk of collision has also increased. Therefore, for non-cooperative targets, that is, those satellites that may deviate from the intended orbit due to failure, loss of control or other reasons, it is of great practical significance to promptly detect orbital anomalies.

[0027] TLE (Two-Line Element Set) data is a specific form of orbital element representation used to describe the orbit of a celestial body. TLE contains orbital elements of satellites, such as orbital semi-major axis (a), orbital inclination (i), orbital period (T), etc., which can be used to calculate the current position, velocity and orbital evolution of celestial bodies. The current methods for detecting orbital anomalies based on the changing characteristics of the selected orbital elements are based on threshold detection technology based on statistical analysis, data fitting technology based on orbital propagation, and machine learning clustering technology.

[0028] For threshold detection technology based on statistical analysis, the square root prediction residuals of the semi-major axis and inclination of the satellite orbit are usually selected as the orbit change detection variables, and the detection threshold is generated by training the historical data of space targets with similar orbital heights, and the n-σ criterion is used to realize orbit change behavior detection. For data fitting technology based on orbit propagation, the SGP4 / SDP4 model is usually used to predict the semi-major axis time series derived from historical TLE orbit information, and the difference between the predicted orbit and the actual orbit is analyzed to realize orbit maneuver identification. For data fitting technology based on orbit propagation and machine learning clustering technology, in the related technologies, three methods, K-means, hierarchical clustering and fuzzy C-means clustering, are used to process TLE historical data for remote sensing satellites. The contour map method based on k-means is used to select feature variables and the number of clusters, which can basically meet the maneuver identification of orbits of different scales.

[0029] As for the above-mentioned related technologies, if they rely too much on the historical data of space targets with similar orbital heights. If the historical data is insufficient or biased, it will affect the accuracy of the detection threshold. Since the detection threshold is generated by training based on historical data under specific conditions, its adaptability may be limited. When the space environment or satellite orbit conditions change significantly, the original detection threshold may no longer be applicable and needs to be retrained and adjusted. There are certain errors in the SGP4 / SDP4 propagation model, which cannot distinguish between model errors, data errors or maneuvering behaviors; the error of this model will be greater as time goes by. In the machine learning method, the selection of feature variables and the number of clusters is based on the silhouette coefficient of the cluster, ignoring the correlation between variables and the characteristics of time series data, and always requires manual parameter selection, such as: manually defining how many clusters to cluster in the specified cluster, which limits the versatility and scalability of the algorithm. In addition, in order to evaluate the accuracy of the anomaly recognition method, data are usually manually labeled, such as: manually labeling a certain data as data generated by an abnormal orbit change of the satellite, but because there is no real label, the result lacks objectivity.

[0030] In view of this, an embodiment of the present invention provides a method for identifying abnormal orbit changes of satellites, including: determining multiple candidate feature data based on a target data set, wherein the candidate feature data is feature data related to the satellite's orbit; filtering the multiple candidate feature data based on variance values ​​corresponding to each of the multiple candidate feature data and correlation values ​​between each of the multiple candidate feature data to obtain target feature data; determining feature change data based on target feature sub-data corresponding to each of the multiple time periods included in the target feature data, wherein the feature change data characterizes changes in the target feature sub-data between adjacent time periods; generating orbit feature vectors corresponding to each of the multiple time periods based on the target feature data and the feature change data; clustering the orbit feature vectors corresponding to each of the multiple time periods to obtain clustering results; and determining that the satellite has an abnormal orbit change phenomenon when it is determined that the clustering results characterize the existence of abnormal feature vectors in the orbit feature vectors corresponding to each of the multiple time periods.

[0031] Figure 1 The application scenario diagram of the satellite abnormal orbit change identification method and system according to an embodiment of the present invention is schematically shown.

[0032] like Figure 1As shown, the application scenario 100 according to this embodiment may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, and a server 105. The network 104 is used to provide a medium for a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103, and the server 105. The network 104 may include various connection types, such as wired, wireless communication links, or optical fiber cables, etc.

[0033] The user can use the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 through the network 104 to receive or send messages, etc. Various communication client applications can be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only for example).

[0034] The first terminal device 101, the second terminal device 102, and the third terminal device 103 may be various electronic devices having display screens and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

[0035] The server 105 may be a server that provides various services, such as a background management server (only as an example) that provides support for websites browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103. The background management server may analyze and process the received data such as user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.

[0036] It should be noted that the satellite abnormal orbit change identification method provided in the embodiment of the present invention can generally be executed by the server 105. Accordingly, the satellite abnormal orbit change identification system provided in the embodiment of the present invention can generally be set in the server 105. The satellite abnormal orbit change identification method provided in the embodiment of the present invention can also be executed by a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105. Correspondingly, the satellite abnormal orbit change identification system provided in the embodiment of the present invention can also be set in a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105.

[0037] It should be understood that Figure 1The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0038] The following will be based on Figure 1 The scene described by Figure 2~Figure 8 The method for identifying abnormal satellite orbit changes according to an embodiment of the present invention is described in detail.

[0039] Figure 2 The flowchart of the method for identifying abnormal satellite orbit changes according to an embodiment of the present invention is schematically shown.

[0040] like Figure 2 As shown, the method includes operations S210 to S260.

[0041] In operation S210, a plurality of candidate feature data are determined based on the target data set, wherein the candidate feature data is feature data related to the satellite orbit.

[0042] In operation S220 , data filtering is performed on the plurality of candidate feature data based on the variance values ​​corresponding to the plurality of candidate feature data and the correlation values ​​between the plurality of candidate feature data to obtain target feature data.

[0043] In operation S230, feature change data is determined based on the target feature sub-data corresponding to each of the plurality of time periods and included in the target feature data, wherein the feature change data represents a change in the target feature sub-data between adjacent time periods.

[0044] In operation S240, track feature vectors corresponding to each of the plurality of time periods are generated based on the target feature data and the feature change data.

[0045] In operation S250, the track feature vectors corresponding to each of the plurality of time periods are clustered to obtain a clustering result.

[0046] In operation S260, when it is determined that the clustering result indicates that there are abnormal feature vectors in the orbit feature vectors corresponding to each of the plurality of time periods, it is determined that there is an abnormal orbit change phenomenon of the satellite.

[0047] According to an embodiment of the present invention, there is no limitation on the target data set, and it may be a data set including characteristic data related to the satellite orbit, such as a TLE data set.

[0048] According to the embodiment of the present invention, there is no limitation on the method of obtaining the TLE dataset. The TLE dataset may be obtained from a fixed disk or may be obtained in real time from a public website, such as Space-Track.

[0049] According to an embodiment of the present invention, the candidate feature data may be data included in the target data set, such as inclination data, eccentricity data, etc., or may be data calculated based on the data included in the target data set, such as semi-major axis data, eccentricity angle data, etc.

[0050] According to an embodiment of the present invention, the candidate feature data and the target feature data are both time series data, and the candidate feature data include candidate feature sub-data corresponding to each of the multiple time periods. The target feature data include target feature sub-data corresponding to each of the multiple time periods, and a time period is not limited and can be 1 day. For example, if the candidate feature data is eccentricity data, the candidate feature data includes eccentricity data on July 1, eccentricity data on July 2, etc.

[0051] According to an embodiment of the present invention, there may be multiple target feature data, and the multiple target feature data may be semi-major axis data, eccentricity data or inclination data respectively. The feature type of the semi-major axis data is semi-major axis, the feature type of the eccentricity data is eccentricity, and the feature type of the inclination data is inclination.

[0052] According to an embodiment of the present invention, the feature change data and the target feature data are in a one-to-one correspondence. For example, if there are three target feature data, namely, semi-major axis data, eccentricity data and inclination data, there is feature change data corresponding to each of the semi-major axis data, eccentricity data and inclination data.

[0053] According to an embodiment of the present invention, each feature change data may include a feature change degree sequence, a feature change speed sequence, and a feature change acceleration sequence. Each of the feature change degree sequence, the feature change speed sequence, and the feature change acceleration sequence may include a plurality of feature change sub-data, each of which is calculated based on two target feature sub-data in adjacent time periods.

[0054] According to an embodiment of the present invention, the variance value of the candidate feature data may reflect the degree of dispersion of the candidate feature data in multiple time periods, that is, the difference between a data point and its mean value, so that the importance of the candidate feature data may be evaluated.

[0055] According to an embodiment of the present invention, the orbital feature vector may be a multidimensional feature vector, and the orbital feature vectors corresponding to each of the multiple time periods may be obtained from the target feature sub-data corresponding to each of the multiple time periods and the feature change sub-data corresponding to each of the multiple time periods. For example, the orbital feature vector corresponding to July 2 may be obtained through the target feature sub-data of July 2 and the feature change sub-data representing the change of the target feature sub-data between July 2 and July 1, and in the case where the target feature data is multiple and respectively includes the semi-major axis data, the eccentricity data and the inclination data, then one orbital feature vector may include the semi-major axis data of July 2, the eccentricity data of July 2, the inclination data of July 2, the feature change sub-data representing the change of the semi-major axis data between July 2 and July 1, the feature change sub-data representing the change of the eccentricity data between July 2 and July 1, and the feature change sub-data representing the change of the inclination data between July 2 and July 1, and the above feature change sub-data are multiple, namely the change degree sub-data, the change speed sub-data and the change acceleration sub-data.

[0056] According to the embodiment of the present invention, the clustering processing method is not limited and may be a density-based spatial clustering of applications with noise (Density-Based Spatial Clustering of Applications with Noise, DBSCAN).

[0057] According to an embodiment of the present invention, DBSCAN clustering is selected for multidimensional feature variables. This method can automatically cluster data into clusters of any shape and size based on data density and spatial distribution, without the need to pre-specify the number of clusters, or manually label and define thresholds. That is, by adopting this clustering method, the spatiotemporal information of data distribution is taken into account, and noise points in the data, i.e., abnormal track change points, can be effectively identified.

[0058] According to an embodiment of the present invention, determining a plurality of candidate feature data based on a target data set may include the following operations.

[0059] Determine candidate feature data of a first preset feature type from a target data set; determine target data for generating candidate feature data of a second preset feature type from the target data set; generate candidate feature data of the second preset feature type based on the target data; aggregate the candidate feature data of the first preset feature type and the candidate feature data of the second preset feature type to obtain a plurality of candidate feature data.

[0060] According to an embodiment of the present invention, the first preset feature type may include: eccentricity, inclination, right ascension of ascending node, argument of perigee, mean anomaly, and the like.

[0061] According to an embodiment of the present invention, the second preset feature type may include: semi-major axis, true anomaly angle, etc.

[0062] According to an embodiment of the present invention, the number of bits per row and the feature types of the data included in the target data set may be as shown in Table 1.

[0063]

[0064] Table 1

[0065] According to an embodiment of the present invention, the semi-major axis data of the feature type of the candidate feature data can be calculated by the number of circles around the earth per day, specifically, as shown in the following formula (1).

[0066] ; (1)

[0067] Among them, T is the satellite's operating period, and n is the number of times the satellite orbits the earth every day.

[0068] According to an embodiment of the present invention, based on the relationship between the semi-major axis of a planetary orbit and the satellite orbit period described by Kepler's third law, the semi-major axis data can be obtained, as specifically shown in the following formulas (2) to (3).

[0069] ; (2)

[0070] ; (3)

[0071] Among them, k is the Kepler constant, M is the mass of the central celestial body, G is the gravitational constant, and a is the semi-major axis data.

[0072] According to an embodiment of the present invention, true anomaly data of a feature type of true anomaly in the candidate feature data can be calculated using mean anomaly data, specifically, as shown in the following formulas (4) to (6).

[0073] ; (4)

[0074] Among them, m is the mean anomaly angle data, E is the eccentric anomaly angle data, and e is the eccentricity data.

[0075] According to an embodiment of the present invention, the relationship between the eccentric anomaly angle data and the true anomaly angle data is as shown in formula (5), that is, the true anomaly angle data can be obtained by formula (6).

[0076] ; (5)

[0077] Among them, φ is the true anomaly data.

[0078] ; (6)

[0079] According to an embodiment of the present invention, filtering multiple candidate feature data based on variance values ​​corresponding to each of the multiple candidate feature data and correlation values ​​between the multiple candidate feature data to obtain target feature data may include the following operations.

[0080] Based on the variance values ​​corresponding to the multiple candidate feature data, the multiple candidate feature data are filtered to obtain multiple intermediate feature data; based on the correlation values ​​between the multiple intermediate feature data, the multiple intermediate feature data are filtered to obtain target feature data.

[0081] According to an embodiment of the present invention, for each candidate feature data, the variance value of the candidate feature data may be calculated by including candidate feature sub-data of multiple time periods in the candidate feature data.

[0082] According to an embodiment of the present invention, the variance values ​​corresponding to the plurality of candidate feature data are compared with a preset variance threshold, and the candidate feature data below the preset variance threshold are filtered out.

[0083] According to an embodiment of the present invention, firstly, the variance value of the candidate feature data is used to measure the degree of discreteness of the candidate feature data, that is, the difference between the data point and its mean. Secondly, a threshold is selected by setting the data distribution, and the candidate feature data below the threshold will be regarded as low-variance feature data. These feature data have small changes in samples and contribute little to distinguishing different samples. That is, it is difficult to distinguish and determine whether the satellite has an abnormal orbit change phenomenon through low-variance feature data, so it can be filtered out, thereby reducing the dimension of the data set.

[0084] According to an embodiment of the present invention, based on the correlation values ​​between the multiple intermediate feature data, filtering the multiple intermediate feature data to obtain the target feature data may include the following operations.

[0085] At least two intermediate feature data having correlation values ​​greater than or equal to a preset correlation threshold are determined from correlation values ​​between the plurality of intermediate feature data; and the intermediate feature data having a feature type identical to the preset feature type among the at least two intermediate feature data is used as the target feature data.

[0086] According to an embodiment of the present invention, the correlation values ​​between the plurality of intermediate feature data may be calculated using the following formula (7).

[0087] ; (7)

[0088] Wherein, r is the correlation value between the intermediate feature data X and the intermediate feature data Y, N is the number of data of the intermediate feature data X and the intermediate feature data Y, is the ith intermediate feature sub-data included in the intermediate feature data X, is the ith intermediate feature sub-data included in the intermediate feature data Y, is the average value of multiple intermediate feature sub-data included in the intermediate feature data X, It is the average value of multiple intermediate feature sub-data included in the intermediate feature data Y.

[0089] According to an embodiment of the present invention, taking the example that multiple intermediate feature data include intermediate feature data A, intermediate feature data B and intermediate feature data C, the "correlation values ​​between each of the multiple intermediate feature data" may include the correlation value between the intermediate feature data A and the intermediate feature data B, the correlation value between the intermediate feature data B and the intermediate feature data C, and the correlation value between the intermediate feature data A and the intermediate feature data C.

[0090] According to an embodiment of the present invention, the preset feature type can be obtained based on the six elements of the artificial satellite orbit. The six elements of the artificial satellite orbit are parameters used to characterize the properties of the satellite orbit such as shape, position and motion, also known as the six orbital elements, which can be used to determine the orbit and position of the satellite at any time. Specifically, the preset feature type may include semi-major axis, eccentricity, orbital inclination, periapsis argument, ascending node right ascension and true anomaly.

[0091] According to the embodiment of the present invention, the preset correlation threshold is not limited and can be set according to actual conditions. For example, it is set to 0.6 based on multiple experiments.

[0092] According to an embodiment of the present invention, at least two intermediate feature data having correlation values ​​greater than or equal to a preset correlation threshold are screened out, and intermediate feature data having a feature type that is the same as a preset feature type is determined from the at least two feature data as target feature data. Intermediate feature data with a high correlation with other intermediate feature data can be screened out, and intermediate feature data of a preset data type that can better characterize attributes such as satellite orbit shape, position and motion can be selected as target feature data. Feature dimensionality reduction can be achieved again, and redundant feature data can be removed.

[0093] According to an embodiment of the present invention, the target feature data may be multiple and the target feature data may be semi-major axis data, eccentricity data and inclination data, that is, subsequent analysis is performed through the multiple target feature data, which contains more information than single variable analysis and removes redundant information compared to using all variables.

[0094] Figure 3 A schematic diagram schematically shows correlation values ​​between multiple intermediate feature data according to an embodiment of the present invention.

[0095] like Figure 3As shown, the correlation values ​​between multiple intermediate feature data are calculated by the above formula to form a Pearson correlation coefficient matrix, including the mean angular velocity data, namely mean_motion, the eccentricity data, namely eccentricity, the inclination data, namely inclination, the right ascension data of the ascending node, namely ra_of_asc_node, the true anomaly data, namely arg_of_pericenter, the mean anomaly data, namely mean_anomaly, the number of circles around the earth per day, namely rev_at_epoch, the BSTAR coefficient data, namely bstar, the derivative data of the mean motion, namely mean_motion_dot, the semi-major axis data, namely semimajor_axis, the period data, namely period, the perigee data, namely apoapsis, and the apogee data, namely periapsis. When the preset correlation threshold is 0.6, it can be determined that at least two intermediate feature data whose correlation values ​​are greater than or equal to the preset correlation threshold are: eccentricity, apoapsis, inclination, rev_at_epoch, bstar, mean_motion_dot, semimajor_axis, period, periapsis, and apoapsis.

[0096] According to an embodiment of the present invention, the feature types of the at least two intermediate feature data are compared with preset feature types, and the determined target feature data are multiple: semi-major axis data semimajor_axis, eccentricity data eccentricity and inclination data inclination.

[0097] According to an embodiment of the present invention, a time period includes at least one time point; a time point corresponds one-to-one to a target feature sub-data; determining feature change data based on the target feature sub-data included in the target feature data and corresponding to each of the multiple time periods may include the following operations.

[0098] For each time period, determine the latest time point in the time period and the target feature sub-data corresponding to the latest time point; delete the target feature sub-data except the target feature sub-data corresponding to the latest time point to obtain new target feature data; determine the feature change data based on the target feature sub-data corresponding to each of the multiple time periods included in the new target feature data.

[0099] According to an embodiment of the present invention, the target characteristic sub-data corresponding to a time period may include target characteristic sub-data corresponding to at least one time point, and one time point may correspond one-to-one to one target characteristic sub-data. For example, the eccentricity data on July 1 may include: the eccentricity data at 5 o'clock on July 1...the eccentricity data at 14 o'clock on July 1, etc.

[0100] According to an embodiment of the present invention, the target feature data can be sorted first, and only the latest target feature sub-data at the latest time point is retained in a time period, which usually represents the latest orbital state of the satellite on that day. In this way, a time series data set containing only one latest record per day can be constructed, that is, new target feature data.

[0101] According to an embodiment of the present invention, since TLE data usually comes from different observation sources, their collection frequencies are often inconsistent, resulting in the target data set having multiple records in one day or no records at all on some days. Therefore, before determining multiple candidate feature data based on the target data set, the missing rate of the data can be detected.

[0102] According to an embodiment of the present invention, the missing rate of the target data set is obtained by calculating the ratio of missing days in the target data set to the total number of days. The target data set whose missing rate exceeds a preset missing rate threshold is regarded as an unreliable data set and is excluded from the analysis scope to ensure the accuracy and reliability of the analysis results.

[0103] According to an embodiment of the present invention, the feature change data includes a feature change degree sequence, a feature change speed sequence and a feature change acceleration sequence; the feature change degree sequence includes multiple change degree sub-data; the feature change speed sequence includes multiple change speed sub-data; the feature change acceleration sequence includes multiple change acceleration sub-data; determining the feature change data based on the target feature sub-data corresponding to each of the multiple time periods included in the new target feature data may include the following operations.

[0104] Based on the target characteristic sub-data of the i-th time period and the target characteristic sub-data of the i+1-th time period, determining the change degree sub-data of the target characteristic sub-data between the i-th time period and the i+1-th time period, wherein i is a positive integer greater than zero; based on the multiple change degree sub-data, obtaining a characteristic change degree sequence; based on the target characteristic sub-data of the i-th time period, the target characteristic sub-data of the i+1-th time period, the time difference between the latest time point of the i-th time period and the latest time point of the i+1-th time period, determining the change rate of the target characteristic sub-data between the i-th time period and the i+1-th time period Based on the change speed sub-data of the target characteristic sub-data between the i-th time period and the i+1-th time period, the change acceleration sub-data of the target characteristic sub-data between the i-th time period and the i+1-th time period are determined; based on the multiple change acceleration sub-data, a characteristic change acceleration sequence is obtained; based on the characteristic change degree sequence, the characteristic change speed sequence and the characteristic change acceleration sequence, the characteristic change data is obtained.

[0105] According to an embodiment of the present invention, the feature change degree sequence includes a plurality of change degree sub-data, and the change degree sub-data may be a percentage of change of the target feature sub-data of the current time period relative to the previous time period, that is, it reflects the fluctuation of the target feature sub-data, i.e., the orbital parameters in a short period of time, which may be specifically shown as formula (8).

[0106] ; (8)

[0107] Among them, C t is the sub-data of the degree of change corresponding to the current time period t, is the target feature sub-data of the current time period t, It is the target feature sub-data of the previous time period t-1 of the current time period.

[0108] According to an embodiment of the present invention, the characteristic change speed sequence may include a plurality of change speed sub-data, and the change speed sub-data characterizes the ratio of the change degree to the time interval, and reveals how fast the target characteristic sub-data, i.e., the orbital parameters, change over time. Specifically, as shown in the following formula (9).

[0109] ; (9)

[0110] Among them, V t is the change speed sub-data corresponding to the current time period t, is the difference between the latest time point in the current time period t and the latest time point in the previous time period t-1, It can be measured in units of one day.

[0111] According to an embodiment of the present invention, the characteristic change acceleration sequence may include a plurality of change acceleration sub-data, and the measurement speed change explains the acceleration of the target characteristic sub-data, i.e., the orbit parameter changing with time, as specifically shown in the following formula (10).

[0112] ; (10)

[0113] Among them, a t is the acceleration sub-data corresponding to the current time period t, V t-1 is the change speed sub-data corresponding to the previous time period t-1.

[0114] According to an embodiment of the present invention, through the above-mentioned characteristic change degree sequence, characteristic change speed sequence and characteristic change acceleration sequence, new characteristics of each orbital element changing with time can be characterized on the basis of the original data, that is, the target characteristic data. That is, the above-mentioned characteristic change degree sequence, characteristic change speed sequence and characteristic change acceleration sequence contain the time information hidden in the data, thereby improving the detection accuracy of whether the satellite has an abnormal orbit change phenomenon in the subsequent satellite abnormal orbit change analysis and reducing the detection error.

[0115] According to an embodiment of the present invention, clustering the orbit feature vectors corresponding to each of the plurality of time periods to obtain a clustering result may include the following operations.

[0116] The orbital feature vectors corresponding to each of the multiple time periods are mapped into the vector space to obtain multiple space points; based on a preset neighborhood range and the number of neighboring points of each space point in the preset neighborhood range, at least one core space point is determined from the multiple space points; for each core space point, based on the core space point and the neighboring points of the core space point, the space points with density reachable between each other are determined; the density reachable space points are connected to obtain a clustering result, wherein the clustering result includes noise points and / or clusters.

[0117] According to an embodiment of the present invention, the orbital feature vectors corresponding to each of the plurality of time periods are clustered, and specifically, the orbital feature vectors corresponding to each of the plurality of time periods are clustered in a density-based manner with noise.

[0118] According to an embodiment of the present invention, the parameters eps and MinPts can be determined first. EPS is the neighborhood radius, which specifies the size of the neighborhood of a spatial point. MinPts is the minimum number of points in the neighborhood, which specifies the minimum number of neighbors required for a spatial point to become a core point in its eps-neighborhood, including the point itself. The eps-neighborhood of all spatial points can be calculated, and the core spatial points can be marked. Specifically, for each point P in the data set, the number of neighboring points in its eps-neighborhood is calculated. If the number of points in the eps-neighborhood of a spatial point is greater than or equal to MinPts, then the point is marked as a core spatial point. Density-reachable spatial points can be found. Specifically, for each core spatial point, a clustering cluster is generated by finding a spatial point whose density is reachable from its starting point. If point P is in the eps-neighborhood of point O, and O is a core point, then P is said to be directly reachable by O density. For points P and Q, if there is a sequence P1, P2, ... Pn, where P1 = P, Pn = Q, and Pi+1 is directly reachable by Pi density, then P is reachable by Q density.

[0119] According to an embodiment of the present invention, clusters are constructed. For points P and Q, if there exists a point O such that both P and Q are density-reachable by O, then P and Q are density-connected. Clusters can be formed by grouping together all density-connected spatial points. These clusters are the largest set of points containing at least one core point, and any two points in the set are density-connected.

[0120] According to an embodiment of the present invention, clustering results can be obtained by connecting density-reachable spatial points. After the clusters are constructed, points in the data set that do not belong to any cluster will be marked as noise points, which are neither core spatial points nor density-reachable points of any core spatial points. That is, the clustering results are constructed clusters and / or noise points that cannot be clustered with other spatial points.

[0121] According to an embodiment of the present invention, when it is determined that the clustering result includes noise points, it is considered that there are abnormal feature vectors in the orbital feature vector, that is, it can be determined that the satellite has an abnormal orbit change phenomenon. Conversely, when it is determined that the clustering result does not include noise points, it is considered that there are no abnormal feature vectors in the orbital feature vector, that is, it can be determined that the satellite has no abnormal orbit change phenomenon.

[0122] According to an embodiment of the present invention, a spatial clustering of density of multidimensional feature variables, namely, dbscan clustering, is selected. This method can automatically cluster data into clusters of any shape and size based on data density and spatial distribution, without pre-specifying the number of clusters, or manually labeling labels and defining thresholds. That is, the above clustering process can overcome the disadvantage that distance-based clustering can only find "spherical" clusters. This method takes into account the spatiotemporal information of data distribution. The density is defined by the number of points within a specified radius neighborhood of a point. Data points can be divided into three categories according to the density. If a point's eps-neighborhood contains at least MinPts points, the point is a core point. If a point is not a core point, but is within the eps-neighborhood of a core point, the point is a boundary point. If a point is neither a core point nor a boundary point, it is regarded as a noise point, so that the noise points in the data can be effectively identified, that is, the inspection of whether the satellite has abnormal orbit change is realized.

[0123] According to an embodiment of the present invention, the above-mentioned method for identifying abnormal orbit changes of satellites may further include the following operation: taking the target time period corresponding to the abnormal feature vector as the time period when the satellite has the abnormal orbit change phenomenon.

[0124] Figure 4 The schematic diagram schematically shows a clustering result according to an embodiment of the present invention.

[0125] like Figure 4 As shown, taking GPS satellite NAVSTAR 77 (USA 289) with NORAD ID: 43874 as an example, using its data from 2022-2023 for DBSCAN clustering processing, in order to better visualize the clustering results, a visualization algorithm is used to map the data points in the high-dimensional space to the two-dimensional space, for example: t-SNE (t-Distributed StochasticNeighbor Embedding) is used, which is a nonlinear dimensionality reduction technology for high-dimensional data visualization. In some embodiments, t-SNE can be used to map data points in the high-dimensional space to the two-dimensional space while maintaining the similarity between the data points.

[0126] According to an embodiment of the present invention, Figure 4 It shows clear boundaries between clusters and between clusters and noise points. Clusters are composed of relatively dense and circular points, which are clusters generated by DBSCAN clustering processing. Plus-shaped points are noise points identified in the clusters, namely outliers. It can be seen that the noise points are significantly separated and do not belong to any major cluster. This shows that DBSCAN clustering processing can effectively distinguish normal data from abnormal data.

[0127] According to an embodiment of the present invention, the candidate feature data includes candidate feature sub-data corresponding to multiple time periods; the above-mentioned satellite abnormal orbit change identification method may also include the following operations.

[0128] For each candidate feature data, determine the candidate feature sub-data corresponding to the target time period and the candidate feature sub-data corresponding to the time period before the target time period from the candidate feature data; obtain a first motion trajectory corresponding to the target time period based on the candidate feature sub-data corresponding to the target time period and the celestial dynamics model; obtain a second motion trajectory corresponding to the time period before the target time period based on the candidate feature sub-data corresponding to the time period before the target time period and the celestial dynamics model; and verify the abnormal orbit change phenomenon based on the first motion trajectory and the second motion trajectory.

[0129] According to the embodiment of the present invention, the astrodynamic model is not limited and can be any model that can simulate the trajectory of a satellite, for example, the SGP4 / SDP4 model, where SGP4 (Simplified General Perturbations) is a simplified general perturbation model that can be applied to near-Earth objects with an orbital period of less than 225 minutes. SDP4 (Simplified Deep Space Perturbations) is a simplified deep space perturbation model that can be applied to objects far from the Earth or with an orbital period of more than 225 minutes.

[0130] According to an embodiment of the present invention, firstly, feature data to be input into the celestial dynamics model may be determined from a plurality of candidate feature data.

[0131] According to an embodiment of the present invention, there may be multiple feature data to be input, which may be orbital altitude data, inclination data, eccentricity data, etc., which are candidate feature data for simulating the satellite trajectory.

[0132] According to an embodiment of the present invention, when the target time period is i, the previous time period of the target time period is i-1. For example, when the target time period is July 2, 2022, the previous time period of the target time period is July 1, 2022.

[0133] According to an embodiment of the present invention, the feature data to be input includes candidate feature sub-data of multiple time periods, and the candidate feature sub-data corresponding to the target time period and the candidate feature sub-data corresponding to the previous time period of the target time period can be determined from the feature data to be input.

[0134] According to an embodiment of the present invention, since the target feature data is preprocessed before clustering processing, only the target feature sub-data related to the latest time point included in the target time period is retained. Therefore, the time point when the abnormal orbit change phenomenon occurs can be the latest time point. By inputting the candidate feature sub-data corresponding to the latest time point of the target time period into the celestial body dynamics model, trajectory data of a satellite orbit period with the latest time point of the target time period as the starting point can be obtained, that is, multiple first satellite position coordinates. By connecting the multiple first satellite position coordinates, the first motion trajectory of the satellite corresponding to the target time period can be obtained. The first satellite position coordinates can be recorded as (x, y, z).

[0135] According to an embodiment of the present invention, a target time point in a time period last than a target time period that differs from the latest time point of the target time period by one satellite orbit cycle, or approximately one satellite orbit cycle, can be used as the time point at which the candidate feature sub-data corresponding to the time period last than the target time period is located. By inputting the candidate feature sub-data corresponding to the target time point of the last time period into the celestial body dynamics model, a plurality of second satellite position coordinates corresponding to the target time point of the last time period can be obtained. By connecting the plurality of second satellite position coordinates, a second motion trajectory of the satellite corresponding to the target time period can be obtained. Since the candidate feature sub-data of each time point in a time period are not much different, if there is no target time point in the target data set that differs from the latest time point of the target time period by one satellite orbit cycle, the time point closest to the target time point can be used as the target time point.

[0136] According to an embodiment of the present invention, by comparing the first motion trajectory with the second motion trajectory, it can be more intuitively determined whether there is an abnormal orbit change phenomenon of the satellite in the target time period.

[0137] According to an embodiment of the present invention, the celestial dynamics model is mainly based on the orbital parameters of the satellite, namely candidate characteristic data, such as orbital altitude data, orbital inclination data, orbital eccentricity data, etc., and takes into account the influence of time dependence and various perturbation factors. The accurate position information of the satellite at a specified time point can be calculated, so as to obtain the 3D trajectory coordinate data of the satellite within a satellite orbit period at a specific time, and a trajectory diagram can be drawn based on the coordinate data.

[0138] According to an embodiment of the present invention, taking the satellite with NORAD ID 43874 as an example, the time point of abnormal orbit change is marked on the time series curve graph of semi-major axis data, inclination data and eccentricity data, as shown in Figure 5 (a) to Figure 7 (b). It can be seen from Figure 5 (a) to Figure 7 (b) that the noise points are mostly sudden turning points of curve changes, and the points with large and small changes can be accurately identified. According to Figure 5 (a) to Figure 7 (b), typical noise points are selected to construct the first motion trajectory and the second motion trajectory, so as to verify the abnormal satellite orbit change phenomenon, as shown in Figure 5 (b) to Figure 7 (b).

[0139] FIG5(a) schematically shows a semi-major axis variation curve diagram constructed based on an abnormal feature vector according to an embodiment of the present invention; FIG5(b) schematically shows a first schematic diagram of a first motion trajectory and a second motion trajectory according to an embodiment of the present invention.

[0140] As shown in Figure 5 (a), the semi-major axis change curve of the satellite with NORAD ID 43874 is taken as an example. The semi-major axis data is semimajor_axis, and the noise points can also be called outliers. The semi-major axis change curve shows the change of the semi-major axis data from January 2022 to January 2023. The noise points intersecting with the dotted line are the points corresponding to 2022-07-1910:24:06. As shown in Figure 5 (b), the trajectory comparison before and after 2022-07-19 10:24:06 is shown. The dotted line part is the second motion trajectory generated based on the candidate feature sub-data corresponding to 2022-07-18 11:22:34, and the solid line part is the first motion trajectory generated based on the candidate feature sub-data corresponding to 2022-07-19 10:24:06 after the satellite changes its orbit.

[0141] According to an embodiment of the present invention, according to FIG. 5 (a), the semi-major axis data of the satellite with NORAD ID 43874 decreased by 50 km from July 18 to July 19, 2022. Correspondingly, in FIG. 5 (b), a cycle of the satellite's orbit change point before and after is generated according to the astrodynamic model. It can be seen from FIG. 5 (b) that the semi-major axis data of the first motion trajectory corresponding to 2022-07-19 is relatively small, and the corresponding trajectory is located on the inside. The change in the semi-major axis data reflects the orbital change in the orbital plane. Since the size of the satellite's semi-major axis has exceeded 40,000 km, this figure uses larger horizontal and vertical coordinate units to more comprehensively display the data trend. Despite this, the relative changes between the trajectories are still significant.

[0142] FIG6 (a) schematically shows a graph of a change in inclination angle constructed according to an abnormal feature vector according to an embodiment of the present invention; FIG6 (b) schematically shows a second schematic diagram of a first motion trajectory and a second motion trajectory according to an embodiment of the present invention;

[0143] In Figure 6 (a), the inclination change curve diagram of the satellite with NORAD ID 43874 is taken as an example. The inclination data is inclination, and the noise points are outliers. The inclination change curve diagram shows the change of inclination data from January 2022 to January 2023. The points intersecting with the dotted line are the noise points corresponding to 2022-07-05 10:25:59. In Figure 6 (b), the trajectory comparison before and after 2022-07-05 10:25:59 is shown. The dotted part is the second motion trajectory generated based on the candidate feature sub-data corresponding to 2022-07-04 10:40:57, and the satellite trajectory of the solid line part is the first motion trajectory generated based on the candidate feature sub-data corresponding to 2022-07-05 10:25:59. In Figure 6(a), the satellite undergoes a sudden change on July 5, 2022 compared to the data point on July 4. Correspondingly, in Figure 6(b), it can be seen that the inclination of the satellite's orbital plane has changed significantly.

[0144] FIG. 7( a ) schematically shows an eccentricity variation curve diagram constructed based on an abnormal feature vector according to an embodiment of the present invention; FIG. 7( b ) schematically shows a third schematic diagram of a first motion trajectory and a second motion trajectory according to an embodiment of the present invention.

[0145] In Figure 7 (a), the eccentricity change curve diagram of the satellite with NORAD ID 43874 is taken as an example. The eccentricity data is eccentricity, and the noise points are outliers. The eccentricity change curve diagram shows the change of eccentricity data from January 2022 to January 2023. The points intersecting with the dotted line are the noise points corresponding to 2022-11-04 11:29:29. In Figure 7 (b), the trajectory comparison before and after 2022-11-04 11:29:29 is shown. The dotted line part is the second motion trajectory generated based on the candidate feature sub-data corresponding to 2022-11-03 10:47:10, and the solid line part is the first motion trajectory generated based on the candidate feature sub-data corresponding to 2022-11-04 11:29:29. In Figure 7(a), the eccentricity of the satellite suddenly increased from November 3 to November 4, 2022. Correspondingly, in Figure 7(b), it can be seen that the shape of the satellite's orbit has changed significantly. The larger the eccentricity data, the flatter the ellipse becomes.

[0146] Figure 8A flowchart of a method for identifying abnormal satellite orbit changes according to another embodiment of the present invention is schematically shown.

[0147] like Figure 8 As shown, the method includes operations S801 to S809.

[0148] In operation S801 , a missing rate calculation is performed on a target data set to obtain a missing rate corresponding to the target data set.

[0149] In operation S802, it is determined whether the missing rate is greater than or equal to a preset missing rate threshold. If the missing rate is greater than or equal to the preset missing rate threshold, operation S809 is performed, and if the missing rate is less than the preset missing rate threshold, operation S803 is performed.

[0150] In operation S803, a plurality of candidate feature data are determined based on the target data set, wherein the candidate feature data is feature data related to the satellite orbit.

[0151] In operation S804, the plurality of candidate feature data are filtered based on the variance values ​​corresponding to the plurality of candidate feature data and the correlation values ​​between the plurality of candidate feature data to obtain target feature data, wherein the target feature data includes target feature sub-data corresponding to the plurality of time periods.

[0152] In operation S805 , for each time period, the target feature sub-data except the target feature sub-data corresponding to the latest time point are deleted to obtain new target feature data.

[0153] In operation S806, feature change data is determined based on the target feature sub-data corresponding to each of the plurality of time periods and included in the new target feature data.

[0154] In operation S807, the track feature vectors corresponding to each of the multiple time periods are subjected to density-based clustering with noise to obtain a clustering result.

[0155] In operation S808, when it is determined that the clustering result indicates that there are abnormal feature vectors in the orbit feature vectors corresponding to each of the plurality of time periods, it is determined that the satellite has an abnormal orbit change phenomenon.

[0156] In operation S809 , a new target data set is acquired, and the process returns to operations S801 to S808 .

[0157] According to an embodiment of the present invention, in the absence of a new target data set, the identification of abnormal orbit change phenomena for the satellite is suspended.

[0158] Based on the above satellite abnormal orbit change identification method, the present invention also provides a satellite abnormal orbit change identification system. Fig. 9The system is described in detail.

[0159] Fig. 9 The structure block diagram of the satellite abnormal orbit change identification system according to an embodiment of the present invention is schematically shown.

[0160] like Fig. 9 As shown, the satellite abnormal orbit change identification system 900 of this embodiment includes a first determination module 910, a filtering module 920, a second determination module 930, a generation module 940, a processing module 950 and a phenomenon determination module 960.

[0161] The first determination module 910 is used to determine a plurality of candidate feature data based on the target data set, wherein the candidate feature data is feature data related to the satellite orbit.

[0162] The filtering module 920 is used to filter the plurality of candidate feature data based on the variance values ​​corresponding to the plurality of candidate feature data and the correlation values ​​between the plurality of candidate feature data to obtain target feature data.

[0163] The second determination module 930 is used to determine feature change data based on the target feature sub-data corresponding to each of the multiple time periods included in the target feature data, wherein the feature change data represents the change of the target feature sub-data between adjacent time periods.

[0164] The generating module 940 is used to generate track feature vectors corresponding to each of the multiple time periods based on the target feature data and the feature change data.

[0165] The processing module 950 is used to perform clustering processing on the orbit feature vectors corresponding to each of the multiple time periods to obtain a clustering result.

[0166] The phenomenon determination module 960 is used to determine that the satellite has an abnormal orbit change phenomenon when it is determined that the clustering result indicates that there are abnormal feature vectors in the orbit feature vectors corresponding to each of the multiple time periods.

[0167] According to an embodiment of the present invention, the program code for executing the computer program provided by the embodiment of the present invention can be written in any combination of one or more programming languages. Specifically, these computing programs can be implemented using high-level process and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, Java, C++, python, "C" language or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on the remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (e.g., using an Internet service provider to connect through the Internet).

[0168] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present invention. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0169] It will be appreciated by those skilled in the art that the features described in the various embodiments of the present invention may be combined and / or combined in various ways, even if such combinations or combinations are not explicitly described in the present invention. In particular, without departing from the spirit and teachings of the present invention, the features described in the various embodiments of the present invention may be combined and / or combined in various ways. All of these combinations and / or combinations fall within the scope of the present invention.

[0170] The embodiments of the present invention are described above. However, these embodiments are only for the purpose of illustration, and are not intended to limit the scope of the present invention. Although each embodiment is described above, it does not mean that the measures in each embodiment cannot be used in combination advantageously. Without departing from the scope of the present invention, those skilled in the art may make various substitutions and modifications, which should all fall within the scope of the present invention.

Claims

1. A method for identifying abnormal satellite orbit changes, characterized in that: The method comprises: Based on the target data set, a plurality of candidate feature data are determined, wherein the candidate feature data are feature data related to the satellite orbit; Based on the variance values ​​corresponding to the plurality of candidate feature data and the correlation values ​​between the plurality of candidate feature data, filtering the plurality of candidate feature data to obtain target feature data; Determine feature change data based on target feature sub-data corresponding to each of the multiple time periods included in the target feature data, wherein the feature change data represents a change of the target feature sub-data between adjacent time periods, and the feature change data includes a feature change degree sequence, a feature change speed sequence, and a feature change acceleration sequence; Based on the target feature data and the feature change data, generating a track feature vector corresponding to each of the plurality of time periods; Clustering the orbital feature vectors corresponding to each of the plurality of time periods to obtain a clustering result; When it is determined that the clustering result indicates that there are abnormal feature vectors in the orbit feature vectors corresponding to each of the plurality of time periods, it is determined that the satellite has an abnormal orbit change phenomenon; Wherein, the time period includes at least one time point; the time point corresponds to the target characteristic sub-data one by one; and the determining of the characteristic change data based on the target characteristic sub-data corresponding to each of the multiple time periods included in the target characteristic data includes: For each time period, Determine the latest time point in the time period and the target characteristic sub-data corresponding to the latest time point; Deleting the target feature sub-data except the target feature sub-data corresponding to the latest time point to obtain new target feature data; The characteristic change data is determined based on the target characteristic sub-data corresponding to each of the plurality of time periods and included in the new target characteristic data.

2. The method according to claim 1, characterized in that The method of filtering the plurality of candidate feature data based on the variance values ​​corresponding to the plurality of candidate feature data and the correlation values ​​between the plurality of candidate feature data to obtain the target feature data includes: Based on the variance values ​​corresponding to the plurality of candidate feature data respectively, filtering the plurality of candidate feature data to obtain a plurality of intermediate feature data; Based on the correlation values ​​between the plurality of intermediate feature data, data filtering is performed on the plurality of intermediate feature data to obtain the target feature data.

3. The method according to claim 2, characterized in that The filtering of the plurality of intermediate feature data based on the correlation values ​​between the plurality of intermediate feature data to obtain the target feature data includes: According to the correlation values ​​between the plurality of intermediate feature data, at least two intermediate feature data having correlation values ​​greater than or equal to a preset correlation threshold are determined among the plurality of intermediate feature data; The intermediate feature data of the at least two intermediate feature data whose feature type is the same as the preset feature type is used as the target feature data.

4. The method according to claim 1, characterized in that: The characteristic change speed sequence includes a plurality of change speed sub-data; the characteristic change acceleration sequence includes a plurality of change acceleration sub-data; The determining of the characteristic change data based on the target characteristic sub-data corresponding to each of the multiple time periods included in the new target characteristic data includes: Determine the sub-data of the degree of change of the target feature sub-data between the i-th time period and the i+1-th time period based on the target feature sub-data of the i-th time period and the target feature sub-data of the i+1-th time period, wherein i is a positive integer greater than zero; Based on the plurality of the change degree sub-data, obtaining the feature change degree sequence; Determine the change speed sub-data of the target characteristic sub-data between the i-th time period and the i+1-th time period based on the target characteristic sub-data of the i-th time period, the target characteristic sub-data of the i+1-th time period, and the time difference between the latest time point of the i-th time period and the latest time point of the i+1-th time period; Based on the plurality of the change speed sub-data, obtaining the characteristic change speed sequence; Determine the change acceleration sub-data of the target characteristic sub-data between the i-th time period and the i+1-th time period based on the change speed sub-data of the target characteristic sub-data between the i-th time period and the i+1-th time period, the change speed sub-data of the target characteristic sub-data between the i-th time period and the i-1-th time period, and the time difference; Based on the plurality of the change acceleration sub-data, obtaining the characteristic change acceleration sequence; The feature change data is obtained based on the feature change degree sequence, the feature change speed sequence and the feature change acceleration sequence.

5. The method according to claim 1, characterized in that The method further comprises: The target time period corresponding to the abnormal feature vector is used as the time period during which the satellite has an abnormal orbit change phenomenon.

6. The method according to claim 5, characterized in that The candidate feature data includes candidate feature sub-data corresponding to a plurality of time periods; the method further includes: For each of the candidate feature data, determining, from the candidate feature data, candidate feature sub-data corresponding to the target time period and candidate feature sub-data corresponding to a time period before the target time period; Based on the candidate feature sub-data corresponding to the target time period and the celestial dynamics model, a first motion trajectory corresponding to the target time period is obtained; Based on the candidate feature sub-data corresponding to the previous time period of the target time period and the celestial body dynamics model, a second motion trajectory corresponding to the previous time period of the target time period is obtained; Based on the first motion trajectory and the second motion trajectory, the abnormal track change phenomenon is verified.

7. The method according to claim 1, characterized in that The clustering process of the orbital feature vectors corresponding to each of the plurality of time periods to obtain a clustering result includes: Mapping the orbital feature vectors corresponding to each of the plurality of time periods into a vector space to obtain a plurality of space points; Determine at least one core spatial point from a plurality of spatial points based on a preset neighborhood range and the number of neighboring points of each spatial point in the preset neighborhood range; For each of the core space points, based on the core space point and neighboring points of the core space point, determine the space points that are density-reachable between each other; The density-reachable spatial points are connected to obtain the clustering result, wherein the clustering result includes noise points and / or clusters.

8. The method according to claim 1, characterized in that The characteristic types of the target characteristic data include semi-major axis, eccentricity or inclination; the target data set includes two rows of orbital element sets.

9. A satellite abnormal orbit change identification system, characterized in that: The system comprises: A first determination module is used to determine a plurality of candidate feature data based on a target data set, wherein the candidate feature data is feature data related to a satellite orbit; A filtering module, configured to filter the plurality of candidate feature data based on the variance values ​​corresponding to the plurality of candidate feature data and the correlation values ​​between the plurality of candidate feature data to obtain target feature data; A second determination module is used to determine feature change data based on target feature sub-data corresponding to each of the multiple time periods included in the target feature data, wherein the feature change data represents the change of the target feature sub-data between adjacent time periods, and the feature change data includes a feature change degree sequence, a feature change speed sequence, and a feature change acceleration sequence; A generating module, configured to generate a track feature vector corresponding to each of the plurality of time periods based on the target feature data and the feature change data; A processing module, used for clustering the orbital feature vectors corresponding to each of the plurality of time periods to obtain a clustering result; A phenomenon determination module, configured to determine that the satellite has an abnormal orbit change phenomenon when it is determined that the clustering result indicates that there are abnormal feature vectors in the orbit feature vectors corresponding to each of the plurality of time periods; Wherein, the time period includes at least one time point; the time point corresponds to the target characteristic sub-data one by one; the second determination module is used for: For each time period, determining the latest time point in the time period and the target characteristic sub-data corresponding to the latest time point; Deleting the target feature sub-data except the target feature sub-data corresponding to the latest time point to obtain new target feature data; The characteristic change data is determined based on the target characteristic sub-data corresponding to each of the plurality of time periods and included in the new target characteristic data.