A low earth orbit satellite orbit prediction method based on feature screening and related device

Abstract

The application discloses a low-orbit satellite orbit prediction method based on feature screening and a related device, and the method comprises the following steps: defining a pseudo-drag coefficient according to an atmospheric drag perturbation formula; using a SHAP method to analyze the influence value of different space environment features on the pseudo-drag coefficient prediction result of a target prediction model, and generating an influence bee colony diagram; screening out an optimal feature combination based on the influence bee colony diagram, and predicting the pseudo-drag coefficient according to the optimal feature combination and the target prediction model to obtain a pseudo-drag prediction value; and predicting and calculating the low-orbit satellite orbit state according to the pseudo-drag prediction value to obtain an orbit prediction result. This process does not depend on the orbit prediction error and is not affected by the uncertainty of the divergence direction of the error. Therefore, the application can solve the technical problem that the existing orbit prediction depends on the prediction error, but the divergence direction of the error relative to the true atmospheric drag is uncertain, resulting in the lack of accuracy and reliability of the prediction result.

Description

Technical Field

[0001] This application relates to the field of satellite technology, and in particular to a method and related apparatus for predicting the orbit of low-Earth orbit satellites based on feature screening. Background Technology

[0002] With the rapid development of space technology and the increasing complexity of satellite missions, orbit prediction technology plays a crucial role in maintaining the operational safety of space targets and optimizing orbit management. Current orbit prediction methods for space targets are based on a framework of orbit determination and prediction. The orbit determination process utilizes observational data to obtain accurate historical orbital data and estimate relevant parameters. However, in the orbit prediction stage, observational data is unavailable, necessitating the use of the same physical model and estimated atmospheric drag coefficients as in the orbit determination process to achieve orbit prediction.

[0003] Low-Earth orbit (LEO) satellites, due to their low orbital altitude, are subject to the combined effects of conservative forces such as Earth's gravity and third-body gravity, as well as non-conservative forces such as atmospheric drag. The accuracy of their orbit predictions is highly susceptible to environmental uncertainties, with atmospheric drag being a major source of error due to its significant influence from the space environment. In medium- and long-term orbit predictions, errors accumulate significantly over time.

[0004] Existing orbit prediction methods based on machine learning modeling usually require the orbit prediction error to follow a certain rule, such as the fluctuation of the prediction error amplitude or the direction of error divergence. However, the orbital conditions of low-Earth orbit satellites are difficult to predict accurately in the real space environment, so the direction of error divergence relative to real atmospheric drag is uncertain, and the prediction results lack accuracy and reliability. Summary of the Invention

[0005] This application provides a feature-based method and related apparatus for predicting the orbit of low-Earth orbit satellites, which addresses the technical problem that existing orbit predictions rely on prediction errors, but the divergence direction of these errors relative to the actual atmospheric drag is uncertain, resulting in a lack of accuracy and reliability in the prediction results.

[0006] In view of this, the first aspect of this application provides a method for predicting the orbit of low-Earth orbit satellites based on feature screening, comprising:

[0007] The pseudo-drag coefficient is defined based on the atmospheric drag perturbation formula;

[0008] The SHAP method is used to analyze the influence of different spatial environment characteristics on the pseudo drag coefficient prediction results of the target prediction model, and an influence beehive diagram is generated.

[0009] The optimal feature combination is selected based on the influence bee colony diagram, and the pseudo resistance coefficient is predicted according to the optimal feature combination and the target prediction model to obtain the pseudo resistance prediction value.

[0010] The orbital state of the low-Earth orbit satellite is predicted and calculated based on the pseudo-drag prediction value to obtain the orbital prediction result.

[0011] Preferably, the step of using the SHAP method to analyze the influence of different spatial environment characteristics on the pseudo drag coefficient prediction results of the target prediction model and generating an influence bee colony diagram further includes:

[0012] The components of multiple measured spatial environment features are combined into a spliced ​​feature vector;

[0013] Based on the spliced ​​feature vector and the preset pseudo drag coefficient, a pseudo drag coefficient prediction model is constructed to obtain the target prediction model.

[0014] Preferably, the step of constructing a pseudo-drag coefficient prediction model based on the spliced ​​feature vector and the preset pseudo-drag coefficient to obtain the target prediction model further includes:

[0015] The orbital state of low-Earth orbit satellites is predicted using numerical integration, and the predicted ephemeris is obtained.

[0016] Traditional forecast errors are calculated based on the orbital dynamics model and the predicted ephemeris.

[0017] The preset pseudo drag coefficient is obtained by iteratively optimizing and analyzing the traditional forecast error using the golden section search algorithm.

[0018] Preferably, the step of predicting the orbital state of the low-Earth orbit satellite based on the pseudo-drag prediction value to obtain the orbital prediction result includes:

[0019] Calculate the atmospheric drag perturbation value based on the pseudo-drag prediction value;

[0020] The atmospheric drag perturbation value is substituted into the orbital dynamics model formula to predict the orbital state of the low-Earth orbit satellite and obtain the orbital prediction result.

[0021] The second aspect of this application provides a low-Earth orbit prediction device based on feature screening, comprising:

[0022] The coefficient definition unit is used to define the pseudo drag coefficient based on the atmospheric drag perturbation formula;

[0023] The impact analysis unit is used to analyze the impact of different spatial environment characteristics on the pseudo drag coefficient prediction results of the target prediction model using the SHAP method, and to generate an impact beehive diagram.

[0024] The resistance prediction unit is used to select the optimal feature combination based on the influence bee colony diagram, and predict the pseudo resistance coefficient according to the optimal feature combination and the target prediction model to obtain the pseudo resistance prediction value.

[0025] The orbit prediction unit is used to predict and calculate the orbital state of low-Earth orbit satellites based on the pseudo-drag prediction values, and obtain orbit prediction results.

[0026] Preferably, it further includes:

[0027] The feature processing unit is used to combine the components of multiple measured spatial environment features into a spliced ​​feature vector;

[0028] The model building unit is used to construct a pseudo-drag coefficient prediction model based on the spliced ​​feature vector and the preset pseudo-drag coefficient, thereby obtaining the target prediction model.

[0029] Preferably, it further includes:

[0030] The ephemeris prediction unit is used to predict the orbital state of low-Earth orbit satellites using numerical integration to obtain predicted ephemeris.

[0031] An error calculation unit is used to calculate traditional prediction errors based on the orbital dynamics model and the predicted ephemeris.

[0032] The iterative calculation unit is used to perform iterative optimization analysis based on the traditional forecast error using the golden section search algorithm to obtain the preset pseudo drag coefficient.

[0033] Preferably, the orbit prediction unit is specifically used for:

[0034] Calculate the atmospheric drag perturbation value based on the pseudo-drag prediction value;

[0035] The atmospheric drag perturbation value is substituted into the orbital dynamics model formula to predict the orbital state of the low-Earth orbit satellite and obtain the orbital prediction result.

[0036] A third aspect of this application provides a low-Earth orbit satellite orbit prediction device based on feature screening, the device including a processor and a memory;

[0037] The memory is used to store program code and transmit the program code to the processor;

[0038] The processor is used to execute the feature-based low-Earth orbit prediction method described in the first aspect according to the instructions in the program code.

[0039] The fourth aspect of this application provides a computer-readable storage medium for storing program code for executing the feature-based low-Earth orbit satellite orbit prediction method described in the first aspect.

[0040] As can be seen from the above technical solutions, the embodiments of this application have the following advantages:

[0041] This application provides a feature-based method for predicting the orbit of a low-Earth orbit (LEO) satellite, comprising: defining a pseudo-drag coefficient based on the atmospheric drag perturbation formula; using the SHAP method to analyze the influence of different space environment features on the predicted pseudo-drag coefficient of the target prediction model, and generating an influence beehive diagram; selecting the optimal feature combination based on the influence beehive diagram, and predicting the pseudo-drag coefficient based on the optimal feature combination and the target prediction model to obtain the predicted pseudo-drag value; and predicting and calculating the LEO satellite orbit state based on the predicted pseudo-drag value to obtain the orbit prediction result.

[0042] The low-Earth orbit (LEO) satellite orbit prediction method provided in this application defines the concept of a pseudo-drag coefficient, transforming the LEO satellite orbit prediction problem into a pseudo-drag coefficient prediction problem. Then, the SHAP method is used to analyze the pseudo-drag coefficient prediction process based on space environment parameters and a prediction model. This process can determine the influence of different space environment characteristics on the model prediction results. Based on this influence value, combinations of space environment characteristics that are more helpful to the prediction results can be selected. The pseudo-drag prediction value obtained based on this feature combination more closely matches the characteristics of the actual environment, ensuring the accuracy and reliability of the LEO satellite orbit prediction results. Furthermore, this process does not depend on orbit prediction errors, and therefore is not affected by the uncertainty of the error divergence direction. Therefore, this application can solve the technical problem that existing orbit predictions rely on prediction errors, but the divergence direction of the errors relative to the actual atmospheric drag is uncertain, leading to a lack of accuracy and reliability in the prediction results. Attached Figure Description

[0043] Figure 1 A flowchart illustrating a low-Earth orbit prediction method based on feature filtering, provided for an embodiment of this application;

[0044] Figure 2 A schematic diagram of a low-Earth orbit prediction device based on feature screening is provided in an embodiment of this application;

[0045] Figure 3 A schematic diagram illustrating the radial error of trajectory prediction based on the pseudo drag coefficient and the traditional numerical method, provided for embodiments of this application;

[0046] Figure 4 A schematic diagram illustrating the trajectory prediction error based on the pseudo drag coefficient and the traditional numerical method provided for embodiments of this application;

[0047] Figure 5 A schematic diagram illustrating the trajectory prediction normal error based on the pseudo drag coefficient and the traditional numerical method provided for embodiments of this application;

[0048] Figure 6 An impact beehive diagram based on Shapley values ​​is provided for embodiments of this application. Detailed Implementation

[0049] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0050] For easier understanding, please refer to Figure 1 This application provides an embodiment of a feature-based method for predicting the orbit of low-Earth orbit satellites, comprising:

[0051] Step 101: Define the pseudo drag coefficient based on the atmospheric drag perturbation formula.

[0052] The atmospheric drag perturbation formula is expressed as:

[0053]

[0054] Among them, a drag Atmospheric drag perturbation refers to the acceleration caused by atmospheric drag. For the current satellite quality, For windward area, For the ratio of flour to texture, This is the atmospheric drag coefficient. Atmospheric density, which is related to orbital altitude. Indicates v e The modulus, v e This represents the spacecraft's velocity vector relative to the atmosphere. All parameters on the right-hand side of the equation are uncertain, and most studies focus on atmospheric density. The modeling of this system has a complex dependence on space environment parameters such as the geomagnetic index and solar radiation flux index. Meanwhile, the atmospheric drag coefficient... In traditional precision orbit determination and prediction, it is estimated along with the spacecraft's state variables to absorb errors from atmospheric density models, satellite surface-to-mass ratio, and satellite attitude changes. Atmospheric drag perturbation a drag right The partial derivatives are expressed as:

[0055]

[0056] Based on the two formulas above, it can be seen that changes in space environment parameters affect atmospheric density. The calculation of this will also affect the atmospheric drag coefficient. The calculation. Furthermore, in the event of medium- to long-term orbital forecasts or significant changes in the space environment, atmospheric density... It is inherently difficult to estimate accurately, which in turn affects the atmospheric drag coefficient. The calculations lead to a decrease in the accuracy of orbit predictions and an increase in errors.

[0057] It can be observed that this embodiment uses space environment parameters to calculate atmospheric density. Then, using real orbital data, changes in the space environment are incorporated into the atmospheric drag coefficient. In this context, the spatial environmental parameters for calculating atmospheric density are set to fixed values, and the atmospheric drag coefficient is... Since orbit determination is no longer necessary and it has lost its original physical meaning, this embodiment defines it as a pseudo-drag coefficient. .

[0058] Step 102: Use the SHAP method to analyze the influence of different spatial environment characteristics on the pseudo drag coefficient prediction results of the target prediction model, and generate an influence beehive diagram.

[0059] Furthermore, step 102, preceding the following, also includes:

[0060] The components of multiple measured spatial environment features are combined into a spliced ​​feature vector;

[0061] A pseudo-drag coefficient prediction model is constructed based on the spliced ​​feature vectors and the preset pseudo-drag coefficients to obtain the target prediction model.

[0062] Furthermore, a pseudo-drag coefficient prediction model is constructed based on the concatenated feature vectors and preset pseudo-drag coefficients to obtain the target prediction model. This process also includes:

[0063] The orbital state of low-Earth orbit satellites is predicted using numerical integration, and the predicted ephemeris is obtained.

[0064] Traditional forecast errors are calculated based on orbital dynamics models and predicted ephemeris.

[0065] By using the golden section search algorithm to iteratively optimize and analyze the traditional forecast error, a preset pseudo drag coefficient is obtained.

[0066] It should be noted that the target prediction model is pre-constructed based on space environment characteristics and a pre-set pseudo-drag coefficient. This model can predict the pseudo-drag coefficient based on space environment characteristics. The pre-set pseudo-drag coefficient here is calculated using traditional methods; please refer to the detailed process. Figure 3 In this embodiment, changes in the space environment are incorporated into the atmospheric drag coefficient using real orbital data. By fixing the space environment parameters for calculating atmospheric density, a pseudo-drag coefficient can be calculated. And the pseudo-drag coefficient Solving this requires a high-precision orbital dynamics model and real satellite orbital data; this embodiment uses an orbital dynamics model in the geocentric inertial ECI coordinate system:

[0067]

[0068] in, and These represent the position and velocity of a spatial target in an inertial frame of reference, respectively. , The Euclidean norm of a vector; The gravitational constant of Earth; Represents the perturbation acceleration vector, caused by non-spherical perturbation. Three-body gravitational perturbation Atmospheric drag perturbation Solar radiation pressure perturbation Other unmodeled perturbations It consists of, etc. Apart from atmospheric drag perturbation, the other perturbation models are already accurate enough or have little impact on low-Earth orbit satellites, so they will not be discussed in detail.

[0069] The spatial environment parameters for calculating atmospheric density are set to fixed values, given an initial epoch. Initial orbital state of low-Earth orbit satellites , The orbital state at any future epoch It can be regarded as a univariate function of the pseudo-drag coefficient. The specific formula is as follows:

[0070]

[0071] The predicted ephemeris can be obtained by numerical integration calculation using the orbital dynamics model described above.

[0072] Using satellite measurement data as the actual satellite orbit data, known as precise ephemeris, and taking precise ephemeris as the benchmark, the objective function for minimizing the prediction error e(t) can be constructed as follows:

[0073]

[0074] Where, x true e(t) represents the precise ephemeris data in the inertial frame, and e(t) represents the error values ​​in the orbital coordinate system along the track, normal, and radial directions, i.e., the traditional prediction error.

[0075] pseudo drag coefficient The optimization calculation process can be viewed as finding the minimum value of a single-variable function within a certain interval. The golden section search is a method that uses the golden ratio to search within a known interval, continuously narrowing down the maximum and minimum values ​​of the function within that interval, thus finding the extreme value of the function. Therefore, this embodiment chooses to use the golden section search algorithm to iteratively optimize the calculation based on the objective function of the prediction error, obtaining the pseudo-drag coefficient when the prediction error is minimized. It is understandable that the pseudo-drag coefficient calculated using the above method can be denoted as the preset pseudo-drag coefficient. It can also be called the historical pseudo drag coefficient, because this coefficient is calculated by combining real satellite orbit data.

[0076] As can be seen from the solution process of the pseudo-drag coefficient, its determination is related to the forecast duration. This embodiment considers improving the medium- and long-term orbit forecast of low-Earth orbit satellites, and the forecast duration is selected as 10 days. This can be used as a basis to solve for the corresponding pseudo-drag coefficient. Taking a certain low-Earth orbit satellite as an example, its orbital altitude is approximately 400 km. Under the same conditions, the orbit forecast errors in the RTN coordinate system are compared between the preset pseudo-drag coefficient forecast and the traditional method forecast. For the results, please refer to [link to relevant documentation]. Figure 3 , Figure 4 and Figure 5 .

[0077] Comparing the forecast errors, it can be found that the forecast based on the preset pseudo drag coefficient in this embodiment has a significant improvement in the total position error compared with the traditional forecast. This is mainly reflected in the fact that the forecast based on the preset pseudo drag coefficient is more consistent with the traditional forecast along the track direction.

[0078] In this embodiment, the forecast error comparison is implemented based on a unified normalized scaling operation, using the pseudo-drag coefficient. The maximum forecast error is a unit quantity, and the scaled error is expressed as:

[0079]

[0080] in, This is the scaling error. The selected pseudo drag coefficient Forecast error, For other forms of forecast error, These represent radial, trace, and normal directions, respectively.

[0081] Analysis reveals that the main errors in low-Earth orbit satellite orbit prediction are along the track direction, with the track error being on a much larger scale than the radial and normal errors; the pseudo-drag coefficient... It can effectively absorb trajectory prediction errors, mainly affecting the track error. The radial error is less significant than the track error but can still be improved, while the normal error has virtually no impact. A pre-set pseudo-drag coefficient is also included. The forecast can approximate the forecast results of traditional methods, and also illustrates the pseudo drag coefficient proposed in this embodiment. The scientific validity and rationality of the concept.

[0082] Using spatial environment features as the model input vector and a pre-set pseudo drag coefficient as the output vector, a prediction model for the pseudo drag coefficient, i.e., a target prediction model, can be constructed. Since this embodiment needs to analyze the impact of different spatial environment features on the model output, the feature vector input to the model is a concatenated vector composed of components of multiple measured spatial environment features; the model can analyze the importance of different feature components and then select the features corresponding to the optimal output prediction result as the optimal feature combination.

[0083] It should be noted that this embodiment uses the SHAP method to quantify the importance of different spatial environment features to the model output, specifically expressed as influence values, i.e., Shapley values. The SHAP (SHapley Additive exPlanations) method is used to interpret the predictions of machine learning models, based on Shapley values ​​in game theory. A Shapley value is a payoff in a fair game, used to determine each player's contribution to the overall game outcome; in this embodiment, the players are the different spatial environment features.

[0084] For feature set The Shapley value for the features in the image is calculated as follows:

[0085]

[0086] in, It is the set of all features. It does not contain features. any subset of features; Represents a set The number of features in Represents the feature set Contribution to the model's predicted output Indicates the inclusion of features feature set Contribution to the model's predicted output.

[0087] This embodiment uses Shapley values ​​to quantify the impact of different space environment features on the model output. Specific space environment features include, but are not limited to, the solar radiation flux index, geomagnetic index Kp, all-day planetary magnetic index Cp, Bart solar rotation period, and international sunspot number. These specific features can be obtained and selected from the Space Environment Prediction Center (SEPC); this is only an example and not a limitation. Any feature parameters related to low-Earth orbit satellite orbit prediction can be considered to analyze their impact on the model output; subsequently, an impact swarm diagram is generated. Please refer to [link to relevant documentation]. Figure 6 In the graph, dark dots represent features that had a positive impact on the model's predictions in that observation; light dots represent features that had a negative impact. The horizontal axis represents the SHAP value, showing the magnitude of the impact; the further the point is from the central zero line, the greater the feature's impact on the model's output, and vice versa. Vertically, features are ordered by their influence from top to bottom. Features at the top have a greater overall impact on the model's output, while features at the bottom have a smaller impact. Features in the middle show dots of two different shades, but their distribution is more concentrated, indicating a relatively smaller impact. Features at the bottom have the least impact on the model, and most of their impact is close to zero, indicating that these features contribute little to the model's predictions.

[0088] Step 103: Select the optimal feature combination based on the influence bee colony diagram, and predict the pseudo resistance coefficient according to the optimal feature combination and the target prediction model to obtain the pseudo resistance prediction value.

[0089] The optimal feature combination can be selected based on the influence bee colony diagram, and based on this feature combination, a pseudo drag coefficient that is more consistent with the actual situation can be predicted, i.e., the pseudo drag prediction value. This embodiment uses raw space environment data, which does not require preprocessing. The model can analyze and select a richer and more effective optimal feature combination, ensuring the accuracy and reliability of the pseudo drag prediction value.

[0090] Step 104: Calculate the orbital state of the low-Earth orbit satellite based on the pseudo-drag prediction value to obtain the orbital prediction result.

[0091] Further, step 104 includes:

[0092] Calculate atmospheric drag perturbation values ​​based on pseudo-drag predictions;

[0093] By substituting atmospheric drag perturbation values ​​into the orbital dynamics model formula, the orbital state of low-Earth orbit satellites is predicted and calculated, resulting in orbital forecasts.

[0094] Atmospheric drag perturbation values ​​can be calculated using the atmospheric drag perturbation formula. Substituting these values ​​into the orbital dynamics model formula allows for the calculation of the satellite's specific position and velocity in its orbit, thus yielding a predicted orbital state. Since the pseudo-drag prediction values ​​are obtained through an optimized selection process, their reliability is ensured. Therefore, the orbital prediction results calculated based on these values ​​possess both accuracy and reliability.

[0095] The low-Earth orbit (LEO) satellite orbit prediction method based on feature selection provided in this application defines the concept of a pseudo-drag coefficient, transforming the LEO satellite orbit prediction problem into a pseudo-drag coefficient prediction problem. Then, the SHAP method is used to analyze the pseudo-drag coefficient prediction process based on space environment parameters and a prediction model. This process can determine the influence of different space environment features on the model prediction results. Based on this influence value, combinations of space environment features that are more helpful to the prediction results can be selected. The pseudo-drag prediction value obtained based on this feature combination more closely matches the characteristics of the actual environment, ensuring the accuracy and reliability of the LEO satellite orbit prediction results. Furthermore, this process does not rely on orbit prediction errors and is therefore unaffected by the uncertainty of the error divergence direction. Therefore, this application can solve the technical problem that existing orbit predictions rely on prediction errors, but the divergence direction of the error relative to the actual atmospheric drag is uncertain, leading to a lack of accuracy and reliability in the prediction results.

[0096] For easier understanding, please refer to Figure 2 This application provides an embodiment of a low-Earth orbit prediction device based on feature screening, comprising:

[0097] Coefficient definition unit 201 is used to define pseudo drag coefficients based on atmospheric drag perturbation formula;

[0098] The impact analysis unit 202 is used to analyze the impact of different spatial environment characteristics on the pseudo drag coefficient prediction results of the target prediction model using the SHAP method, and generate an impact beehive diagram.

[0099] The resistance prediction unit 203 is used to select the optimal feature combination based on the influence bee colony diagram, and predict the pseudo resistance coefficient according to the optimal feature combination and the target prediction model to obtain the pseudo resistance prediction value.

[0100] The orbit prediction unit 204 is used to predict and calculate the orbital state of low-Earth orbit satellites based on pseudo-drag prediction values, and obtain orbit prediction results.

[0101] Furthermore, it also includes:

[0102] The feature processing unit 205 is used to combine the components of multiple measured spatial environment features into a spliced ​​feature vector.

[0103] Model building unit 206 is used to build a pseudo drag coefficient prediction model based on spliced ​​feature vectors and preset pseudo drag coefficients to obtain the target prediction model.

[0104] Furthermore, it also includes:

[0105] Ephemeris prediction unit 207 is used to predict the orbital state of low-Earth orbit satellites using numerical integration to obtain predicted ephemeris;

[0106] Error calculation unit 208 is used to calculate traditional forecast errors based on orbital dynamics model and predicted ephemeris;

[0107] The iterative calculation unit 209 is used to perform iterative optimization analysis based on the traditional forecast error using the golden section search algorithm to obtain the preset pseudo drag coefficient.

[0108] Furthermore, the orbit prediction unit 204 is specifically used for:

[0109] Calculate atmospheric drag perturbation values ​​based on pseudo-drag predictions;

[0110] By substituting atmospheric drag perturbation values ​​into the orbital dynamics model formula, the orbital state of low-Earth orbit satellites is predicted and calculated, resulting in orbital forecasts.

[0111] This application also provides a low-Earth orbit satellite orbit prediction device based on feature screening, the device including a processor and a memory;

[0112] The memory is used to store program code and transfer the program code to the processor;

[0113] The processor is used to execute the feature-based low-Earth orbit prediction method in the above method embodiment according to the instructions in the program code.

[0114] This application also provides a computer-readable storage medium for storing program code for executing the feature-based low-Earth orbit satellite orbit prediction method in the above method embodiments.

[0115] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0116] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0117] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0118] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for executing all or part of the steps of the methods described in the various embodiments of this application through a computer device (which may be a personal computer, server, or network device, etc.). The aforementioned storage medium includes: USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media capable of storing program code.

[0119] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for predicting the orbit of low-Earth orbit satellites based on feature selection, characterized in that, include: The pseudo-drag coefficient is defined based on the atmospheric drag perturbation formula, which is expressed as follows: ； Among them, a drag Atmospheric drag perturbation refers to the acceleration caused by atmospheric drag. For the current satellite quality, For windward area, For the ratio of flour to texture, This is the atmospheric drag coefficient. Atmospheric density, which is related to orbital altitude. V represents e The modulus, V e This represents the velocity vector of a spacecraft relative to the atmosphere. The atmospheric drag perturbation a drag right The partial derivatives are expressed as: ； The atmospheric density was calculated using space environment parameters. Then, using real orbital data, changes in the space environment are incorporated into the atmospheric drag coefficient. In this context, the spatial environmental parameters for calculating the atmospheric density are set to fixed values, and the atmospheric drag coefficient... Since orbit determination is no longer necessary and it has lost its original physical meaning, it is defined as a pseudo-drag coefficient. ; The components of multiple measured spatial environment features are combined into a spliced ​​feature vector; The orbital state of low-Earth orbit satellites is predicted using numerical integration, and the predicted ephemeris is obtained. Traditional forecast errors are calculated based on the orbital dynamics model and the predicted ephemeris. The preset pseudo drag coefficient is obtained by iterative optimization analysis based on the traditional forecast error using the golden section search algorithm. Based on the spliced ​​feature vector and the preset pseudo drag coefficient, a pseudo drag coefficient prediction model is constructed to obtain the target prediction model; The SHAP method is used to analyze the influence of different spatial environment characteristics on the pseudo drag coefficient prediction results of the target prediction model, and an influence beehive diagram is generated. The optimal feature combination is selected based on the influence bee colony diagram, and the pseudo resistance coefficient is predicted according to the optimal feature combination and the target prediction model to obtain the pseudo resistance prediction value. The orbital state of the low-Earth orbit satellite is predicted and calculated based on the pseudo-drag prediction value to obtain the orbital prediction result.

2. The low-Earth orbit prediction method based on feature screening according to claim 1, characterized in that, The prediction calculation of the low-Earth orbit satellite's orbital state based on the pseudo-drag prediction value yields orbital prediction results, including: Calculate the atmospheric drag perturbation value based on the pseudo-drag prediction value; The atmospheric drag perturbation value is substituted into the orbital dynamics model formula to predict the orbital state of the low-Earth orbit satellite and obtain the orbital prediction result.

3. A low-Earth orbit satellite orbit prediction device based on feature screening, characterized in that, include: The coefficient definition unit is used to define the pseudo-drag coefficient based on the atmospheric drag perturbation formula, which is expressed as: ； Among them, a drag Atmospheric drag perturbation refers to the acceleration caused by atmospheric drag. For the current satellite quality, For windward area, For the ratio of flour to texture, This is the atmospheric drag coefficient. Atmospheric density, which is related to orbital altitude. V represents e The modulus, V e This represents the velocity vector of a spacecraft relative to the atmosphere. The atmospheric drag perturbation a drag right The partial derivatives are expressed as: ； The atmospheric density was calculated using space environment parameters. Then, using real orbital data, changes in the space environment are incorporated into the atmospheric drag coefficient. In this context, the spatial environmental parameters for calculating the atmospheric density are set to fixed values, and the atmospheric drag coefficient... Since orbit determination is no longer necessary and it has lost its original physical meaning, it is defined as a pseudo-drag coefficient. ; The feature processing unit is used to combine the components of multiple measured spatial environment features into a spliced ​​feature vector; The ephemeris prediction unit is used to predict the orbital state of low-Earth orbit satellites using numerical integration to obtain predicted ephemeris. An error calculation unit is used to calculate traditional prediction errors based on the orbital dynamics model and the predicted ephemeris. The iterative calculation unit is used to perform iterative optimization analysis based on the traditional forecast error using the golden section search algorithm to obtain the preset pseudo drag coefficient. The model building unit is used to construct a pseudo-drag coefficient prediction model based on the spliced ​​feature vector and the preset pseudo-drag coefficient to obtain the target prediction model. The impact analysis unit is used to analyze the impact of different spatial environment characteristics on the pseudo drag coefficient prediction results of the target prediction model using the SHAP method, and to generate an impact beehive diagram. The resistance prediction unit is used to select the optimal feature combination based on the influence bee colony diagram, and predict the pseudo resistance coefficient according to the optimal feature combination and the target prediction model to obtain the pseudo resistance prediction value. The orbit prediction unit is used to predict and calculate the orbital state of low-Earth orbit satellites based on the pseudo-drag prediction values, and obtain orbit prediction results.

4. The low-Earth orbit prediction device based on feature screening according to claim 3, characterized in that, The orbit prediction unit is specifically used for: Calculate the atmospheric drag perturbation value based on the pseudo-drag prediction value; The atmospheric drag perturbation value is substituted into the orbital dynamics model formula to predict the orbital state of the low-Earth orbit satellite and obtain the orbital prediction result.

5. A low-Earth orbit satellite orbit prediction device based on feature screening, characterized in that, The device includes a processor and a memory; The memory is used to store program code and transmit the program code to the processor; The processor is used to execute the low-Earth orbit prediction method based on feature screening as described in any one of claims 1-2 according to the instructions in the program code.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store program code, which, when executed by an executor, is used to implement the feature-based low-Earth orbit prediction method according to any one of claims 1-2.

Images