A Multi-Granularity and Multi-Pulse Observation and Tracking Training Method for Satellite Digital-Twin Fusion Testing

Abstract

The present invention discloses a multi-granularity multi-pulse observation and tracking training method for satellite digital-physical fusion testing, belonging to the fields of spacecraft digitization and computer science. The method includes: building a digital twin test scenario environment, constructing a digital twin satellite observation and tracking model, conducting digital twin multi-granularity multi-pulse observation and tracking training, and performing satellite digital-physical fusion testing of multi-granularity multi-pulse observation and tracking; designing a multi-granularity multi-pulse observation and tracking training method for satellite digital-physical fusion, including a digital twin tracking satellite multi-granularity multi-pulse observation and tracking maneuvering strategy and a digital twin target satellite single-granularity multi-pulse escape maneuvering strategy, to achieve satellite tracking training of multi-model granularity and multi-pulse orbit transfer in the observation task and digital-physical fusion correction of the model.

Description

Technical Field

[0001] The present invention belongs to the fields of spacecraft digitization and computer science, and particularly relates to a multi-granularity multi-pulse observation and tracking training method for satellite digital-physical fusion testing. Background Art

[0002] Space power is an important factor related to national security and prosperity. Each country is vigorously developing the construction of space situation awareness and space information reconnaissance systems. Space surveillance has become an important field of concern for each country's space activities. With the increasingly complex space environment, the number of individuals threatening space property (including debris, out-of-control satellites, space weapons, small celestial bodies, etc.) has increased sharply. The movement trajectories of non-cooperative targets are often variable and difficult to predict. Compared with two-pulse orbit transfer, the pulses of small observation satellites are dispersed in the orbit transfer path during multi-pulse orbit transfer, which can effectively reduce the fuel consumption accumulated by the increase in the number of orbit transfer wheels. In order to ensure the task of observing non-cooperative moving targets while minimizing fuel consumption, it is of great value to carry out research on multi-pulse observation and tracking training.

[0003] The current multi-pulse observation and tracking training has the following problems: ① Existing mainstream multi-pulse tracking models often adopt a single fixed model scheme, with unchanged computational complexity. The model calculation error will increase with the increase of the observation distance. During the close-range tracking process, it will be passive due to the inability of the computational complexity to keep up with the changes in the tracking situation. ② For the satellite multi-pulse observation and tracking training model, the current fixed simplified modeling method is used, which cannot meet the needs of real-time on-orbit dynamic emergency observation tasks. The accuracy of the model directly affects whether the task can be successfully completed, and there is a lack of model correction processing for the observation model. Summary of the Invention

[0004] The technical problems to be solved by the present invention are as follows: ① Existing mainstream multi-pulse tracking models often adopt a single fixed model scheme, with unchanged computational complexity. The model calculation error will increase with the increase of the observation distance. During the close-range tracking process, it will be passive due to the inability of the computational complexity to keep up with the changes in the tracking situation. ② For the satellite multi-pulse observation and tracking training model, the current fixed simplified modeling method is used, which cannot meet the needs of real-time on-orbit dynamic emergency observation tasks. The accuracy of the model directly affects whether the task can be successfully completed, and there is a lack of model correction processing for the observation model. To solve the above problems, the present invention introduces a multi-granularity model method and digital-physical fusion testing technology, designs a training method, and improves the response ability of the observation and tracking task to a certain extent. The present invention solves its technical problems by adopting the following technical solutions: A multi-granularity multi-pulse observation and tracking training method for satellite digital-physical fusion testing, and the specific method flow is as follows:

[0005] Step S101: Build the digital twin test scenario environment for satellite digital-physical integration testing, and add digital twin tracking satellite entities, digital twin target satellite entities, and digital twin property satellite entities, specifically including:

[0006] ① The digital twin tracking satellite entity, which is the training subject on our side and is responsible for approaching and observing suspicious target satellites within the scenario;

[0007] ② The digital twin target satellite entity, which is a non-cooperative subject and is responsible for continuously approaching our property satellite within the scenario and avoiding close observation by the tracking satellite during the approach;

[0008] ③ The digital twin property satellite entity, which is an important property target on our side. Referencing the space station or large GEO relay satellites, it orbits along a fixed orbit and has no maneuvering ability;

[0009] Step S102: Build the satellite digital twin observation and tracking model for satellite digital-physical integration testing, specifically including:

[0010] ① Build the digital twin orbit model and the multi-pulse orbit maneuver models for the tracking satellite and the target satellite;

[0011] ② Build the multi-granularity observation calculation model library for the digital twin tracking satellite, and establish an on-board real-time optional multi-granularity calculation observation model library for the tracking satellite, which can be used to select the most suitable calculation granularity according to the tracking situation during the training and testing of the tracking satellite;

[0012] ③ Build the multi-granularity escape calculation model library for the digital twin target satellite. It is defined that the target satellite has no real-time optional granularity ability, but a single calculation model granularity can be configured for the target satellite before each training. Select from the multi-granularity escape calculation model library and fix this model granularity during this training process;

[0013] ④ Build the tracking model for the digital twin tracking satellite and the target satellite. Define that the multi-granularity observation calculation model and the tracking model of the tracking satellite are collectively called the digital twin observation and tracking model;

[0014] Step S103: Conduct digital twin multi-granularity multi-pulse observation and tracking training for satellite digital-physical integration testing, and train the digital twin observation and tracking model, specifically including:

[0015] ① Configure the multi-granularity multi-pulse observation and tracking maneuver strategy for the digital twin tracking satellite;

[0016] ② Configure the single-granularity multi-pulse escape maneuver strategy for the digital twin target satellite;

[0017] ③ Modify the initial orbit parameters of the three, modify the maneuver constraint conditions of the tracking satellite and the target satellite, and modify the observation constraint conditions, and conduct batch digital twin observation and tracking training to determine the parameter range of the multi-granularity multi-pulse observation and tracking maneuver strategy of the tracking satellite for use in digital-physical integration testing;

[0018] Step S104: Conduct the satellite number-reality fusion test for multi-granularity multi-pulse observation and tracking. Deploy the trained multi-granularity multi-pulse digital twin observation and tracking model in step S103 to real observation satellites, correct the multi-granularity multi-pulse digital twin observation and tracking model according to the telemetry parameters of real satellite tests, and conduct multiple fusion tests to continuously correct the on-board multi-granularity multi-pulse observation and tracking model.

[0019] The advantages of the present invention compared with the prior art are as follows:

[0020] (1) Design the number-reality fusion test process for satellite multi-granularity multi-pulse observation and tracking training. It includes the construction of the digital twin test scenario environment, the construction of the digital twin satellite observation and tracking model, the digital twin multi-granularity multi-pulse observation and tracking training, and the satellite number-reality fusion test for multi-granularity multi-pulse observation and tracking. A scheme for correcting the model during the satellite observation task is given, which improves the reliability and energy efficiency of on-board observation and tracking training to a certain extent.

[0021] (2) Design the method for multi-granularity multi-pulse observation and tracking training of satellite number-reality fusion, including the multi-granularity multi-pulse observation and tracking maneuver strategy of the digital twin tracking satellite, the single-granularity multi-pulse escape maneuver strategy of the digital twin target satellite, and the batch digital twin observation and tracking training and number-reality fusion test. A tracking training process for the satellite to select the observation model granularity in real time on board is given, which improves the response efficiency and decision-making effectiveness of on-board observation tasks to a certain extent. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is the flowchart of the number-reality fusion test for multi-granularity multi-pulse observation and tracking training of the present invention;

[0023] Figure 2 It is the maneuver strategy diagram of the multi-granularity multi-pulse observation and tracking training of the digital twin tracking satellite of the present invention;

[0024] Figure 3 It is the maneuver strategy of the single-granularity multi-pulse escape training of the digital twin target satellite of the present invention;

[0025] Figure 4 It is the schematic diagram of the maneuver mode of the target satellite to avoid the tracking satellite of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0026] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0027] According to an embodiment of the present invention, a multi-granularity and multi-pulse observation and tracking training method for satellite digital-physical fusion testing of the present invention, Figure 1 is the digital-physical fusion test flow chart of the multi-granularity and multi-pulse observation and tracking training of the present invention, Figure 2 is the maneuver strategy diagram of the multi-granularity and multi-pulse observation and tracking training of the digital twin tracking satellite of the present invention, Figure 3 is the maneuver strategy of the single-granularity and multi-pulse escape training of the digital twin target satellite of the present invention, Figure 4 is the schematic diagram of the maneuver mode of the target satellite to avoid the tracking satellite, introduced with a typical three-satellite scenario as an example, such as Figure 1 shown, specifically including:

[0028] Step S101, build the digital twin test scenario environment for satellite digital-physical fusion testing, and add digital twin tracking satellite entities, digital twin target satellite entities and digital twin property satellite entities, specifically including:

[0029] ① The digital twin tracking satellite entity, which is the training subject on our side and is responsible for approaching and observing suspicious target satellites in the scenario;

[0030] ② The digital twin target satellite entity, which is a non-cooperative party subject and is responsible for continuously approaching our property satellite in the scenario and avoiding close observation by the tracking satellite during the approach;

[0031] ③ The digital twin property satellite entity, which is our property target, and runs along a fixed orbit with reference to a GEO large relay satellite and has no maneuvering ability;

[0032] Step S102, build the satellite digital twin observation and tracking model for satellite digital-physical fusion testing, specifically including:

[0033] ① Build the digital twin orbit model and the multi-pulse orbit maneuver models of the tracking satellite and the target satellite;

[0034] ② Build the multi-granularity observation calculation model library of the digital twin tracking satellite, and establish an on-board real-time optional multi-granularity calculation observation model library for the tracking satellite, so that the most suitable calculation granularity can be selected according to the tracking situation during the training and testing of the tracking satellite;

[0035] ③ Build the multi-granularity escape calculation model library of the digital twin target satellite, define that the target satellite has no real-time optional granularity ability, and configure a single calculation model granularity for the target satellite before each training, select from the multi-granularity escape calculation model library and fix this model granularity during this training process;

[0036] ④ Build the tracking models of the digital twin tracking satellite and the target satellite, and define that the multi-granularity observation calculation model and the tracking model of the tracking satellite are collectively called the digital twin observation and tracking model;

[0037] Step S103: Conduct digital twin multi-granularity multi-pulse observation and tracking training for satellite digital-physical fusion testing, and train the digital twin observation and tracking model, specifically including:

[0038] ① Configure the digital twin tracking satellite multi-granularity multi-pulse observation and tracking maneuver strategy;

[0039] ② Configure the digital twin target satellite single-granularity multi-pulse escape maneuver strategy;

[0040] ③ Modify the initial orbital parameters of the tracking satellite, target satellite, and property satellite, modify the maneuver constraints of the tracking satellite and target satellite, and modify the observation constraints, and conduct batch digital twin observation and tracking training to determine the parameter range of the tracking satellite multi-granularity multi-pulse observation and tracking maneuver strategy for digital-physical fusion testing;

[0041] Step S104: Conduct satellite digital-physical fusion testing for multi-granularity multi-pulse observation and tracking, deploy the trained multi-granularity multi-pulse digital twin observation and tracking model in step S103 to a real observation satellite, correct the multi-granularity multi-pulse digital twin observation and tracking model according to the telemetry parameters of real satellite testing, and conduct multiple fusion tests to continuously correct the on-board multi-granularity multi-pulse observation and tracking model.

[0042] The tracking model in item ④ of step S102 specifically includes:

[0043] Define the safe approach area geometrically. Denote the position of the sun in the Earth's equatorial inertial system as , which is obtained according to the ephemeris in the calculation. The position of the target satellite in the Earth's equatorial inertial system is , then the position vector from the target satellite to the sun at time t: ;

[0044] The tracking satellite optimization strategy flies into the safe approach area of the target satellite, and the target satellite also continuously optimizes the strategy to prevent the tracking satellite from flying into its own safe approach area, and constructs a two-star zero-sum model: , where x p , x e represents the states of the pursuer and the evader, x p represents the tracking satellite, x e represents the target satellite, u is a pair of strategies, and the initial state constitutes the initial boundary conditions of the game: ; The terminal time constraint , in order for the tracking satellite to enter the safe approach area of the target satellite at the terminal time, the terminal position constraint: , that is, the distance from the tracking satellite to the target satellite at the terminal time is less than the radius of the safe approach area , and the angle between the tracking satellite and the sun with the target satellite as the vertex at the terminal time is less than the angle of the safe approach area ; The target star maneuvers its orbit so that the tracking star is not within its own safety zone at the terminal time ; Meanwhile, the total fuel constraint is expressed by the velocity increment , is the orbit-changing pulse vector of the tracking star, is the orbit-changing pulse vector of the target star, and both need to be less than the pulse upper limit of the fuel constraint;

[0045] Define as the policy benefit. If the following inequalities are all satisfied for u p and all u e , are said to form a saddle-point equilibrium, or saddle-point strategy:

[0046] .

[0047] The maneuvering strategy for the multi-granularity and multi-pulse observation tracking training of the digital twin tracking star in step S103 is as Figure 2 shown, and the specific implementation includes:

[0048] Step S301: Set the minimum distance identifier of the target star to 1, and select an appropriate orbit determination granularity model. First, perform model consistency verification. If the verification fails, correct the model. After passing, determine the orbit of the target star and execute step S302; Set the minimum distance identifier of the target star to 1, and select an appropriate orbit determination granularity model. First, perform model consistency verification. If the verification fails, correct the model. After passing, determine the orbit of the target star and execute step S302;

[0049] Step S302: Solve the minimum distance between the target star and the property star according to the orbit determination result in step S301 , and the time

[0050] to reach the minimum distance, and execute step S303; Whether the minimum distance between the target star and the property star is less than the threat distance of the property star

[0051] ① If the minimum distance between the target star and the property star is less than the threat distance of the property star, the tracking star selects an appropriate orbit-changing granularity model. First, perform model consistency verification. If the verification fails, correct the model. After passing, calculate the orbit-changing pulse to approach the target star, and store the orbit determination parameters, the time to reach the minimum distance, the pulse maneuver time, and the pulse maneuver components, and execute step S304;

[0052] ② If the minimum distance between the target star and the property star is greater than the threat distance of the property star, return to step S301 to re-determine the orbit of the target star;

[0053] Step S304: Determine whether the maneuver time has arrived according to the stored impulse maneuver time.

[0054] ① If the recorded time is reached, perform the impulse maneuver at the corresponding time.

[0055] ② If the impulse maneuver time has not arrived, set the identifier to 0, and then execute Step S305.

[0056] Step S305: Continue to select an appropriate granularity to determine the orbit of the target star, and recursively push the current orbit determination result and the previously stored orbit determination result to the stored time of reaching the minimum distance , and compare the recursive results of the two orbit determinations:

[0057] ① If the difference is less than the set maneuver judgment threshold, it is considered that the target star has not made an evasive maneuver, and return to Step S305 to re-determine the orbit of the target star.

[0058] ② If the difference is greater than the set maneuver judgment threshold, it is considered that the target star has made an evasive maneuver, and execute Step S306.

[0059] Step S306: Further determine whether the recursive result of the current orbit determination is less than the threat distance of the property star :

[0060] ① If the recursive result of the current orbit determination is less than the threat distance of the property star, it is considered that the target star has made a maneuver and has not changed its threat intention. Clear the stored minimum distance, minimum distance time, maneuver time, and maneuver impulse, set the identifier to 1, return to Step S301 to re-determine the orbit of the target star, and re-calculate the minimum distance and the time of reaching the minimum distance.

[0061] ② If the recursive result of the current orbit determination is greater than the threat distance of the property star , it is considered that the target star has made a maneuver, but the end position does not threaten the property star. There are two possibilities. One is that the target star has abandoned the task of threatening the property star, and the other is that the target star is performing the first few pulses of a multi-pulse orbit change and is on a transfer orbit where the threat cannot be judged. In either case, the target star had the intention of threatening the property star before. Therefore, the tracking star adopts the strategy of approaching the property star for both possibilities, and at the same time selects an appropriate orbit maneuver granularity to solve and store the maneuver time and the impulse maneuver component, and execute Step S304.

[0062] The maneuver strategy of the digital twin target star single-granularity multi-pulse escape training in ② of the above Step S103 is as Figure 3 shown, and the specific implementation includes:

[0063] Step S401: First, perform model consistency verification. If the verification fails, correct the model. If the verification passes, determine the orbit of the target star as the property star and solve for the multi-pulse orbit transfer to approach the property star.

[0064] Step S402: Determine the orbit of the target star as the tracking star and execute Step S403.

[0065] Step S403: Solve for the minimum distance between the tracking star and the target star based on the orbit determination result in Step S402 , and the time to reach the minimum distance , and execute Step S404.

[0066] Step S404: Determine whether the minimum distance between the tracking star and the target star is less than the threat distance of the target star , and make a judgment:

[0067] ① If the minimum distance between the tracking star and the target star is less than the threat distance of the target star , the target star calculates the orbit transfer pulse to avoid the tracking star using a fixed orbit transfer granularity, and stores the orbit determination parameters, the time to reach the minimum distance, the pulse maneuver time, and the pulse maneuver components, and execute Step S405;

[0068] ② If the minimum distance between the tracking star and the target star is greater than the threat distance of the target star , return to Step S401 to re-determine the orbit of the tracking star;

[0069] Step S405: Determine whether the maneuver time has been reached based on the stored pulse maneuver time:

[0070] ① If the recorded time has been reached, perform the pulse maneuver at the corresponding time;

[0071] ② If the pulse maneuver time has not been reached, execute Step S402.

[0072] In the above Step S404, the maneuver method for the target star to avoid the tracking star is as Figure 4 shown, and the specific implementation includes: The tracking star drifts past the target star at the minimum distance , and divides the circle with the target star as the center and the threat distance of the target star as the radius into a semi-circle containing the minimum distance and a semi-circle region not containing the minimum distance. The target star moves one threat distance of the target star towards the semi-circle region not containing the minimum distance, and the target star must ensure that it is within the threat distance of the property star , and makes a small-distance pulse maneuver under this scheme.

[0073] In summary, the present invention discloses a multi-granularity multi-pulse observation and tracking training method for satellite digital-physical fusion testing, including designing a digital-physical fusion test process for observation and tracking training and designing a multi-granularity multi-pulse observation and tracking training method, which can, to a certain extent, solve the problems existing in current satellite observation and tracking training and improve the response efficiency and decision-making effectiveness in on-board observation tasks.

[0074] The content not detailedly described in the specification of the present invention belongs to the prior art well-known to those skilled in the art.

[0075] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art in the technical field, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A multi-granularity multi-pulse observation tracking training method for satellite digital-real fusion testing, characterized in that: The specific process of the method is as follows: Step S101: construct a digital twin test scenario environment for satellite digital-real fusion testing, add a digital twin tracking star entity, a digital twin target star entity, and a digital twin property star entity, specifically including: ① The digital twin tracking star entity is our training subject, responsible for approaching and observing suspicious target stars in the scene; ② The digital twin target star entity is a non-cooperative entity that is responsible for continuously approaching our property star in the scene and avoiding close observation by the tracking star on the way; ③ The digital twin property star entity is our property target, which is based on the space station or GEO large relay satellite and runs along a fixed orbit without maneuverability; Step S102, constructing a satellite digital twin observation and tracking model for satellite digital-real fusion testing, specifically includes: ① Construction of digital twin orbit model, multi-pulse orbit maneuver model of tracking satellite and target satellite; ② Construction of a multi-granularity observation computing model library for digital twin tracking stars. Establish a real-time optional multi-granularity observation computing model library on the tracking star, so that the tracking star can select the most appropriate computing granularity according to the tracking situation during training and testing; ③ Construction of a multi-granularity escape computing model library for the digital twin target satellite. Define that the target satellite has no real-time selectable granularity capability. Configure a single computing model granularity for the target satellite before each training. Select from the multi-granularity escape computing model library and fix this model granularity during this training process. ④Construct the tracking model of the digital twin tracking star and the target star, and define the tracking star multi-granularity observation calculation model and tracking model as the digital twin observation tracking model; Step S103: Performing digital twin multi-granularity multi-pulse observation and tracking training for satellite digital-real fusion testing, and training the digital twin observation and tracking model, specifically including: ①Configure the digital twin tracking satellite multi-granularity and multi-pulse observation and tracking maneuvering strategy; ② Configure the digital twin target satellite single-granularity multi-pulse escape maneuver strategy; ③ Modify the initial orbital parameters of the tracking satellite, target satellite and property satellite, modify the maneuvering constraints of the tracking satellite and target satellite, and modify the observation constraints, and conduct batch digital twin observation and tracking training to determine the parameter range of the multi-granularity and multi-pulse observation and tracking maneuvering strategy of the tracking satellite for use in digital-physical fusion testing; Step S104: Perform satellite digital-real fusion test of multi-granularity multi-pulse observation and tracking, deploy the multi-granularity multi-pulse digital twin observation and tracking model trained in step S103 to the real observation satellite, correct the multi-granularity multi-pulse digital twin observation and tracking model according to the telemetry parameters of the real satellite test, and perform multiple fusion tests to continuously correct the on-board multi-granularity multi-pulse observation and tracking model.

2. The multi-granularity multi-pulse observation tracking training method for satellite digital-real fusion testing according to claim 1 is characterized in that: The tracking model in step S102 specifically includes: Define the safe approach area geometrically, and note that the position of the sun in the Earth's equatorial inertial system is , in the calculation, according to the ephemeris, the position of the target star in the Earth's equatorial inertial system is , then the position vector pointing from the target star to the sun at time t is: ; The tracking star optimizes its strategy to fly into the safe approach zone of the target star, and the target star also continuously optimizes its strategy to avoid the tracking star from flying into its own safe approach zone, thus constructing a two-star zero-sum model: , where x p , x e Indicates the status of the two people chasing and fleeing, x p Indicates tracking star, x e represents the target star, u is a pair of strategies, and the initial state constitutes the initial boundary condition of the game: ;Terminal time constraints In order for the tracking star to enter the safe approach area of ​​the target star at the terminal time, the terminal position constraint is: , that is, the terminal always tracks the star to the target star The distance is less than the safe approach area radius At the terminal moment, the angle between the target star and the sun is tracked with the target star as the vertex Smaller than the safe approach area angle ; The target satellite maneuvers in orbit so that the terminal always tracks the satellite outside its own safety zone ; At the same time, the total fuel constraint is expressed by the speed increment , To track the pulse vector of the star's orbit change, The pulse vector for the target star’s orbit change, both of which must be smaller than the upper pulse limit of the fuel constraint; definition is the strategy return, if the following inequalities all satisfy u p and all u e , This is called a saddle point equilibrium, or a saddle point strategy: 。 3. The multi-granularity multi-pulse observation tracking training method for satellite digital-real fusion testing according to claim 1 is characterized in that: The maneuvering strategy of the multi-granularity multi-pulse observation and tracking training of the digital twin tracking star in step S103 specifically includes: Step S301, select a suitable orbit determination granularity model, first perform model consistency verification, if the model fails the verification, perform model correction, if it passes the verification, orbit determination for the target satellite is performed, and execute step S302; Step S302: Calculate the minimum distance between the target star and the property star based on the orbit determination result in step S301 , and the time to reach the minimum distance , execute step S303; Step S303: Determine the minimum distance between the target star and the property star Is it smaller than the threat distance of the property star? ,judge: ① If the minimum distance between the target satellite and the property satellite is less than the threat distance of the property satellite, the tracking satellite selects an appropriate orbit change granularity model, first performs model consistency verification, and corrects the model if it fails the verification. After passing the verification, the orbit change pulse is calculated to approach the target satellite, and the orbit determination parameters, the time of reaching the minimum distance, the pulse maneuver time and the pulse maneuver component are stored, and step S304 is executed; ② If the minimum distance between the target star and the property star is greater than the threat distance of the property star, return to step S301 to re-orbit the target star; Step S304: Determine whether the maneuvering time has been reached according to the stored pulse maneuvering time: ① If the recorded time is reached, the pulse maneuver at the corresponding time is executed; ② If the pulse maneuvering time has not been reached, execute step S305; Step S305: Continue to select the appropriate granularity to determine the orbit of the target star and The last orbit determination result stored All are recursively extended to the stored minimum distance arrival time , and compare the recursive results of the two orbit determinations: ① If the difference is less than the set maneuver judgment threshold, it is considered that the target satellite has not made an evasive maneuver, and the process returns to step S305 to re-determine the orbit of the target satellite; ② If the difference is greater than the set maneuver judgment threshold, it is considered that the target satellite has made an evasive maneuver, and step S306 is executed; Step S306: further determine whether the recursive result of this orbit determination is less than the threat distance of the property star : ① If the recursive result of this orbit determination is less than the threat distance of the property satellite, it is considered that the target satellite has made a maneuver and has not changed its threat intention. The stored minimum distance, minimum distance time, maneuver time and maneuver pulse are cleared, and the process returns to step S301. The orbit of the target satellite is re-determined, and the minimum distance and the time of reaching the minimum distance are recalculated. ② If the recursive result of this orbit determination is greater than the threat distance of the property star , it is considered that the target star has made a maneuver, but the terminal position does not threaten the property star. There are two situations. One is that the target star has given up the task of threatening the property star, and the other is that the target star is executing the first few pulses of the multi-pulse orbit change and is in the transfer orbit and no threat can be determined. Both situations indicate that the target star has the intention of threatening the property star before. Therefore, the tracking star adopts the strategy of approaching the property star for both situations, and at the same time selects the appropriate orbital maneuver granularity to solve and store the maneuver time and pulse maneuver component, and executes step S304.

4. The multi-granularity multi-pulse observation tracking training method for satellite digital-real fusion testing according to claim 1 is characterized in that: The maneuvering strategy of the single-granularity multi-pulse escape training of the digital twin target satellite in step S103 specifically includes: Step S401, firstly, the model consistency verification is performed, and if the verification fails, the model is corrected, and after the verification passes, the target star is determined as the property star, and the multi-pulse orbit change is solved to approach the property star; Step S402: The target star is a tracking star, and the orbit is determined. Step S403 is executed; Step S403: Calculate the minimum distance between the tracking star and the target star based on the orbit determination result in step S402 , and the time to reach the minimum distance , execute step S404; Step S404: Determine the minimum distance between the tracking star and the target star Is it smaller than the threat distance of the target star? ,judge: ① If the minimum distance between the tracking star and the target star Less than the target star's threat distance The target star uses a fixed orbit change granularity to calculate the orbit change pulse to avoid the tracking star, and stores the orbit determination parameters, the time of reaching the minimum distance, the pulse maneuvering time and the pulse maneuvering component, and executes step S405; ② If the minimum distance between the tracking star and the target star Greater than the target star's threat distance , return to step S401 to re-orbit the tracking star; Step S405: Determine whether the maneuvering time has been reached according to the stored pulse maneuvering time: ① If the recorded time is reached, the pulse maneuver at the corresponding time is executed; ② If the pulse maneuvering time has not been reached, execute step S402.

5. The multi-granularity multi-pulse observation tracking training method for satellite digital-real fusion testing according to claim 4 is characterized in that: The maneuvering method of the target star to avoid the tracking star in step S404 specifically includes: When flying over the target star, the target star will be the center of the circle. The circle with a radius of is divided into a semicircle containing the minimum distance and a semicircle area not containing the minimum distance. The target star moves a target star threat distance toward the semicircle area not containing the minimum distance. , and the target star must be within the threat distance of the property star Within the range of , short-distance pulse maneuvers are performed under this scheme.

Images