A prediction pursuit game control method and system for approaching non-cooperative giant stars in space

Abstract

The present invention provides a prediction pursuit game control method and system for approaching non-cooperative giant stars in space, belonging to the field of spacecraft guidance and control. In order to solve the problem that the trajectory of existing spacecraft tracking games is unstable and not smooth, especially for a large number of targets and complex space environments, the applicability of spacecraft attitude control is poor. After establishing the dynamic model of the tracker and the target, the present invention uses the RRT* algorithm and the Bezier curve method to perform initial path planning for the tracker, and then optimizes the trajectory tracking through the PEG game strategy under the MPC framework to achieve the pursuit game approximation control of non-cooperative giant star targets in space. When targeting autonomous and dynamically changing targets, the present invention can achieve accurate tracking of targets in complex dynamic environments. It has been verified by simulation experiments that the desired trajectory can be tracked, proving the high accuracy and stability in the pursuit game trajectory tracking.

Description

Technical Field

[0001] The present invention relates to the field of spacecraft guidance and control technology, and in particular to a prediction pursuit and escape game control method and system applied to approaching a non-cooperative giant star in space. Background Art

[0002] Low-orbit giant satellites have become an important trend in the development of space technology due to their low launch cost, wide global coverage, strong anti-destruction, low latency and low power consumption. Low-orbit giant satellites such as Starlink achieve global link interconnection through dense deployment, play a key role in global reconnaissance and communications, and at the same time bring potential risks to my country's network, data and military security.

[0003] Non-cooperative giant stars usually have a certain degree of autonomous maneuvering and obstacle avoidance capabilities, which poses a challenge to the research of approaching targets and conducting non-cooperative pursuit technology. In the traditional pursuit process, single-objective optimization techniques suitable for approximate linearity, such as PID (Proportional-Integral-Derivative) control, are usually used. However, non-cooperative giant star targets involve dynamic problems such as nonlinear maneuvering obstacle avoidance and approach time uncertainty in complex space environments, which makes traditional methods lack autonomy and flexibility when facing system changes.

[0004] In recent years, the Pursuit-Evasion Game (PEG) technology has developed rapidly in the field of space non-cooperative target approach. PEG technology is based on methods such as differential game, reinforcement learning and game tree search, and has shown excellent performance in strategy optimization, dynamic decision-making and complex environment response. The PEG intelligent control method combined with model predictive control (MPC) can not only effectively solve the target optimization problem under uncertain pursuit strategy, but also overcome the limitations of traditional control methods in tracking the dynamics of non-cooperative giant stars.

[0005] There are many non-cooperative giant star constellations and they cover a wide range of orbits, so trajectory planning research is needed to approach the target. Sampling-based path planning algorithms, such as Probabilistic Roadmap (PRM) and Rapidly-exploring Random Tree (RRT), are widely used for path planning in unstructured environments. However, the trajectories generated by sampling-based methods are often unstable and not smooth enough, and have poor applicability to spacecraft attitude control. Faced with problems such as a large number of targets and a complex space environment, it is necessary to further optimize the trajectory planning method of the spacecraft tracking game to improve its application effect in complex environments. Summary of the invention

[0006] The technical problems to be solved by the present invention are:

[0007] In order to solve the problem that the trajectory planning process of the existing spacecraft tracking game is unstable and not smooth, especially for the situation with a large number of targets and complex space environment, the applicability of spacecraft attitude control is poor.

[0008] The present invention adopts the following technical solutions to solve the above technical problems:

[0009] The present invention provides a prediction pursuit game control method applied to approaching a non-cooperative giant star constellation in space, comprising the following steps:

[0010] S100, establish dynamic models and constraints according to the states of the target and the tracker,

[0011] It includes abstracting the environment in which the tracker and the target are located during actual flight into a specific three-dimensional mathematical space model through discrete points, defining the body coordinate system and the geocentric inertial system, and describing the conversion relationship between the tracker and the target position through the attitude transformation matrix; establishing the connection between the tracker and the target according to Newton's second law; mapping the orbital motion of the tracker and the orbital motion of the target to this system, assuming that the target is in a near-circular orbit and the distance between the target and the tracker is very small compared to the orbital radius of the target, and then deriving the linear model of relative translational dynamics;

[0012] S200, using the RRT* algorithm and the Bezier curve to perform path planning, including using the RRT* algorithm to generate a path, and then using the Bezier curve to optimize and smooth the path;

[0013] S300. Using the PEG game strategy to optimize trajectory tracking under the MPC framework includes building a pursuit-escape game model between the target and the tracker by combining the dynamic model, input constraints and objective functions. In the pursuit-escape game model, the target and the tracker act independently to meet their respective objective functions. Then, a predictive pursuit-escape game controller is built. Under the MPC framework, the multi-objective constraint optimization problem is transformed into a quadratic programming problem, and the saddle point of the game is solved to process the input constraints to achieve path tracking.

[0014] Furthermore, in step S100, establishing a dynamic model according to the state of the target and the tracker includes:

[0015] The environment in which the tracker and the target are located during actual flight is abstracted into a specific three-dimensional mathematical space model through discrete points. Using the definition of the northeast sky, a certain point in the mission space is located in the geocentric inertial system O I X I Y I Z I The following is expressed as (x, y, z), defining the body coordinate system O b X b Y b Z b With the Earth-centered inertial system O I X I Y I Z I Describe the translation of the target and tracker positions and the rotation of the attitude respectively. The attitude transformation matrix is ​​used to describe the conversion relationship between the two coordinate systems.

[0016] According to Newton's second law, the mass m of each body, the moment of inertia I in the body coordinate system, the external force F, the external torque M, the acceleration a, the angular acceleration α, and the angular velocity ω of the body are related:

[0017] F=ma

[0018] M=Iα+ω×Iω

[0019] Describe the orbital motion of the tracker and target and for:

[0020]

[0021] Among them, u c and u t are the input control accelerations of the tracker and the target, respectively; μ is the gravitational constant of the central celestial body; r c and r t are the position vectors of the tracker and the target in the geocentric inertial system at the current moment respectively;

[0022] Then, the motion vector between the tracker and the target is r = r c -r t , mapped to the body coordinate system:

[0023]

[0024] Assuming that the target is in a near-circular orbit and the distance between the target and the tracker is much smaller than the orbital radius of the target, the linear model of the target's relative translational dynamics is derived:

[0025]

[0026] in, B t =-B c ,

[0027] is the state vector, Indicates the tracker's orbital velocity.

[0028] Furthermore, in step S100, the following assumptions and constraints are made on the tracker and the target:

[0029] (1) Assume that the geometric centers of the tracker and target are connected to their respective centers of gravity;

[0030] (2) Assume that the trajectories of the tracker and the target are smooth enough for tracking;

[0031] (3) Assume that the tracker and the target are symmetrical rigid bodies;

[0032] (4) Assume that the speed is low and the starting effect is not considered;

[0033] (5) Constraint trackers must follow the desired trajectory;

[0034] (6) The non-cooperative target is constrained to have unknown maneuvers, and the tracker has better maneuverability than the target;

[0035] (7) Set input constraints for the target and tracker to ensure that the control input is within a reasonable range.

[0036] Furthermore, in step S200, it includes:

[0037] S210, initialize parameters and tree structure, define distance calculation function d(x,y), and set the maximum number of iterations N max , step size Δs, counter k and threshold ε; starting point x start As the root node, initialize the path planning tree structure T:

[0038] d(x,y)=(x1-y1) 2+(x2-y2) 2 +(x3-y3) 2

[0039] Among them, the parameters are set to: Maximum number of iterations N max =1000; step size Δs=0.1; counter k=0; threshold ε=0.05;

[0040] S220, iteratively generating a path, generating and optimizing the path through a cyclic iteration process, specifically including:

[0041] S221, random sampling, randomly generate sampling points x in the planning space rand =(x rand1 ,x rand2 ,x rand3 );

[0042] S222, find the nearest node, find the node x closest to the random point in the current tree nearest =argmin x∈T d(x,x rand );

[0043] S223, generate extension points, according to the step size Δs, from the nearest node x nearest To randomly sample point x rand Generate a new extension point x in the direction new ,

[0044]

[0045] S224, collision detection, collision detection, ensure the extension point x new Do not collide with any obstacles;

[0046] S230, insert a new node. If the extension point does not collide, insert it into the tree structure T and record its parent node information; if the extension point is x away from the target point goal If it is less than the set threshold ε, the path is considered to be found;

[0047] S240, searching for a path, starting from the target point and tracing back along the parent node pointer through the backtracking tree structure, to generate a path from the starting point to the target point;

[0048] S250, path optimization and smoothing, after the path is generated in step S240, the path is optimized and smoothed by using a Bezier curve, specifically including:

[0049] S251, select control points, select several control points according to the path points;

[0050] S252. Construct a Bezier curve. Use the Bezier curve formula to construct a smooth path through control points. The Bezier curve formula is:

[0051]

[0052] Among them, P0, P1, ..., P n is the control point, and P0=x start ,P n =x goal ;

[0053] According to the actual path requirements, the control points and the step size Δt of the Bezier curve are dynamically adjusted to optimize the smoothness and feasibility of the path and generate a smooth path suitable for actual flight.

[0054] Furthermore, in step S300, constructing the pursuit-escape game model includes:

[0055] S311, establish a dynamic model of the target and the tracker. According to the dynamic model of the target and the tracker established in step S100, use differential equations to describe their motion state and control input:

[0056]

[0057] Among them, x T and x P Represent the state vectors of the target and tracker respectively; u T and u P represent the control input of the target and the tracker respectively; f T and f P denote the dynamics functions of the target and the tracker, respectively;

[0058] S312: Set input constraints for the target and the tracker to ensure that the control input is within a reasonable range; the input constraints are physical constraints of the target and the tracker:

[0059] u T,min ≤u T ≤u T,max

[0060] u P,min ≤u P ≤u P,max

[0061] Among them, u T,min and u T,max Respectively represent the minimum and maximum values ​​of the target control input; u P,min and u P,max They represent the minimum and maximum values ​​of the tracker control input respectively;

[0062] S313. Define the objective functions of the target and the tracker to describe the control effects that the target and the tracker want to achieve, respectively. The objective function of the tracker is used to minimize the distance to the target, and the objective function of the target is used to maximize the distance to the tracker:

[0063]

[0064] Among them, J T and J P Represent the objective functions of the target and tracker respectively; L T and L P denote the instantaneous cost functions of the target and tracker respectively; Φ T and Φ P denote the terminal cost functions of the target and tracker, respectively.

[0065] Further, in step S300, constructing a prediction pursuit and escape game controller includes:

[0066] S321, prediction model, using the MPC framework to predict the optimal strategy of the target and adjust the optimal strategy of the tracker; MPC uses the dynamic model and constraints of the target and tracker to predict the future N at each control moment p System behavior over time and optimize control inputs:

[0067]

[0068] Among them, N p is the prediction time domain; N c To control the time domain; R x is the control weight matrix; x P (k|t) and x T (k|t) represents the states of the tracker and target predicted at time t at future time k; Δu P (k|t) represents the increment of control input;

[0069] S322, game solving, by repeatedly predicting the opponent's actions and optimizing one's own strategy, solving the saddle point of the game, including:

[0070] (1) Initialize the initial strategy of the tracker and target

[0071] (2) Iterative optimization, where the tracker and target update their strategies respectively, predict the opponent's response and adjust their own strategies;

[0072]

[0073] (3) Until the strategies converge and the equilibrium point of the game is reached:

[0074]

[0075] Among them, ε max is the convergence threshold.

[0076] The present invention provides a prediction pursuit and escape game control system applied to approaching non-cooperative giant star constellations in space. The system has a program module corresponding to the above steps, and executes the steps in the above prediction pursuit and escape game control method applied to approaching non-cooperative giant star constellations in space during operation.

[0077] The present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, wherein the computer program is configured to implement the steps of a predictive pursuit and escape game control method applied to approaching a non-cooperative superstar constellation in space when called by a processor.

[0078] Compared with the prior art, the present invention has the following beneficial effects:

[0079] (1) Efficient path planning and dynamic adaptability

[0080] In a complex spatial environment, the effectiveness of path planning is crucial. The present invention adopts the RRT* algorithm for path planning, and can quickly generate the optimal path from the initial position to the target position through random sampling and path optimization. The path is further smoothed by combining the Bezier curve method to ensure the adaptability and smoothness of the path in a dynamic environment. The present method shows extremely high efficiency and reliability when dealing with dynamic obstacles and complex environmental changes.

[0081] (2) Intelligent pursuit and escape game control

[0082] Traditional control methods have limited effects when facing autonomous and dynamically changing targets. The present invention uses a PEG (prediction pursuit game) control model to simulate the game process between the tracker and the target, optimize the strategies of both parties, and enable the tracker to adjust the path and strategy in real time according to the target with a constantly changing trajectory, ensuring effective approach to the target in a complex and changeable spatial environment. The PEG control model is repeatedly predicted and optimized to achieve the optimal control effect in a dynamic game environment.

[0083] (3) Accurate model predictive control

[0084] In order to achieve accurate tracking of the target, the present invention combines MPC (model predictive control), uses the dynamic model and constraints of the tracker and the target, predicts the tracker and target behaviors in the next several steps at each control moment, and optimizes the control input; MPC can handle multi-objective constraint optimization problems, ensure that the tracker achieves optimal control in a complex environment, and effectively avoids obstacles; through this method, the tracker can achieve high-precision path tracking and target capture in a dynamic environment;

[0085] (4) Test performance

[0086] In the simulation test, the pursuit-escape game control algorithm under the MPC framework involved in the present invention achieved the tracking of the desired trajectory, and the actual trajectory overlapped and was similar to the desired trajectory to a great extent, demonstrating the high accuracy and stability of the present invention in tracking the trajectory of the pursuit-escape game. BRIEF DESCRIPTION OF THE DRAWINGS

[0087] Figure 1 It is a flow chart of a prediction pursuit and escape game control method applied to approaching a non-cooperative giant star constellation in space in an embodiment of the present invention;

[0088] Figure 2 A schematic diagram of establishing a geocentric inertial coordinate system and a body coordinate system in an embodiment of the present invention;

[0089] Figure 3 This is a flow chart of the MPC principle in an embodiment of the present invention;

[0090] Figure 4 The initial path planning diagram obtained by using the RRT* algorithm and the Bezier curve in the embodiment of the present invention, wherein the green line is the original trajectory generated, and the red line is the trajectory obtained after smoothing and optimization;

[0091] Figure 5 It is a position tracking diagram of the actual trajectory on the x-axis, y-axis and z-axis in an embodiment of the present invention. DETAILED DESCRIPTION

[0092] In the description of the present invention, it should be noted that the terms "first", "second", and "third" mentioned in the embodiments of the present invention are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first", "second", and "third" may explicitly or implicitly include one or more of the features.

[0093] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0094] Specific implementation plan 1: Combine Figures 1 to 5 As shown, the present invention provides a prediction pursuit game control method applied to approaching a non-cooperative giant star constellation in space, comprising the following steps:

[0095] S100, establishing dynamic models and constraints according to the states of the target and the tracker, including:

[0096] S110, combined Figure 2As shown in the figure, the environment in which the tracker and the target are located in the actual flight is abstracted into a specific three-dimensional mathematical space model through discrete points. Using the definition of the northeast sky, a certain point in the mission space is located in the geocentric inertial system O I X I Y I Z I The following is expressed as (x, y, z), defining the body coordinate system O b X b Y b Z b The position translation and attitude rotation are described with the geocentric inertial system respectively. The attitude transformation matrix R is used to describe the conversion relationship between the two coordinate systems.

[0097] According to Newton's second law, the mass m of each body, the moment of inertia I in the body coordinate system, the external force F, the external torque M, and the body's acceleration a, angular acceleration α, and angular velocity ω are related:

[0098] F=ma

[0099] M=Iα+ω×Iω

[0100] Describe the orbital motion of the tracker and target and for:

[0101]

[0102] Among them, u c and u t are the input control accelerations of the tracker and the target, respectively; μ is the gravitational constant of the central celestial body; r c and r t are the position vectors of the tracker and the target in the geocentric inertial system at the current moment respectively;

[0103] Then, the motion vector between the tracker and the target is r = r c -r t , mapped to the body coordinate system:

[0104]

[0105] in, is the first-order derivative of the angular velocity, is the first-order derivative of the motion vector between the tracker and the target;

[0106] Assuming that the target is in a near-circular orbit and the distance between the target and the tracker is much smaller than the orbital radius of the target, the linear model of the target's relative translational dynamics is derived:

[0107]

[0108] in, B t =-B c , is the state vector, represents the tracker's orbital velocity;

[0109] S120, the following assumptions and constraints are made for the tracker and the target:

[0110] (1) Assume that the geometric centers of the tracker and target are connected to their respective centers of gravity;

[0111] (2) Assume that the trajectories of the tracker and the target are smooth enough for tracking;

[0112] (3) Assume that the tracker and the target are symmetrical rigid bodies;

[0113] (4) Assume that the speed is low and the starting effect is not considered;

[0114] (5) Constraint trackers must follow the desired trajectory;

[0115] (6) The non-cooperative target is constrained to have unknown maneuvers, and the tracker has better maneuverability than the target;

[0116] (7) Set input constraints for the target and tracker to ensure that the control input is within a reasonable range;

[0117] S200, using RRT* algorithm and Bezier curve for path planning, including,

[0118] S210, initialization parameters and tree structure,

[0119] By defining the distance calculation function d(x,y) and setting the maximum number of iterations N max , step size Δs, counter k and threshold ε; then, starting from the starting point x start As the root node, initialize the path planning tree structure T:

[0120] d(x,y)=(x1-y1) 2 +(x2-y2) 2 +(x3-y3) 2

[0121] Among them, the parameters are set to: Maximum number of iterations N max =1000; step size Δs=0.1; counter k=0; threshold ε=0.05;

[0122] S220, iteratively generate a path,

[0123] The path is generated and optimized through a cyclic iterative process, including:

[0124] S221, random sampling: randomly generate sampling points x in the planning space rand =(x rand1 ,x rand2 ,x rand3 );

[0125] S222, find the nearest node: find the node x closest to the random point in the current tree nearest =argmin x∈T d(x,x rand );

[0126] S223, generate extension points: according to the step size Δs, from the nearest node x nearest To randomly sample point x rand Generate a new extension point x in the direction new ,

[0127]

[0128] S224, Collision detection: Perform collision detection to ensure that the extension point x new Do not collide with any obstacles;

[0129] S230, insert new node

[0130] If the extension point does not collide, insert it into the tree structure T and record its parent node information; if the extension point is x away from the target point goal If it is less than the set threshold ε, the path is considered to be found;

[0131] S240, Finding a Path

[0132] Through the backtracking tree structure, starting from the target point and backtracking along the parent node pointer, a path from the starting point to the target point is generated;

[0133] S250, path optimization and smoothing

[0134] After the path is generated, the path is optimized and smoothed by the Bezier curve to improve the operability and smoothness of the path. The Bezier curve is a curve suitable for path smoothing. Through the dynamic adjustment of control points, the path smoothing and obstacle avoidance can be achieved.

[0135] The specific steps are as follows:

[0136] S251, select control points: select a number of control points according to the path points, and these points determine the shape of the Bezier curve;

[0137] S252. Construct Bezier curve: Use the Bezier curve formula to construct a smooth path through control points. The Bezier curve formula is:

[0138]

[0139] Among them, P0, P1, ..., P n is the control point, and P0=x start ,P n =x goal ;

[0140] According to the actual path requirements, the step size Δt of the Bezier curve is dynamically adjusted to ensure the smoothness and obstacle avoidance capability of the path; then the smoothness and feasibility of the path are optimized by adjusting the control points and step size to generate a smooth path suitable for actual flight;

[0141] In summary, the RRT* algorithm combined with the Bezier curve method can realize the path planning of the tracker tracking the non-cooperative target giant star in a complex environment, and smooth and optimize the generated path to improve the smoothness and operability of the path to meet the needs of actual flight missions;

[0142] Combination Figure 4 As shown in FIG. 1 , an initial path planning diagram obtained by using the RRT* algorithm and the Bezier curve in an embodiment of the present invention, wherein the green line is the original trajectory generated, and the red line is the trajectory obtained after smoothing and optimization. This figure intuitively reflects that the tracker can achieve the approach to a specific target in the giant star system;

[0143] S300, using PEG game strategy for trajectory tracking optimization under the MPC framework, including,

[0144] By combining the dynamic model, input constraints and objective function, a pursuit-escape game model between the target and the tracker is constructed, and a predictive pursuit-escape game controller is designed to achieve path tracking.

[0145] S310, constructing a pursuit-escape game model, based on the dynamic model, input constraints and objective function, constructing a pursuit-escape game model between the target and the tracker; in this model, the target and the tracker act independently to satisfy their respective objective functions;

[0146] S311, establish the dynamic model of the target and the tracker,

[0147] According to the dynamic model of the target and tracker established in step S100, the motion state and control input are described by differential equations:

[0148]

[0149] Among them, xT and x P Represent the state vectors of the target and tracker respectively; u T and u P represent the control input of the target and the tracker respectively; f T and f P denote the dynamics functions of the target and the tracker, respectively;

[0150] S312, setting input constraints for the target and the tracker,

[0151] Set the input constraints of the target and tracker to ensure that the control input is within a reasonable range; the input constraints specifically include velocity, acceleration, and angular velocity:

[0152] u T,min ≤u T ≤u T,max

[0153] u P,min ≤u P ≤u P,max

[0154] Among them, u T,min and u T,max Respectively represent the minimum and maximum values ​​of the target control input; u P,min and u P,max They represent the minimum and maximum values ​​of the tracker control input respectively;

[0155] S313, define the objective function of the target and the tracker,

[0156] Describe the control effects that the target and tracker want to achieve respectively. The objective function of the tracker usually includes minimizing the distance to the target, while the objective function of the target includes maximizing the distance to the tracker:

[0157]

[0158] Among them, J T and J P Represent the objective functions of the target and tracker respectively; L T and L P denote the instantaneous cost functions of the target and tracker respectively; Φ T and Φ P denote the terminal cost functions of the target and tracker respectively;

[0159] S320, constructing a prediction and pursuit game controller,

[0160] Combination Figure 3The MPC principle flow chart shown in the figure uses a double closed-loop system to compare the expected position with the actual position to obtain a reference trajectory, which is then subtracted from the position deviation after the predicted output and subjected to rolling optimization to obtain a control input, which is input into the controlled object and the prediction model. The actual position output by the controlled object and the position output by the prediction model are processed and re-predicted and output, and then an inner-loop control system is formed using negative feedback. The actual position output by the controlled object directly enters the link of forming the reference trajectory to form an outer-loop control system.

[0161] Through the pursuit-escape game model, a predictive pursuit-escape game controller is designed to track the reference trajectory; under the model predictive control (MPC) framework, the multi-objective constrained optimization problem is transformed into a quadratic programming problem to deal with the input constraints;

[0162] S321, prediction model,

[0163] MPC is used to predict the optimal strategy of the target and adjust the optimal strategy of the tracker. MPC uses the dynamic model and constraints of the target and tracker to predict the future N at each control moment. p System behavior over time and optimize control inputs:

[0164]

[0165] Among them, N p is the prediction time domain; N c To control the time domain; R x is the control weight matrix; x P (k|t) and x T (k|t) represents the states of the tracker and target predicted at time t at future time k; Δu P (k|t) represents the increment of control input;

[0166] S322, game solution,

[0167] By repeatedly predicting the opponent's actions and optimizing one's own strategy, the saddle point of the game is solved; the saddle point is the equilibrium point between the tracker and the target, at which the strategies of both parties are optimal, including:

[0168] (1) Initialize the initial strategy of the tracker and target

[0169] (2) Iterative optimization, where the tracker and target update their strategies respectively, predict the opponent's response and adjust their own strategies;

[0170]

[0171] (3) Until the strategies converge and the equilibrium point of the game is reached:

[0172]

[0173] Among them, ε max is the convergence threshold;

[0174] Through the above steps, combined with MPC and PEG game strategies, trajectory tracking and control optimization of non-cooperative targets in approaching space can be achieved; this method can not only handle complex dynamic environments, but also effectively deal with the target's escape strategy, thereby improving the accuracy and robustness of trajectory tracking.

[0175] Combination Figure 5 The figure shows the position tracking diagram of the actual trajectory in the x-axis, y-axis and z-axis in the embodiment of the present invention. It can be seen from the figure that the position of the tracker has a good effect in the obstacle avoidance and pursuit game when approaching a specific target in the giant star system.

[0176] Specific implementation scheme two: The present invention provides a prediction and pursuit game control system applied to approaching non-cooperative giant star constellations in space. The system has a program module corresponding to the above steps, and executes the steps in the above-mentioned prediction and pursuit game control method applied to approaching non-cooperative giant star constellations in space during operation.

[0177] The other combinations and connection relationships of this embodiment are the same as those of the first embodiment.

[0178] Specific implementation scheme three: The present invention provides a computer-readable storage medium, which stores a computer program, and the computer program is configured to implement the steps of a prediction and pursuit game control method applied to approaching a non-cooperative superstar constellation in space when called by a processor.

[0179] The other combinations and connection relationships of this embodiment are the same as those of the first embodiment.

[0180] Although the present invention is disclosed as above, the protection scope of the present invention is not limited thereto. Those skilled in the art may make various changes and modifications without departing from the spirit and scope of the present invention, and these changes and modifications will fall within the protection scope of the present invention.

Claims

1. A prediction and pursuit game control method for approaching non-cooperative giant stars in space, characterized in that: The following steps are involved: S100, establish dynamic models and constraints according to the states of the target and the tracker, It includes abstracting the environment in which the tracker and the target are located during actual flight into a specific three-dimensional mathematical space model through discrete points, defining the body coordinate system and the geocentric inertial system, and describing the conversion relationship between the tracker and the target position through the attitude transformation matrix; establishing the connection between the tracker and the target according to Newton's second law; mapping the orbital motion of the tracker and the orbital motion of the target to this system, assuming that the target is in a near-circular orbit and the distance between the target and the tracker is very small compared to the orbital radius of the target, and then deriving the linear model of relative translational dynamics; S200, using the RRT* algorithm and the Bezier curve to perform path planning, including using the RRT* algorithm to generate a path, and then using the Bezier curve to optimize and smooth the path; S300. Using the PEG game strategy to optimize trajectory tracking under the MPC framework includes building a pursuit-escape game model between the target and the tracker by combining the dynamic model, input constraints and objective functions. In the pursuit-escape game model, the target and the tracker act independently to meet their respective objective functions. Then, a predictive pursuit-escape game controller is built. Under the MPC framework, the multi-objective constraint optimization problem is transformed into a quadratic programming problem, and the saddle point of the game is solved to process the input constraints to achieve path tracking.

2. The prediction pursuit and escape game control method for approaching non-cooperative giant stars in space according to claim 1, characterized in that: In step S100, a dynamic model is established according to the state of the target and the tracker, including: The environment in which the tracker and the target are located during actual flight is abstracted into a specific three-dimensional mathematical space model through discrete points. Using the definition of the northeast sky, a certain point in the mission space is located in the geocentric inertial system O I X I Y I Z I The following is expressed as (x, y, z), defining the body coordinate system O b X b Y b Z b With the Earth-centered inertial system O I X I Y I Z I Describe the translation of the target and tracker positions and the rotation of the attitude respectively. The attitude transformation matrix is ​​used to describe the conversion relationship between the two coordinate systems. According to Newton's second law, the mass m of each body, the moment of inertia I in the body coordinate system, the external force F, the external torque M, the acceleration a, the angular acceleration α, and the angular velocity ω of the body are related: F=ma M=Iα+ω×Iω Describe the orbital motion of the tracker and target and for: Among them, u c and u t are the input control accelerations of the tracker and the target, respectively; μ is the gravitational constant of the central celestial body; r c and r t are the position vectors of the tracker and the target in the geocentric inertial system at the current moment respectively; Then, the motion vector between the tracker and the target is r = r c -r t , mapped to the body coordinate system: in, is the first-order derivative of the angular velocity, is the first-order derivative of the motion vector between the tracker and the target; Assuming that the target is in a near-circular orbit and the distance between the target and the tracker is much smaller than the orbital radius of the target, the linear model of the target's relative translational dynamics is derived: in, B t =-B c , is the state vector, Indicates the tracker's orbital velocity.

3. The prediction pursuit and escape game control method for approaching non-cooperative giant stars in space according to claim 2, characterized in that: In step S100, the following assumptions and constraints are made on the tracker and the target: (1) Assume that the geometric centers of the tracker and target are connected to their respective centers of gravity; (2) Assume that the trajectories of the tracker and the target are smooth enough for tracking; (3) Assume that the tracker and the target are symmetrical rigid bodies; (4) Assume that the speed is low and the starting effect is not considered; (5) Constraint trackers must follow the desired trajectory; (6) The non-cooperative target is constrained to have unknown maneuvers, and the tracker has better maneuverability than the target; (7) Set input constraints for the target and tracker to ensure that the control input is within a reasonable range.

4. The prediction pursuit and escape game control method for approaching non-cooperative giant stars in space according to claim 3, characterized in that: In step S200, it includes: S210, initialize parameters and tree structure, define distance calculation function d(x,y), and set the maximum number of iterations N max , step size Δs, counter k and threshold ε; starting point x start As the root node, initialize the path planning tree structure T: d(x,y)=(x1-y1) 2 +(x2-y2) 2 +(x3-y3) 2 Among them, the parameters are set to: Maximum number of iterations N max =1000; step size Δs=0.1; counter k=0; threshold ε=0.05; S220, iteratively generating a path, generating and optimizing the path through a cyclic iteration process, specifically including: S221, random sampling, randomly generate sampling points x in the planning space rand =(x rand1 ,x rand2 ,x rand3 ); S222, find the nearest node, find the node x closest to the random point in the current tree nearest =argmin x∈T d(x,x rand ); S223, generate extension points, according to the step size Δs, from the nearest node x nearest To randomly sample point x rand Generate a new extension point x in the direction new , S224, collision detection, collision detection, ensure the extension point x new Do not collide with any obstacles; S230, insert a new node. If the extension point does not collide, insert it into the tree structure T and record its parent node information; if the extension point is x away from the target point goal If it is less than the set threshold ε, the path is considered to be found; S240, searching for a path, starting from the target point and tracing back along the parent node pointer through the backtracking tree structure, to generate a path from the starting point to the target point; S250, path optimization and smoothing, after the path is generated in step S240, the path is optimized and smoothed by using a Bezier curve, specifically including: S251, select control points, select several control points according to the path points; S252. Construct a Bezier curve. Use the Bezier curve formula to construct a smooth path through control points. The Bezier curve formula is: Among them, P0, P1, ..., P n is the control point, and P0=x start ,P n =x goal ; According to the actual path requirements, the control points and the step size Δt of the Bezier curve are dynamically adjusted to optimize the smoothness and feasibility of the path and generate a smooth path suitable for actual flight.

5. The prediction pursuit and escape game control method for approaching non-cooperative giant stars in space according to claim 4, characterized in that: In step S300, constructing the pursuit-escape game model includes: S311, establish a dynamic model of the target and the tracker. According to the dynamic model of the target and the tracker established in step S100, use differential equations to describe their motion state and control input: Among them, x T and x P Represent the state vectors of the target and tracker respectively; u T and u P represent the control input of the target and the tracker respectively; f T and f P denote the dynamics functions of the target and the tracker, respectively; S312: Set input constraints for the target and the tracker to ensure that the control input is within a reasonable range; the input constraints are physical constraints of the target and the tracker: in T,min in T in T,max in P,min in P in P,max Among them, u T,min and u T,max Respectively represent the minimum and maximum values ​​of the target control input; u P,min and u P,max They represent the minimum and maximum values ​​of the tracker control input respectively; S313. Define the objective functions of the target and the tracker to describe the control effects that the target and the tracker want to achieve, respectively. The objective function of the tracker is used to minimize the distance to the target, and the objective function of the target is used to maximize the distance to the tracker: Among them, J T and J P Represent the objective functions of the target and tracker respectively; L T and L P denote the instantaneous cost functions of the target and tracker respectively; Φ T and Φ P denote the terminal cost functions of the target and tracker, respectively.

6. The prediction pursuit and escape game control method for approaching non-cooperative giant stars in space according to claim 5, characterized in that: In step S300, constructing a prediction pursuit game controller includes: S321, prediction model, using the MPC framework to predict the optimal strategy of the target and adjust the optimal strategy of the tracker; MPC uses the dynamic model and constraints of the target and tracker to predict the future N at each control moment p System behavior over time and optimize control inputs: Among them, N p is the prediction time domain; N c To control the time domain; R x is the control weight matrix; x P (k|t) and x T (k|t) represents the states of the tracker and target predicted at time t at future time k; Δu P (k|t) represents the increment of control input; S322, game solving, by repeatedly predicting the opponent's actions and optimizing one's own strategy, solving the saddle point of the game, including: (1) Initialize the initial strategy of the tracker and target (2) Iterative optimization, where the tracker and target update their strategies respectively, predict the opponent's response and adjust their own strategies; (3) Until the strategies converge and the equilibrium point of the game is reached: Among them, ε max is the convergence threshold.

7. A prediction and pursuit game control system for approaching non-cooperative giant stars in space, characterized by: The system has a program module corresponding to the steps of any one of claims 1 to 6 above, and executes the steps of the above-mentioned prediction pursuit and escape game control method applied to approaching non-cooperative giant star constellations in space when running.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the computer program is configured to implement the steps of the prediction pursuit and escape game control method applied to approaching a non-cooperative superstar constellation in space as described in any one of claims 1-6 when called by a processor.

Images