An active tracking and capturing system and method for an unmanned aerial vehicle

Abstract

The present invention relates to the technical field of unmanned aerial vehicles, and specifically provides an active tracking and capture system and method for unmanned aerial vehicles; the system includes a target recognition module, a path planning module, and a capture module; the target recognition module includes an active radar detection device for identifying, positioning, and determining the motion state of the captured target i; the path planning module is used for extracting and processing the system position information and target information, and calculating and dynamically optimizing the flight path; the capture module includes an airborne anti-unmanned aerial vehicle target capture device, and the device includes four traction head launching nets, which are automatically aimed through the captured target position and a three-degree-of-freedom pan-tilt, and are used for capturing black-flying unmanned aerial vehicles. The advantages are as follows: it has the capabilities of vertical takeoff and landing and high-speed level flight, can capture invading unmanned aerial vehicles of different sizes, has high positioning accuracy, is flexible in response, can be captured repeatedly, and has no secondary damage; the capture method is flexible and efficient, and there is no risk of secondary damage to the ground.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicles, and in particular to an active tracking and capturing system and method for unmanned aerial vehicles. Background Art

[0002] In recent years, unmanned aerial vehicles have been widely used in various civilian occasions, bringing convenience to the general public. At the same time, they have also caused social problems such as frequent occurrence of illegal flights and increasing threats to important targets and important areas. In addition, the excellent performance of unmanned aerial vehicles in several local wars has accelerated the development and use of military unmanned aerial vehicles, and also posed a great threat to the existing air defense system. At present, there are many countermeasures for unmanned aerial vehicles, such as suppression radio interference, deceptive unmanned aerial vehicle interference, and physical countermeasures such as direct destruction or capture.

[0003] The particle swarm optimization algorithm has the characteristics of high search efficiency, strong versatility, and easy implementation, and has a large application space in the field of solving the route planning of unmanned aerial vehicles. However, the particle swarm optimization algorithm has problems such as local optimum and lack of dynamic adjustment of speed. The accuracy of the planned path given is low, and the kinematics and dynamics characteristics of the unmanned aerial vehicle are not considered, and it only has reference significance. Summary of the Invention

[0004] The present invention provides an active tracking and capturing system and method for unmanned aerial vehicles to solve the above problems.

[0005] The first object of the present invention is to provide an active tracking and capturing system for unmanned aerial vehicles, including a target recognition module, a path planning module, and a capturing module;

[0006] The target recognition module includes an active radar detection device for identifying, positioning, and determining the motion state of the captured target i;

[0007] The path planning module is used for extracting and processing the system position information and target information, and calculating and dynamically optimizing the flight path;

[0008] The capturing module includes an on-board anti-unmanned aerial vehicle target capturing device, and the device includes four traction head launched capture nets, which are automatically aimed through the captured target position and a three-degree-of-freedom pan-tilt, and are used for capturing black-flying unmanned aerial vehicles.

[0009] Preferably, the path planning module includes:

[0010] (1) Establish a dynamic model and a kinematic model of a tail-sitter vertical takeoff unmanned aerial vehicle;

[0011] (2) Obtain the current position, attitude information, captured target information, and environmental information of the unmanned aerial vehicle;

[0012] (3) Use the improved particle swarm optimization algorithm to obtain the global path, and give the capture order and initial flight route for multiple capture targets;

[0013] (4) Taking the drone dynamics and kinematics models as constraint conditions, and taking the drone's initial flight route, the principle of the shortest path, and attitude stability as multiple optimization objectives, use the model predictive control algorithm to give all state variables and energy values in the prediction domain based on the constraint conditions, determine the expected state variables of the drone at the next moment, and dynamically optimize the flight route;

[0014] (5) Convert the output expected state variables into actuator control input quantities, and determine whether the capture conditions are met; if the capture is satisfied, execute the capture, and if not, re-plan the path.

[0015] The second object of the present invention is to provide a method for active tracking and capturing of an unmanned aerial vehicle. Using the above-mentioned unmanned aerial vehicle active tracking and capturing system, the method specifically includes the following steps:

[0016] S1. Use the target recognition module to identify, locate, and determine the motion state of the capture target i; the target recognition module uses an active radar detection device;

[0017] S2. Use the path planning module to extract and process the system position information and target information, and calculate and dynamically optimize the flight path; specifically including:

[0018] S21. Establish the dynamics model and kinematics model of the tail-sitter vertical takeoff unmanned aerial vehicle: Establish the kinematics and dynamics equations of the tail-sitter vertical takeoff unmanned aerial vehicle in the inertial coordinate system {I} = {x i , y i , z i}, and the body coordinate system {B} = {x b , y b , z b}, and then obtain the kinematics model and non-linear dynamics model of the tail-sitter unmanned aerial vehicle;

[0019] S22. Obtain the current position, attitude information, capture target information, and environmental information of the unmanned aerial vehicle; the current position coordinates of the unmanned aerial vehicle are (x0, y0, z0), and the attitude information is expressed as (θ0, ψ0), and the capture target information is:

[0020] S23. Use the improved particle swarm optimization algorithm to obtain the global path, and give the capture order and initial flight route for multiple capture targets;

[0021] S24. With the UAV dynamics and kinematics model as the constraint condition, and the UAV's initial flight route, the principle of the shortest path, and attitude stability as multiple optimization objectives, the model predictive control algorithm is used to give all state variables and energy values in the prediction domain based on the constraint condition, determine the expected state variables of the UAV at the next moment, and dynamically optimize the flight route.

[0022] S25. Convert the output expected state variables into actuator control input quantities, and determine whether the capture condition is satisfied; if the capture is satisfied, execute the capture, and if not, re-plan the path.

[0023] S3. Control the attitude of the UAV and use the capture module to capture target i.

[0024] Preferably, the method for establishing the kinematic equation in step S21 is as follows:

[0025] Select the center of gravity position of the tail-sitter vertical takeoff UAV as the origin O of the body coordinate system b , x b axis points in the forward direction of the UAV's head, y b axis points to the right side of the UAV, z b axis direction is determined by the right-hand rule to point to the center of the earth. The coordinate system is fixedly connected to the body, and the inertial coordinate system is used to determine the spatial position coordinates of the UAV; x i axis points north, y i axis points east, z i points to the center of the earth; the rotation matrix from the body coordinate system to the earth coordinate system is:

[0026]

[0027] where, θ is the roll angle, is the pitch angle, is the yaw angle;

[0028] Based on Newton's law, the kinematic model of the tail-sitter UAV is obtained as:

[0029]

[0030] In the formula, F is the total sum of external forces, m is the body weight, u is the flight speed, L is the angular momentum, and M is the total torque;

[0031] The total external force is: F = F G + F T + F L + F D ; where F G is the body gravity; the gravity F Gb in the body coordinate system is F T is the thrust generated by the four rotors:

[0032]

[0033] In the formula, C T is the rotor thrust coefficient, D is the rotor diameter, ω is the rotor speed, and F L +F D is the pitching moment, lift, and drag force received by the wing;

[0034]

[0035]

[0036] In the formula: S is the fin area, ρ is the air density, C L and C D are the lift coefficient and drag coefficient respectively;

[0037] The moment M is composed of the gyroscopic moment M Tg , the rotor aerodynamic moment M Tc , and the wing aerodynamic moment M A ; The obtained nonlinear dynamic model of the tail-sitter UAV is:

[0038]

[0039] In the formula, (a x , a y , a z ), (v x , v y , v z ), (ω x , ω y , ω z ) are the acceleration, velocity, and angular velocity in the inertial coordinate system respectively, α is the flight angle of attack of the UAV, and (I x , I y , I z ) are the moments of inertia of the body coordinate axes.

[0040] Preferably, the specific steps of obtaining the global path by improving the particle swarm optimization algorithm in step S23 include:

[0041] S231. Input the starting point, capture target position, speed information, and terrain information; establish an initial population with the number of particles being N and the dimension being D;

[0042] S232. Calculate the global optimal route using the adaptive fitness function; the adaptive function includes the route length function, the capture target function, the altitude fluctuation function, and the heading fluctuation function;

[0043] The adaptive fitness function f = λ1f1 + λ2f2 + λ3f3 + λ4f4;

[0044] The flight path length function f1: It is used to calculate the flight path length of the UAV from its actual flight position to the completion of the capture task, which can be obtained by applying the differential method. The expression is:

[0045]

[0046] The passing capture target function f2: The flight path of the UAV must ensure passing through all capture targets, otherwise the flight path is invalid. The expression is:

[0047] The altitude fluctuation function f3: Since the flight altitudes of the capture targets are different, the frequent fluctuations of the UAV in the altitude direction should be minimized; the expression of f3 is:

[0048] The heading fluctuation function f4: The frequent changes of the UAV's flight heading in the horizontal direction should be minimized; the expression of f4 is: I i Record the number of heading changes;

[0049] S233. Update the particle curve to generate an optimized population; adjust the value of the inertia weight ω to:

[0050]

[0051]

[0052] In the formula: p i is the individual extreme value, ω error = 0.8, t max is the maximum number of iterations;

[0053] When updating the position, introduce a random factor to improve the randomness of the particle position: S i (t) = S i (t - 1)(1 + r); in the formula, S i (t) is the position of the i-th particle at the t-th iteration, and r is a random value within a certain range;

[0054] S234. Update the velocity and position of the particle and perform boundary condition processing: If the end condition requirements are met, end the iteration and output the global optimal path; otherwise, go to step S232.

[0055] Preferably, step S24 specifically includes:

[0056] S241. Establish a prediction model for the linear model of the tail-sitter vertical takeoff UAV;

[0057] S242. Rolling Optimization: With the drone dynamics and kinematics model as the constraint condition, and the initial flight route of the drone, the principle of the shortest path, and attitude stability as multiple optimization objectives, based on the prediction model and the constraint condition, obtain the optimization objective;

[0058] S243. Feedback Correction: Transmit the optimization results obtained within each control time domain Nc to the flight control, and the flight control issues instructions to the actuator to achieve trajectory tracking and attitude optimization. At the same time, transmit the position and attitude information to the flight control to achieve real-time state feedback.

[0059] Preferably, the method for establishing the prediction model includes:

[0060] Establish a prediction equation:

[0061] Set the sampling period T to obtain the discretized equation:

[0062]

[0063] Take the state prediction variable as Obtain the discretized prediction equation Y(k|k) = C(k|k)X(k|k);

[0064] Take the prediction time domain as N P , the state vector X P is:

[0065] X p = {X(k + 1|k), X(k + 2|k), …, X(k + N x - 1|k), X(k + N p |k)};

[0066] Y p (k) = {Y(k + 1|k), Y(k + 2|k), …, Y(k + N x - 1|k), Y(k + N p |k)}

[0067] X p (k) = P(k|k)·X(k) + H(k|k)V(k)

[0068] Y p (k) = ψ(k|k)X(k) + Θ(k|k){V p (k|k) - V p (k - 1|k)};

[0069] In the above formula, P(k│k) is the prediction time domain conversion matrix of the state X(k), and the expression is:

[0070]

[0071] H(k│k) is the prediction time domain conversion matrix of the control quantity V(k), and the expression is:

[0072]

[0073] ψ(k|k) is the prediction time domain X(k) conversion matrix of the output Y P (k), and the expression is:

[0074]

[0075] Θ(k|k) is the conversion matrix of the output Y P (K) prediction time domain control quantity V(K), and the expression is:

[0076]

[0077] In the formula

[0078] Take the control time domain as N C , and the attitude control vector is:

[0079] Preferably, the optimization result in step S243 is transmitted to the flight control through the MAVLink protocol, and the flight control issues instructions to the actuator in the Gazebo virtual environment.

[0080] Advantages of the present invention:

[0081] Based on the prior art, a highly maneuverable active capture anti-drone platform is provided. This platform can achieve the positioning, tracking and capture of invading drones, and can capture invading drones of different sizes. It has the advantages of high positioning accuracy, flexible response, repeatable capture, and no secondary damage. This flying wing double-body anti-drone platform also has the capabilities of vertical takeoff and landing and high-speed horizontal flight, with flexible, efficient capture methods and no risk of secondary damage to the ground. When there are multiple capture targets, it is required that the capture anti-drone platform has a reasonable flight route and accurate trajectory tracking performance, which can not only ensure the capture success rate, but also improve the platform work efficiency and reduce the platform power consumption. Brief Description of the Drawings

[0082] Figure 1 It is the flow chart of the method for active tracking and capturing of drones provided by the embodiment of the present invention. Detailed Embodiments

[0083] In order to make the objectives, technical solutions and advantages of the present invention more clear and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and do not constitute a limitation to the present invention.

[0084] Embodiment 1

[0085] This embodiment provides an active tracking and capturing system for an unmanned aerial vehicle, which includes a target recognition module, a path planning module, and a capturing module.

[0086] The described target recognition module is used to identify, locate, and determine the motion state of the capturing target i; the target recognition module adopts an active radar detection device, which avoids the problem of difficult recognition of low, slow, and small capturing targets due to low signal-to-noise ratio, weak optoelectronic signals, low sound, etc., and can accurately detect, locate, identify, and capture targets flying at low altitude and low speed.

[0087] The described path planning module is used to extract and process the position information and target information of the system, and calculate and dynamically optimize the flight path; the path planning module includes:

[0088] (1) Establish the dynamic model and kinematic model of the tail-sitting vertical take-off unmanned aerial vehicle;

[0089] (2) Obtain the current position and attitude information of the unmanned aerial vehicle, the capturing target information, and the environmental information;

[0090] (3) Apply an improved particle swarm algorithm to obtain the global path, and give the capturing order and the initial flight route of multiple capturing targets;

[0091] (4) Take the dynamic and kinematic models of the unmanned aerial vehicle as constraint conditions, and take the initial flight route of the unmanned aerial vehicle, the principle of the shortest path, and attitude stability as multiple optimization objectives. Through the MPC algorithm (model predictive control algorithm), all state variables and energy values in the prediction domain based on the constraint conditions are given, and the expected state variables of the unmanned aerial vehicle at the next moment are determined to dynamically optimize the flight route;

[0092] (5) Convert the output expected state variables into actuator control input variables, and judge whether the capturing conditions are met; if the conditions are met, execute the capture, and if not, re-plan the path.

[0093] The advantage of the above path planning module is that when there are multiple capture targets, the capture order can be obtained through the improved particle swarm optimization algorithm, and the flight route that conforms to the terrain and has the shortest flight distance can be obtained, reducing the battery energy loss caused by unreasonable planning of the capture targets; and the problem that the particle swarm optimization algorithm may fall into local optimum is avoided through the local jump-out strategy. Taking the current capture target as the flight end point of this period, the control quantity range is designed based on the UAV kinematic model, and the MPC control algorithm is applied to obtain the local optimal path. By combining the particle swarm optimization algorithm with the MPC control algorithm, a flight route with high flight route segment, flight safety and stable flight route can be obtained.

[0094] The described capture module includes an on-board anti-UAV target capture device, which is disclosed in the Chinese patent with the authorization announcement number CN217049039U. It ejects four towing heads by a spring to launch a capture net, captures the target position through a camera, and conducts automatic aiming through a three-degree-of-freedom pan-tilt to achieve the capture of unlicensed flying UAVs.

[0095] Embodiment 2

[0096] This embodiment provides a UAV active tracking and capture method, which adopts the above UAV active tracking and capture system, and specifically includes the following steps:

[0097] S1. Use the target recognition module to identify, locate and determine the motion state of the capture target i; the target recognition module adopts an active radar detection system.

[0098] S2. Use the path planning module to extract and process the system position information and target information, and calculate and dynamically optimize the flight path; specifically including:

[0099] S21. Establish the dynamic model and kinematic model of the tail-sitter vertical take-off UAV:

[0100] First, establish the kinematic and dynamic equations of the tail-sitter vertical take-off UAV in the inertial coordinate system {I} = {x i , y i , z i} and the body coordinate system {B} = {x b , y b , z b} as follows:

[0101] The center of gravity position of the tail-sitter VTOL UAV is selected as the origin O of the body coordinate system b , the body coordinate system conforms to the right-hand rule, the x b axis points in the forward direction of the UAV head, the y b axis points to the right side of the UAV, and the z bThe axis direction is determined by the right-hand rule and points to the center of the earth. The coordinate system is fixed to the airframe, and the inertial coordinate system is used to determine the spatial position coordinates of the UAV; the x i axis points north, the y i axis points east, and the z i points to the center of the earth; the rotation matrix from the airframe coordinate system to the earth coordinate system is:

[0102]

[0103] where, θ is the roll angle, is the pitch angle, is the yaw angle;

[0104] Based on Newton's law, the kinematic equation (kinematic model) of the tail-sitter UAV is obtained as:

[0105]

[0106] where, F is the total resultant external force, m is the airframe weight, u is the flight speed, L is the angular momentum, and M is the total torque;

[0107] The resultant external force is: F = F G + F T + F L + F D ; where F G is the airframe gravity, and the gravity F Gb in the airframe coordinate system is F T is the thrust generated by the four rotors:

[0108]

[0109] In the formula, C T is the rotor thrust coefficient, D is the rotor diameter, ω is the rotor speed, F L + F D is the pitch moment, lift force, and drag force received by the wing:

[0110]

[0111]

[0112] In the formula: S is the wing area, ρ is the air density, C L and C D are the lift coefficient and drag coefficient respectively;

[0113] The torque M is composed of the gyroscopic torque M Tg , the rotor aerodynamic torque M Tc , and the wing aerodynamic torque M A ; the non-linear dynamic model of the tail-sitter UAV is obtained as:

[0114]

[0115] where (a x , a y , a z ), (v x , v y , v z ), (ω x , ω y , ω z ) are the acceleration, velocity and angular velocity in the inertial coordinate system respectively, α is the flight angle of attack of the UAV, and (I x , I y , I z ) is the moment of inertia of the airframe coordinate axis.

[0116] S22. Obtain the current position and attitude information of the UAV, and capture the target information and environmental information; the current position coordinates of the UAV are (x0, y0, z0), and the attitude information is expressed as (θ0, ψ0), and the captured target information is:

[0117] S23. Apply the improved particle swarm optimization algorithm to obtain the global path, and give the capture order of multiple captured targets and the initial flight route; the specific steps for the improved particle swarm optimization algorithm to obtain the global path include:

[0118] S231. Input the starting point, the positions of the captured targets, the velocity information, and the terrain information; establish the initial population, with the number of particles being N and the dimension being D;

[0119] S232. Calculate the global optimal route using the adaptive fitness function; the adaptive function includes the route length function, the function of passing through the captured targets, the altitude fluctuation function, and the heading fluctuation function; the route length function f1: used to calculate the flight path length of the UAV from its actual flight position to the completion of the capture task, which can be obtained by applying the differential method, and the expression is:

[0120]

[0121] The function of passing through the captured targets f2: the flight route of the UAV must ensure passing through all the captured targets, otherwise the route is invalid, and the expression is:

[0122] The altitude fluctuation function f3, since the flight altitudes of the captured targets are different, the frequent fluctuations of the UAV in the altitude direction should be minimized. The expression of f3 is:

[0123] The heading fluctuation function f4, the frequent changes in the flight heading of the UAV in the horizontal direction should be minimized; the expression of f4 is: I i Record the number of course changes.

[0124] The fitness function f = λ1f1 + λ2f2 + λ3f3 + λ4f4;

[0125] S233. Update the particle curve to generate an optimized population; adjust the value of the inertia weight ω to:

[0126]

[0127]

[0128] In the formula: p i is the individual extreme value, ω error = 0.8, t max is the maximum number of iterations;

[0129] To reduce the probability of falling into local optimum, when updating the particle position, introduce a random factor to increase the randomness of the particle position: S i (t) = S i (t - 1)(1 + r); in the formula, S i (t) is the position of the i-th particle at the t-th iteration, and r is a random value within a certain range;

[0130] S234. Update the velocity and position of the particle and perform boundary condition processing: If the end condition requirements are met, end the iteration and output the global optimal path; otherwise, go to step S232.

[0131] S24. Taking the UAV dynamics and kinematics models as constraint conditions, and taking the UAV initial flight route, the shortest path principle, and attitude stability as multiple optimization objectives, use the MPC algorithm (Model Predictive Control algorithm) to give all state variables and energy values in the prediction domain based on the constraint conditions, determine the expected state variables of the UAV at the next moment, and dynamically optimize the flight route; specifically including:

[0132] S241. Establish the prediction equation of the tail-sitter vertical takeoff UAV linear model:

[0133]

[0134] Set the sampling period T to obtain the discretized equation:

[0135]

[0136] Take the state prediction variable as Get the discretized prediction equation as:

[0137]

[0138] Y(k|k) = C(k|k)X(k|k)

[0139] Take the prediction time domain as N P , the state vector X P is:[[]]

[0140] X p = {X(k + 1|k), X(k + 2|k), …, X(k + N x - 1|k), X(k + N p |k)};

[0141] Y p (k) = {Y(k + 1|k), Y(k + 2|k), …, Y(k + N x - 1|k), Y(k + N p |k)}

[0142] X p (k) = P(k|k)·X(k) + H(k|k)V(k)

[0143] Y p (k) = ψ(k|k)X(k) + Θ(k|k){V p (k|k) - V p (k - 1|k)}

[0144] In the above formula, P(K|K) is the prediction time domain conversion matrix of state X(K), and the expression is:

[0145]

[0146] H(K│K) is the prediction time domain conversion matrix of control quantity V(K), and the expression is:

[0147]

[0148] ψ(k|k) is the prediction time domain X(K) conversion matrix of output Y P (K), and the expression is:

[0149]

[0150] Θ(k|k) is the conversion matrix of output Y P (K) prediction time domain control quantity V(K), and the expression is:

[0151]

[0152] In the formula

[0153] Take the control time domain as N C , and the attitude control vector is:

[0154]

[0155]

[0156] S242. Rolling Optimization: The initial flight route of the drone, the principle of the shortest path, and attitude stability are taken as multiple optimization objectives. Based on the prediction model and constraint conditions, the optimization objectives are obtained.

[0157] S243. Feedback Correction: Based on the MAVROS function in ROS, the optimization results obtained within each N c are transmitted to the flight controller through the MAVLink protocol. The flight controller issues commands to the actuators in the Gazebo virtual environment to achieve trajectory tracking and attitude optimization. At the same time, the position and attitude information are transmitted to the flight controller to achieve real-time state feedback.

[0158] S25. Convert the output desired state quantity into the actuator control input quantity and determine whether the capture condition is met. If the condition is met, execute the capture; if not, re-plan the path.

[0159] S3. Control the attitude of the drone and use the capture module (the on-board anti-drone target capture device) to capture the target i.

[0160] Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention.

[0161] The above specific implementation manners of the present invention do not constitute a limitation to the protection scope of the present invention. Any other corresponding changes and deformations made according to the technical concept of the present invention should be included within the protection scope of the claims of the present invention.

Claims

1. An active tracking and capturing method for an unmanned aerial vehicle, characterized in that: An active tracking and capture system for unmanned aerial vehicles (UAVs) is adopted. The active tracking and capture system for UAVs includes a target recognition module, a path planning module, and a capture module. The target recognition module includes an active radar detection device for identifying, positioning, and determining the motion state of the capture target i. The path planning module is used for extracting and processing the system position information and target information, and calculating and dynamically optimizing the flight path. The capture module includes an on-board anti-UAV target capture device, which includes four towing head launched capture nets, and automatically aims through the capture target position and a three-degree-of-freedom pan-tilt head for capturing black-flying UAVs. Specifically, it includes the following steps: S1. Use the target recognition module to identify, position, and determine the motion state of the capture target i. The target recognition module uses an active radar detection device. S2. Use the path planning module to extract and process the system position information and target information, and calculate and dynamically optimize the flight path. Specifically, it includes: S21. Establish the kinematic model and dynamic model of the tail-sitter vertical takeoff and landing UAV: Establish the kinematic and dynamic equations of the tail-sitter vertical takeoff and landing UAV in the inertial coordinate system {I} = {x i , y i , z i} and the body coordinate system {B} = {x b , y b , z b}, and then obtain the kinematic model and non-linear dynamic model of the tail-sitter UAV; S22. Obtain the current position and attitude information of the drone, and capture target information and environmental information; the current position coordinates of the drone are (x0, y0, z0), the attitude information is expressed as (θ0, φ0, ψ0), and the captured target information is: , ; S23. Apply an improved particle swarm optimization algorithm to obtain the global path, and give the capture order and initial flight route of multiple capture targets. The specific steps for the improved particle swarm optimization algorithm to obtain the global path include: S231. Input the starting point, capture target position, speed information, and terrain information. Establish the initial population, with the number of particles being N and the dimension being D. S232. Calculate the global optimal route using the adaptive fitness function. The adaptive fitness function includes a flight route length function, a capture target passing function, a height fluctuation function, and a heading fluctuation function. The described adaptive fitness function ; The route length function f 1: It is used to calculate the flight path length of the UAV from its actual flight position to the completion of the capture mission. It can be obtained by applying the differential method, and the expression is: ； The above-mentioned method of capturing the objective function f 2: The flight route of the drone must ensure that it passes through all the captured targets, otherwise the route is invalid. The expression is: ; The height fluctuation function f 3. Since the flying heights of the captured targets are different, frequent fluctuations in the height direction of the UAV should be minimized; The expression is: ; The heading fluctuation function f 4. The frequent changes in the flight heading of the UAV in the horizontal direction should be minimized; f The expression of 4 is: ; Record the number of heading changes; S233. Update the particle curve to generate an optimized population; adjust the inertia weight The value of: ； where: p i is the individual extreme value, ω error = 0.8, t max is the maximum number of iterations; When updating the position, introduce a random factor to increase the randomness of the particle position: S i (t) = S i (t - 1)(1 + r); where S i (t) is the position of the i-th particle at the t-th iteration, and r is a random value within a certain range; S234. Update the speed and position of the particles, and perform boundary condition processing: If the end condition requirements are met, end the iteration and output the global optimal path; otherwise, go to step S232. S24. With the UAV dynamics and kinematics model as the constraint condition, and the UAV initial flight route, the shortest path principle, and attitude stability as multiple optimization objectives, use the model predictive control algorithm to give all state variables and energy values in the prediction domain based on the constraint conditions, determine the expected state variables of the UAV at the next moment, and dynamically optimize the flight route. S25. Convert the output expected state variables into actuator control input variables, and judge whether the capture condition is met. If the capture is satisfied, execute the capture; if not, re-plan the path. S3. Control the attitude of the UAV and use the capture module to capture target i.

2. The active tracking and capturing method for an unmanned aerial vehicle according to claim 1, wherein: The path planning module includes: (1) Establish the dynamics model and kinematics model of a tail-sitter vertical takeoff UAV. (2) Obtain the current position, attitude information, capture target information, and environmental information of the UAV. (3) Apply an improved particle swarm optimization algorithm to obtain the global path, and give the capture order and initial flight route of multiple capture targets. (4) With the UAV dynamics and kinematics model as the constraint condition, and the UAV initial flight route, the shortest path principle, and attitude stability as multiple optimization objectives, use the model predictive control algorithm to give all state variables and energy values in the prediction domain based on the constraint conditions, determine the expected state variables of the UAV at the next moment, and dynamically optimize the flight route. (5) Convert the expected state quantity of the output into the control input quantity of the actuator, and determine whether the capture condition is satisfied; if the capture is satisfied, execute the capture, and if not, re-plan the path.

3. A method for active tracking and capturing of an unmanned aerial vehicle according to claim 1, characterized in that: The method for establishing the kinematic equation of step S21 is as follows: Select the center of gravity position of the tail-sitting vertical take-off unmanned aerial vehicle as the origin O of the body coordinate system b , x b axis points in the direction of the unmanned aerial vehicle's head advancement, y b axis points to the right side of the unmanned aerial vehicle, z b axis direction is determined by the right-hand rule to point towards the center of the earth. The coordinate system is fixedly connected to the body. The inertial coordinate system is used to determine the spatial position coordinates of the unmanned aerial vehicle; x i axis points north, y i axis points east, z i points towards the center of the earth; The rotation matrix from the body coordinate system to the earth coordinate system is: wherein, is the roll angle, is the pitch angle, is the yaw angle; Based on Newton's law, the kinematic model of the tail-sitter UAV is obtained as: ， ； where F is the total external force, m is the weight of the aircraft, is the flight speed, L is the angular momentum, and M is the total torque; The resultant external force is: F = F G + F T + F L + F D ; where F G is the gravity of the airframe; the gravity in the airframe coordinate system is ; F T is the thrust generated by the four rotors: ； where C T is the rotor thrust coefficient, D is the rotor diameter, ω is the rotor speed, and F L + F D are the pitching moment, lift, and drag forces acting on the wing; S S ； Where: S is the fin area, ρ is the air density, C L and C D are the lift coefficient and the drag coefficient, respectively; The torque M consists of the gyroscopic torque M Tg , the rotor aerodynamic torque M Tc , and the wing aerodynamic torque M A ; The obtained nonlinear dynamic model of the tail-sitter UAV is as follows: where (a x , a y , a z ), (v x , v y , v z ), (ω x , ω y , ω z ) are the acceleration, velocity and angular velocity in the inertial coordinate system respectively, α is the flight angle of attack of the UAV, and (I x , I y , I z ) is the moment of inertia of the body coordinate axis.

4. A method for active tracking and capturing of an unmanned aerial vehicle according to claim 1, characterized in that The specific steps of S24 include: S241. Establish a prediction model for the linear model of the tail-sitter vertical takeoff UAV; S242. Rolling optimization: taking the UAV dynamics and kinematic models as constraint conditions, and taking the UAV initial flight route, the principle of the shortest path, and attitude stability as multiple optimization objectives, based on the prediction model and constraint conditions, obtain the optimization objectives; S243. Feedback correction: Transmit the optimization results obtained within each control time domain Nc to the flight control, and the flight control issues commands to the actuator to achieve trajectory tracking and attitude optimization. At the same time, transmit the position and attitude information to the flight control to achieve real-time state feedback.

5. A method for active tracking and capturing of an unmanned aerial vehicle according to claim 4, characterized in that, The method for establishing the prediction model includes: Establish a prediction equation: ; Set the sampling period T to obtain the discretized equation: ； Take the state prediction variable as , and the discretized prediction equation is obtained as follows: ； Take the prediction horizon as N P , the state vector X P is as follows: ； ； In the above formula, P(k│k) is the prediction time domain conversion matrix of the state X(k), and the expression is: H(k│k) is the prediction time domain conversion matrix of the control quantity V(k), and the expression is: ； The transformation matrix for the prediction time domain X(k) of the output YP(k), and the expression is: ； The conversion matrix for outputting the predictive control quantity V(K) in the prediction time domain YP(K), and the expression is: wherein ; Take the control time domain as N C , and the attitude control vector is: ; .

6. The active tracking and capturing method for an unmanned aerial vehicle according to claim 5, characterized in that: The optimization results in step S243 are transmitted to the flight control through the MAVLink protocol, and the flight control issues commands to the actuator in the Gazebo virtual environment.

Images