Method and system for avoiding tracking object by unmanned aerial vehicle based on particle swarm algorithm

Abstract

The invention provides a method and system for avoiding a tracked object by an unmanned aerial vehicle based on a particle swarm algorithm, and relates to the technical field of parameter optimization of avoiding the tracked object by the unmanned aerial vehicle, and the method comprises the following steps: constructing a dynamic model according to the position, speed, acceleration and guidance law of the tracked object; constructing an unmanned aerial vehicle flight simulation model according to the aerodynamic parameters, the control surface response and the environmental interference factors; constructing an evasion action model of the unmanned aerial vehicle, wherein the evasion action model comprises a plurality of evasion schemes with adjustable parameters; based on the dynamic model, the unmanned aerial vehicle flight simulation model and the avoidance action model, performing simulation of an unmanned aerial vehicle avoidance tracking object, and solving a constructed target function through a particle swarm optimization algorithm to obtain a global optimal avoidance tracking object parameter; and performing offline training based on the optimal avoidance tracking object parameter to generate an avoidance strategy library. The method has the advantage of improving the reliability, adaptability and acquisition efficiency of the avoidance parameters of the unmanned aerial vehicle.

Description

Technical Field

[0001] The present invention relates to the technical field of parameter optimization for unmanned aerial vehicle (UAV) evasion of tracked objects, and in particular to a method and system for UAV evasion of tracked objects based on a particle swarm algorithm. Background Art

[0002] With the widespread application of drones in logistics, transportation, disaster relief and other fields, the security threats they face are becoming increasingly complex, especially the threats from dynamic tracking objects.

[0003] Existing avoidance technologies are often based on fixed parameters or empirical rules. This means that existing avoidance action parameters (such as overload and maneuver time) are static and cannot be dynamically adjusted to meet real-time threats. Optimization methods based on genetic algorithms or dynamic programming suffer from slow convergence, making it difficult to meet real-time requirements. Consequently, existing drone avoidance technologies struggle to adapt to the complex and ever-changing behavior and environmental conditions of tracked objects.

[0004] Therefore, it is urgent to study the parameter optimization method for UAV avoidance of tracking objects, so as to improve the reliability, adaptability and acquisition efficiency of UAV avoidance parameters. Summary of the Invention

[0005] The purpose of the present invention is to provide a method and system for UAV avoidance tracking objects based on particle swarm algorithm, which improves the reliability, adaptability and acquisition efficiency of UAV avoidance parameters.

[0006] The present invention is achieved through the following technical solutions: The UAV avoidance tracking object method based on particle swarm optimization includes the following steps: A dynamic model is constructed based on the position, velocity, acceleration, and guidance law of the tracked object. The dynamic model is used to simulate the real-time behavior of the tracked object. A UAV flight simulation model is constructed based on aerodynamic parameters, control surface response, and environmental interference factors. The UAV flight simulation model is used to simulate the flight state of the UAV. Constructing an evasive maneuver model for the UAV, the evasive maneuver model including multiple evasive schemes with adjustable parameters; Based on the dynamic model, UAV flight simulation model and avoidance action model, the UAV avoidance of the tracked object is simulated. The constructed objective function is solved by the particle swarm algorithm to obtain the global optimal avoidance parameters of the tracked object. Offline training is performed based on the optimal parameters for avoiding tracked objects to generate an avoidance strategy library.

[0007] Preferably, the dynamic model is constructed by: ; ; ; in, is the proportionality coefficient, This is the overload instruction for the pitch direction of the tracked object. is the overload instruction for the yaw direction of the tracked object, is the overload limit value, is the flight height of the tracked object, is the relative approach speed, is the relative angular velocity of the UAV in the horizontal direction, is the relative angular velocity of the UAV in the yaw plane, is the acceleration due to gravity; like , then and The value of is transformed: ; .

[0008] Preferably, the method for constructing the UAV flight simulation model is: ; ; ; ; ; ; in, 、 and are the x-axis, y-axis, and z-axis positions of the drone in the ground coordinate system at time t. The dot on top of the letter represents the time derivative. 、 and are the speed, pitch angle and yaw angle of the UAV respectively, For the longitudinal control overload of the UAV, Control the overload for the yaw direction turning of the drone. Control the overload of the drone's pitch and turn direction. is the acceleration due to gravity.

[0009] Preferably, the circumvention scheme includes: Tail-placement maneuver: The horizontal overload of the UAV is directed to the side with the larger angle between the UAV and the tracking object. The pitch overload controls the aircraft to maintain the vertical speed. The yaw overload and vertical speed are the adjustable parameters. Downward maneuver: The UAV performs accelerated linear motion, with the longitudinal size being the maximum value under the constraint adjustment, and the vertical speed being an adjustable parameter; Terminal maneuver: The direction of the drone's horizontal overload is the side with the larger angle between the tracked object and the drone's vector. The pitch direction overload is 0, the longitudinal overload takes the maximum value, and the yaw direction overload is the adjustable parameter.

[0010] Preferably, the model construction method of the tail placement maneuver is: ; ; ; in, 、 and They are the horizontal overload, pitch overload and yaw overload of the drone. is the maximum longitudinal overload of the UAV, Represents the vertical speed proportional coefficient and is a set fixed coefficient. is the adjustment target of the vertical speed as an adjustable parameter, is the actual current vertical speed, is the angle between the UAV velocity vector and the horizontal projection of the UAV to the tracking object vector, The maximum overload that the drone can withstand. is the adjustment target of the yaw direction overload as an adjustable parameter.

[0011] Preferably, the construction method of the down-height maneuver is: ; ; ; in, is the proportional coefficient.

[0012] Preferably, the method for constructing the model of the terminal maneuver is: ; ; .

[0013] Preferably, the objective function includes: Minimize hit penalty : ; in, Represents the end time of the evasive action The off-target amount, is the preset miss threshold, and P is the penalty value; Minimize energy consumption : ; in, Indicates the actual flight speed of the drone. Indicates the minimum flight speed of the drone. and Represents the overload size of the drone in the y and z directions respectively, The starting time of the evasive action; When using the particle swarm algorithm to solve the problem, the individual velocity of the i-th particle The following formula is used for updating: ; in, is the inertia coefficient, is the individual extreme value coefficient, is the group extreme value coefficient, is the random coefficient, Generate a random number between 0 and 1, Generate a d-dimensional independent Gaussian distribution vector with a mean of 0 and a variance of 1 in each dimension. represents the upper limit of the parameter vector, represents the lower limit of the parameter vector, Indicates the current location of the example, represents individual extreme values, Indicates the population extreme value.

[0014] Preferably, the method for generating the evasion strategy library is: Collecting the corresponding optimal tracking object avoidance parameters calculated under different speeds, directions, and altitudes of the tracking object and the UAV; Offline training is performed using a neural network to memorize the optimal tracking object avoidance parameters for all scenarios.

[0015] The present invention also provides a UAV tracking avoidance system based on a particle swarm algorithm, which is applied to the above-mentioned UAV tracking avoidance method based on a particle swarm algorithm, including: The tracking object simulation module is used to build a dynamic model based on the position, velocity, acceleration and guidance law of the tracking object. The dynamic model is used to simulate the real-time behavior of the tracking object. The UAV flight simulation module is used to build a UAV flight simulation model based on aerodynamic parameters, control surface response, and environmental interference factors. The UAV flight simulation model is used to simulate the UAV flight state; An evasive action module is used to build an evasive action model for the UAV, which includes multiple evasive schemes with adjustable parameters; The stochastic particle swarm optimization module is used to simulate the UAV's avoidance of tracked objects based on the dynamic model, the UAV flight simulation model, and the avoidance action model. The constructed objective function is solved by the particle swarm algorithm to obtain the global optimal avoidance parameters. The neural network learning module is used to perform offline training based on the optimal avoidance tracking object parameters to generate an avoidance strategy library.

[0016] The technical solution of the present invention has at least the following advantages and beneficial effects: Compared to the fixed parameters or empirical rules used in existing avoidance technologies, this invention builds multiple avoidance action models with adjustable parameters and optimizes them based on the dynamic behavior of the tracked object and environmental information. This enables more reliable and targeted adjustment of avoidance parameters, improving the adaptability of drones to complex threat environments. The present invention introduces a particle swarm algorithm to solve the constructed objective function, which has the characteristics of fast convergence speed and strong global search capability. It can quickly obtain high-quality avoidance strategies within a limited time and meet the real-time requirements in practical applications. This invention uses offline training based on the optimization results of the particle swarm algorithm to generate an avoidance strategy library covering various environments. In practical applications, the content of the strategy library can be directly called, and high-quality avoidance decisions can be made without the need for real-time complex calculations, significantly improving the response speed and decision-making efficiency of drones when encountering dynamic threats. The present invention integrates dynamic modeling, simulation deduction and intelligent optimization, and can automatically adjust avoidance parameters according to the different behavioral characteristics of the tracked objects, forming diversified and optimized response strategies, greatly enhancing the autonomous survival capability and mission completion rate of the drone in various work areas. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A schematic diagram of a flow chart of a method for avoiding tracking objects by a drone based on a particle swarm algorithm according to Example 1 of the present invention; Figure 2 Schematic diagram of the principle of the UAV avoidance tracking object system based on the particle swarm algorithm provided in Example 2 of the present invention. DETAILED DESCRIPTION

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.

[0019] Example 1 This embodiment provides a method for UAV to avoid tracking objects based on particle swarm algorithm. Figure 1 , including the following steps: Step S1: Construct a dynamic model based on the position, velocity, acceleration and guidance law of the tracked object. The dynamic model is used to simulate the real-time behavior of the tracked object.

[0020] In this embodiment, based on the proportional guidance principle, the method for constructing the dynamic model is: ; ; ; in, is the proportionality coefficient, This is the overload instruction for the pitch direction of the tracked object. is the overload instruction for the yaw direction of the tracked object, is the overload limit value, is the flight height of the tracked object, is the relative approach speed, is the relative angular velocity of the UAV in the horizontal direction, is the relative angular velocity of the UAV in the yaw plane, is the acceleration due to gravity; like , then and The value of is transformed: ; .

[0021] Step S2: Constructing a UAV flight simulation model based on aerodynamic parameters, control surface responses, and environmental interference factors. The UAV flight simulation model is used to simulate the flight state of the UAV.

[0022] As a preferred solution, the method for constructing the UAV flight simulation model is: ; ; ; ; ; ; in, 、 and are the x-axis, y-axis, and z-axis positions of the drone in the ground coordinate system at time t. The dot on top of the letter represents the time derivative. 、 and are the speed, pitch angle and yaw angle of the UAV respectively, For the longitudinal control overload of the UAV, Control the overload for the yaw direction turning of the drone. Control the overload of the drone's pitch and turn direction. is the acceleration due to gravity.

[0023] Step S3: Constructing an evasive action model for the UAV, wherein the evasive action model includes a plurality of evasive schemes with adjustable parameters.

[0024] On this basis, the workaround solutions include: Tail-placement maneuver: The horizontal overload of the UAV is directed to the side with the larger angle between the UAV and the tracking object. The pitch overload controls the aircraft to maintain the vertical speed. The yaw overload and vertical speed are the adjustable parameters. Downward maneuver: The UAV performs accelerated linear motion, with the longitudinal size being the maximum value under the constraint adjustment, and the vertical speed being an adjustable parameter; Terminal maneuver: The direction of the drone's horizontal overload is the side with the larger angle between the tracked object and the drone's vector. The pitch direction overload is 0, the longitudinal overload takes the maximum value, and the yaw direction overload is the adjustable parameter.

[0025] Specifically, the model construction method of the tail placement maneuver is as follows: ; ; ; in, 、 and They are the horizontal overload, pitch overload and yaw overload of the drone. is the maximum longitudinal overload of the UAV, Represents the vertical speed proportional coefficient and is a set fixed coefficient. is the adjustment target of the vertical speed as an adjustable parameter, is the actual current vertical speed, is the angle between the UAV velocity vector and the horizontal projection of the UAV to the tracking object vector, The maximum overload that the drone can withstand. is the adjustment target of the yaw direction overload as an adjustable parameter.

[0026] In addition, the construction method of the down-height maneuver is: ; ; ; in, is the proportional coefficient.

[0027] On the other hand, the model construction method of the terminal maneuver is: ; ; .

[0028] This step decomposes the drone's evasive action into three stages: tail placement, altitude descent, and terminal maneuver, and optimizes the parameters for each stage.

[0029] Step S4: Based on the dynamic model, the UAV flight simulation model and the avoidance action model, the UAV avoidance of the tracked object is simulated, and the constructed objective function is solved by the particle swarm algorithm to obtain the global optimal avoidance tracking object parameters.

[0030] The objective functions include: Minimize hit penalty : ; in, Represents the end time of the evasive action The off-target amount, is the preset miss threshold, and P is the penalty value; Minimizing the hit penalty In the algorithm, if the tracked object hits the drone, J1 is equal to a large penalty value P; if the tracked object misses the target, J1 is equal to the inverse of the miss distance. Minimize energy consumption : ; in, Indicates the actual flight speed of the drone. Indicates the minimum flight speed of the drone. and Represents the overload size of the drone in the y and z directions respectively, The start time of the evasive action.

[0031] The objective function consists of maximizing survival probability and minimizing energy consumption. Maximizing survival probability ensures that the drone prioritizes risk avoidance when facing a tracking threat, preventing being hit or disrupted. Minimizing energy consumption reduces energy consumption during the evasion process, extending flight time and avoiding the waste of resources caused by excessive maneuvers. Combining these two factors enables efficient flight while ensuring safety.

[0032] The random particle swarm algorithm is used to balance maneuverability and energy consumption. Reducing energy consumption provides a larger operating space for subsequent maneuvers and can also prevent individuals from falling into local optimality, thereby increasing the reliability and accuracy of the solution.

[0033] As a further optimization solution, when using the particle swarm algorithm to solve the problem, the individual velocity of the i-th particle The following formula is used for updating: ; in, is the inertia coefficient, is the individual extreme value coefficient, is the group extreme value coefficient, is the random coefficient, Generate a random number between 0 and 1, Generate a d-dimensional independent Gaussian distribution vector with a mean of 0 and a variance of 1 in each dimension. represents the upper limit of the parameter vector, represents the lower limit of the parameter vector, Indicates the current location of the example, represents individual extreme values, Indicates the population extreme value.

[0034] Compared with the traditional particle swarm optimization algorithm which only relies on individual and global optimal guided search, this embodiment introduces Gaussian random perturbation term into the particle velocity update formula. , a certain amount of random perturbation can be injected into the search process, thereby effectively preventing the algorithm from falling into the local optimal solution and improving the global optimization ability. At the same time, the introduction of Gaussian perturbation makes the particle update have directional random jump characteristics, combined with the boundary interval , the perturbation amplitude can be dynamically adjusted according to the scale of the search space, maintaining the search diversity while not over-diverging, thereby covering the entire search domain more efficiently. When solving high-dimensional nonlinear problems involving multiple optimization objectives, namely maximizing the survival probability and minimizing energy consumption, traditional particle swarm optimization is prone to premature convergence problems. The improved solution of this embodiment introduces the inertia coefficient With three control coefficients - , and jointly adjust the search balance to achieve a dynamic trade-off between exploration and utilization, which helps to improve the convergence accuracy and robustness of the algorithm in complex environments.

[0035] Step S5: Perform offline training based on the optimal tracking object avoidance parameters to generate an avoidance strategy library; Based on this solution, the method for generating the avoidance strategy library is: Collecting the corresponding optimal tracking object avoidance parameters calculated under different speeds, directions, and altitudes of the tracking object and the UAV; Offline training is performed using a neural network to memorize the optimal tracking object avoidance parameters for all scenarios.

[0036] In other words, based on offline training of neural networks, when the drone needs to avoid a tracking object, by inputting the speed, direction, and altitude information of the tracking object and its own, it can quickly obtain a better avoidance plan, meeting the rapid decision-making needs of dynamic threats.

[0037] Based on the solution of this embodiment, through multi-objective optimization, the survivability of drones in complex threat scenarios is improved. Example 2 The present invention also provides a UAV tracking avoidance system based on particle swarm algorithm, which is applied to the above-mentioned UAV tracking avoidance method based on particle swarm algorithm, see Figure 2 ,include: The tracking object simulation module is used to build a dynamic model based on the position, velocity, acceleration and guidance law of the tracking object. The dynamic model is used to simulate the real-time behavior of the tracking object. The UAV flight simulation module is used to build a UAV flight simulation model based on aerodynamic parameters, control surface response, and environmental interference factors. The UAV flight simulation model is used to simulate the UAV flight state; An evasive action module is used to build an evasive action model for the UAV, which includes multiple evasive schemes with adjustable parameters; The stochastic particle swarm optimization module is used to simulate the UAV's avoidance of tracked objects based on the dynamic model, the UAV flight simulation model, and the avoidance action model. The constructed objective function is solved by the particle swarm algorithm to obtain the global optimal avoidance parameters. The neural network learning module is used to perform offline training based on the optimal avoidance tracking object parameters to generate an avoidance strategy library.

[0038] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A UAV avoidance tracking method based on particle swarm optimization algorithm is characterized by: The following steps are involved: A dynamic model is constructed based on the position, velocity, acceleration, and guidance law of the tracked object. The dynamic model is used to simulate the real-time behavior of the tracked object. A UAV flight simulation model is constructed based on aerodynamic parameters, control surface response, and environmental interference factors. The UAV flight simulation model is used to simulate the flight state of the UAV. Constructing an evasive maneuver model for the UAV, the evasive maneuver model including multiple evasive schemes with adjustable parameters; Based on the dynamic model, UAV flight simulation model and avoidance action model, the UAV avoidance of the tracked object is simulated. The constructed objective function is solved by the particle swarm algorithm to obtain the global optimal avoidance parameters of the tracked object. Offline training is performed based on the optimal parameters for avoiding tracked objects to generate an avoidance strategy library.

2. The method for avoiding tracking objects by using a UAV based on a particle swarm algorithm according to claim 1, characterized in that: The construction method of the dynamic model is: ； ； ； in, is the proportionality coefficient, This is the overload instruction for the pitch direction of the tracked object. is the overload instruction for the yaw direction of the tracked object, is the overload limit value, is the flight height of the tracked object, is the relative approach speed, is the relative angular velocity of the UAV in the horizontal direction, is the relative angular velocity of the UAV in the yaw plane, is the acceleration due to gravity; like , then and The value of is transformed: ； 。 3. The UAV tracking avoidance method based on particle swarm optimization according to claim 1 is characterized in that: The construction method of the UAV flight simulation model is as follows: ； ； ； ； ； ； in, 、 and are the x-axis, y-axis, and z-axis positions of the drone in the ground coordinate system at time t. The dot on top of the letter represents the time derivative. 、 and are the speed, pitch angle and yaw angle of the UAV respectively, For the longitudinal control overload of the UAV, Control the overload for the yaw direction turning of the drone. Control the overload of the drone's pitch and turn direction. is the acceleration due to gravity.

4. The method for avoiding tracking objects by unmanned aerial vehicles based on particle swarm optimization according to claim 1, characterized in that: The workarounds include: Tail-placement maneuver: The horizontal overload of the UAV is directed to the side with the larger angle between the UAV and the tracking object. The pitch overload controls the aircraft to maintain the vertical speed. The yaw overload and vertical speed are the adjustable parameters. Downward maneuver: The UAV performs accelerated linear motion, with the longitudinal size being the maximum value under the constraint adjustment, and the vertical speed being an adjustable parameter; Terminal maneuver: The direction of the drone's horizontal overload is the side with the larger angle between the tracked object and the drone's vector. The pitch direction overload is 0, the longitudinal overload takes the maximum value, and the yaw direction overload is the adjustable parameter.

5. The method for avoiding tracking objects by a UAV based on a particle swarm algorithm according to claim 4 is characterized in that: The model construction method of the tail placement maneuver is as follows: ； ； ； in, 、 and They are the horizontal overload, pitch overload and yaw overload of the drone. is the maximum longitudinal overload of the UAV, Represents the vertical speed proportional coefficient and is a set fixed coefficient. is the adjustment target of the vertical speed as an adjustable parameter, is the actual current vertical speed, is the angle between the UAV velocity vector and the horizontal projection of the UAV to the tracking object vector, The maximum overload that the drone can withstand. is the adjustment target of the yaw direction overload as an adjustable parameter.

6. The method for avoiding tracking objects by using a UAV based on a particle swarm algorithm according to claim 5, characterized in that: The construction method of the descending high maneuver is as follows: ； ； ； in, is the proportional coefficient.

7. The method for avoiding tracking objects by unmanned aerial vehicles based on particle swarm optimization according to claim 6, characterized in that: The model construction method of the terminal maneuver is: ； ； 。 8. The method for avoiding tracking objects by using a UAV based on a particle swarm algorithm according to claim 1, characterized in that: The objective function includes: Minimize hit penalty : ； in, Represents the end time of the evasive action The off-target amount, is the preset miss threshold, and P is the penalty value; Minimize energy consumption : ； in, Indicates the actual flight speed of the drone. Indicates the minimum flight speed of the drone. and Represents the overload size of the drone in the y and z directions respectively, The starting time of the evasive action; When using the particle swarm algorithm to solve the problem, the individual velocity of the i-th particle The following formula is used for updating: ； in, is the inertia coefficient, is the individual extreme value coefficient, is the group extreme value coefficient, is the random coefficient, Generate a random number between 0 and 1, Generate a d-dimensional independent Gaussian distribution vector with a mean of 0 and a variance of 1 in each dimension. represents the upper limit of the parameter vector, represents the lower limit of the parameter vector, Indicates the current location of the example, represents individual extreme values, Indicates the population extreme value.

9. The method for avoiding tracking objects by a UAV based on a particle swarm algorithm according to claim 1, characterized in that: The method for generating the avoidance strategy library is as follows: Collecting the corresponding optimal tracking object avoidance parameters calculated under different speeds, directions, and altitudes of the tracking object and the UAV; Offline training is performed using a neural network to memorize the optimal tracking object avoidance parameters for all scenarios.

10. A UAV tracking avoidance system based on a particle swarm algorithm, applied to the UAV tracking avoidance method based on a particle swarm algorithm according to any one of claims 1 to 9, characterized in that: include: The tracking object simulation module is used to build a dynamic model based on the position, velocity, acceleration and guidance law of the tracking object. The dynamic model is used to simulate the real-time behavior of the tracking object. The UAV flight simulation module is used to build a UAV flight simulation model based on aerodynamic parameters, control surface response, and environmental interference factors. The UAV flight simulation model is used to simulate the UAV flight state; An evasive action module is used to build an evasive action model for the UAV, which includes multiple evasive schemes with adjustable parameters; The stochastic particle swarm optimization module is used to simulate the UAV's avoidance of tracked objects based on the dynamic model, the UAV flight simulation model, and the avoidance action model. The constructed objective function is solved by the particle swarm algorithm to obtain the global optimal avoidance parameters. The neural network learning module is used to perform offline training based on the optimal avoidance tracking object parameters to generate an avoidance strategy library.

Images