Unmanned aerial vehicle DET optimization control method based on HSMS under pursuit game

Abstract

The invention provides an unmanned aerial vehicle DET optimization control method based on an HSMS under a pursuit game, and relates to the technical field of multi-agent system optimization control. Establishing a motion attitude nonlinear dynamic model and a communication network topological relation of the pursuer unmanned aerial vehicle and the escaper unmanned aerial vehicle; constructing an HSMS from the angles of the pursuer unmanned aerial vehicle and the escaper unmanned aerial vehicle, and further constructing a cost function of the pursuer unmanned aerial vehicle and the escaper unmanned aerial vehicle; establishing a pursuit game mechanism with a plurality of participants; dET optimization control laws are respectively designed for the pursuer unmanned aerial vehicle and the escaper unmanned aerial vehicle by utilizing self-adaptive dynamic programming, so that the control on the pursuer unmanned aerial vehicle and the escaper unmanned aerial vehicle is realized; and constructing an evaluation network to carry out online estimation on the DET optimization control law of the pursuer unmanned aerial vehicle and the escaper unmanned aerial vehicle, and then updating the DET optimization control law. According to the method, the communication burden between the unmanned aerial vehicles is reduced, the system response rate is improved, and unnecessary redundant operation is effectively avoided.

Description

Technical Field

[0001] This invention belongs to the field of multi-agent system optimization control technology, and particularly relates to a DET optimization control method for UAVs based on HSMS under a pursuit-escape game. Background Technology

[0002] With the rapid development of cooperative control of multi-agent systems, drone pursuit and escape game theory has become a hot research topic in this field. In typical application scenarios such as reconnaissance, airspace security, and key area defense, the pursuing drone swarm usually needs to make coordinated decisions to effectively capture highly maneuverable escape targets, while the escaping party also possesses the ability to actively evade and counter strategies. This type of problem is essentially a differential game problem with information asymmetry, dynamic conflict, and real-time constraints; the quality of the control strategy directly determines the success or failure of the mission.

[0003] Sliding mode control is widely used in multi-agent cooperative control due to its robustness to system perturbations and external disturbances. However, traditional sliding mode control strategies typically assume that the ideal sliding surface is reachable, which is often difficult to satisfy in practical control systems. On the other hand, while existing research has attempted to introduce event-triggered mechanisms into sliding mode control to reduce communication and computational load, most of these studies use fixed threshold triggering conditions. Once the threshold parameter is set, it cannot be dynamically adjusted, making it difficult to achieve a dynamic balance between system state convergence and reducing communication load. Furthermore, for adversarial bilateral optimization problems such as the chase-escape game, a unified framework for sliding surface construction and event-triggered cooperative design has not yet been established. Therefore, researching the dynamic event-triggered (DET) optimization control problem of UAVs based on hierarchical sliding-mode surfaces (HSMS) in the chase-escape game has significant theoretical and practical value. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention proposes a DET optimization control method for UAVs based on HSMS under a pursuit-escape game. Under the established nonlinear dynamic model of the motion attitude of the pursuer and escaper UAVs and the communication network topology, HSMS and cost functions are constructed from the perspectives of both the pursuer and escaper UAVs. Then, a pursuit-escape game mechanism with multiple participants is established, and adaptive dynamic programming technology is introduced to design the DET optimization control law and the evaluation network weight update law, thereby achieving the DET optimization control objective of UAVs based on HSMS under the pursuit-escape game.

[0005] The technical solution of this invention is as follows:

[0006] On the one hand, this invention provides a DET optimization control method for unmanned aerial vehicles based on HSMS under escape game theory, including the following steps:

[0007] A nonlinear dynamic model of the motion attitude of the pursuer drone and the escaper drone is established, and the communication network topology of the pursuer drone and the escaper drone is defined; the pursuer drone is the drone of the pursuer; the escaper drone is the drone of the escaper.

[0008] Based on the nonlinear dynamic model of the motion attitude of the pursuer drone and the escaper drone and the topology of the communication network, HSMS is constructed from the perspectives of the pursuer drone and the escaper drone respectively, and then the cost functions of the pursuer drone and the escaper drone are constructed.

[0009] Based on the HSMS and cost function of the pursuer drone and the escapee drone, a pursuit-escape game mechanism with multiple participants is established.

[0010] Based on the established pursuit-escape game mechanism and using adaptive dynamic programming, a DET optimal control law is designed for both the pursuer drone and the escaper drone to achieve control of the pursuer drone and the escaper drone.

[0011] A judgment network is constructed to estimate the DET optimal control law of the pursuer drone and the escaper drone online, and then update the DET optimal control law.

[0012] Furthermore, the establishment of a nonlinear dynamic model of the motion attitude of the pursuer UAV and the escaper UAV, and the definition of the communication network topology between the pursuer UAV and the escaper UAV, are specifically as follows:

[0013] A1: Regarding the pursuer... The motion attitude of the drone was modeled to obtain a nonlinear dynamic model of the motion attitude of the pursuer drone.

[0014] The nonlinear dynamic model of the motion attitude of the pursuer drone is established as follows:

[0015] (1);

[0016] in, The drone is used to represent the pursuer. , , They represent the first as the pursuers. A drone orbiting the body coordinate system axis, shaft and Moment of inertia in the axial direction, This indicates the serial number of the drone acting as the pursuer. The coordinate system is based on the drone's center of mass as the origin, and extends along the longitudinal axis of the drone towards the nose. The axis, along the horizontal axis of the fuselage, points towards the right wing of the UAV. Axis, perpendicular to The plane is downward. axis; , , They represent the first as the pursuers. The roll angle, pitch angle, and yaw angle of each drone; , , They represent , and The first derivative; , , They represent , and The second derivative; Indicating the role of the pursuer The moment of inertia of the motor blades of a drone. Indicating the role of the pursuer Additional perturbations from the drone; , , Indicating the role of the pursuer A drone in the body coordinate system axis, shaft and Axial drag coefficient; , , They represent the first as the pursuers. The control parameters for the roll, pitch, and yaw angles of a single UAV; (collection) Defined as , Indicates the number of drones acting as pursuers;

[0017] Defined as the pursuer The system state variables of a drone And let its state components , , , , , , If we denote the transpose, then the nonlinear dynamic model of the motion attitude of the pursuer UAV is further expressed by formula (1) as:

[0018] (2);

[0019] in, express The first derivative, and , , ; ; To control the input signal;

[0020] A2: Regarding the first as an escapee The motion attitude of the drone is modeled to obtain a nonlinear dynamic model of the motion attitude of the escapee drone.

[0021] The nonlinear dynamic model of the escapee drone's motion attitude is constructed as follows:

[0022] (3);

[0023] in, This refers to the drone acting as an escapee; , , These represent the first as an escapee A drone orbiting the body coordinate system axis, shaft and Moment of inertia in the axial direction, The serial number indicating the drone that escaped; , , These represent the first as an escapee The roll angle, pitch angle, and yaw angle of each drone; , , They represent , and The first derivative; , , They represent , and The second derivative; Indicating the first as an escapee The moment of inertia of the motor blades of a drone. Indicating the first as an escapee Additional perturbations from the drone; , , Indicating the first as an escapee A drone in the body coordinate system axis, shaft and Axial drag coefficient; , , These represent the first as an escapee The control parameters for the roll, pitch, and yaw angles of a single UAV; (collection) Defined as , This indicates the number of drones that escaped;

[0024] Defined as the escapee The system state variables of a drone And let its state components , , , , , Then, as the escapee's first The nonlinear dynamic model of the motion attitude of the UAV can be re-expressed by formula (3):

[0025] (4);

[0026] in, express The first derivative, and , , ; ; To control the input signal;

[0027] Considering the escapee Control input signal of a drone Due to the Bouc-Wen hysteresis effect, the control input signal... The mathematical model is as follows:

[0028] (5);

[0029] in, and These are the input and output of the mathematical model, respectively. , and They are respectively the first as escapees The input control quantities of a UAV for roll angle, pitch angle and yaw angle under Bouc-Wen hysteresis; , and They are respectively the first as escapees The output of a UAV in terms of roll angle, pitch angle and yaw angle under the Bouc-Wen hysteresis effect; , ,and , , , , and All are constants, and they satisfy , and ; ,and , and All are auxiliary variables;

[0030] According to formula (5), we can obtain The time derivative is:

[0031] (6);

[0032] in, and They are respectively and The first derivative; ,and , and The shape stiffness proportionality coefficient of the hysteresis loop representing the Bouc-Wen hysteresis effect satisfies , and ; , and This represents the magnitude of the hysteresis loop; additionally, auxiliary variables... , and Each satisfies , and ; , and The smoothness index represents the hysteresis loop; , and They are defined as , and ,and , and They are respectively , and The first derivative;

[0033] Based on formulas (4), (5), and (6), the first as the escapee The nonlinear dynamic model of the motion attitude of the UAV is further expressed as follows:

[0034] (7);

[0035] A3: Define the communication network topology between the pursuer drone and the escapee drone;

[0036] Specifically, the communication network topology between the pursuer drones is defined as an undirected graph. ,in, and They are the vertex set and the undirected edge set, respectively. Indicating the role of the pursuer Each drone corresponds to a vertex, and each edge represents a communication link between the corresponding pursuer drones; define... As vertices The neighborhood group, Indicates the number of the neighboring vertex. Representing neighboring vertices, Represents vertices and The edge between; It is an adjacency matrix. Representing Euclidean space, Representing the adjacency matrix The Line number Column elements, and satisfying and , Representing the adjacency matrix The Line number Column elements;

[0037] Define the communication network topology between escapee drones as an undirected graph. ,in, and They are the vertex set and the undirected edge set, respectively. Indicating the first as an escapee Each drone corresponds to a vertex, and each edge represents a communication link between the escapee drones corresponding to that vertex; Define... As vertex The neighborhood group, Indicates the number of the neighboring vertex. Representing neighboring vertices, Represents vertices and The edge between; It is an adjacency matrix. Representing the adjacency matrix The Line number Column elements, and satisfying and , Adjacency matrix The Line number Column elements;

[0038] Define the communication network topology between the pursuer drone and the escapee drone as an undirected graph. ,in, and These are the vertex set and the undirected edge set, respectively. A hunter drone and The escapee drones are located in two different sectors. and All drones exchange information between two different sectors; for those containing A sector of a hunter drone In other words, if the first The hunter drone was able to capture the first Information about each escapee drone, and the connectivity weights between them. ,on the contrary, For those containing A sector of an escaped drone In other words, if the first The escapee can obtain the first Information about each pursuer, and the connection weights between them. ,on the contrary, .

[0039] Furthermore, based on the nonlinear dynamic model of the motion attitude of the pursuer UAV and the escaper UAV and the communication network topology, the HSMS is constructed from the perspectives of the pursuer UAV and the escaper UAV respectively, and then the cost functions of the pursuer UAV and the escaper UAV are constructed, specifically including:

[0040] B1: Based on the nonlinear dynamic model of the hunter drone's motion attitude and the communication network topology, construct the HSMS of the hunter drone's angles, and further design the control input signals in the nonlinear dynamic model of the hunter drone's motion attitude based on the HSMS of the hunter drone's angles. ;

[0041] From the perspective of the hunter drone, the HSMS is constructed as follows:

[0042] (8);

[0043] in, Indicates from the first The HSMS is constructed from the perspective of a hunter drone, and for The One element, Indicates element number and ; Defined as , for The One element, , ,in, For the system state variables of the hunter drone corresponding to the neighboring vertex; for The One element, For time variables, For integration variables, express The The elements are in the integral variable. The value at time, It is a positive definite matrix; Let be a diagonal matrix, where, and All are positive constants; Defined as and According to the iterative algorithm, we get and Further , , For number index, for The One element, If it is a constant, ,but , for The One element, and vice versa. ;

[0044] From formula (8), we can see that the first... The HSMS of the individual hunter drone is re-expressed as follows:

[0045] (9);

[0046] in, It is a lower triangular matrix and ;

[0047] Based on formulas (2) and (9), the dynamic calculation of the HSMS of the hunter drone angle is as follows:

[0048] (10);

[0049] in, for The first derivative, , , For the control input signal of the hunter drone corresponding to the neighboring vertex;

[0050] According to formulas (9) and (10), the first The hunter drone in its first Control quantities on layer HSMS , and This can be expressed as:

[0051] (11);

[0052] in, , , They represent the first The hunter drone in its first Control parameters for roll angle, pitch angle and yaw angle on the layer HSMS; , , They represent the first The hunter drone in its first The control parameters for roll angle, pitch angle, and yaw angle on the layer HSMS, and they satisfy... ; , , They represent the first The hunter drone in its first The equivalent control values ​​for roll angle, pitch angle, and yaw angle on the layer HSMS, and they satisfy... ; , , They represent the first The hunter drone in its first The switching control values ​​for roll angle, pitch angle, and yaw angle on the layer HSMS, and satisfying ;

[0053] According to the iterative algorithm, formula (11) can be rewritten as:

[0054] (12);

[0055] in, Indicates the index number;

[0056] Based on formula (12), we can further obtain the result in formula (2). for:

[0057] (13);

[0058] in, ; ;

[0059] B2: Based on the nonlinear dynamic model of the escaped UAV's motion attitude and the communication network topology, construct the HSMS of the escaped UAV's angles, and further design the control input signals in the nonlinear dynamic model of the escaped UAV's motion attitude based on the HSMS of the escaped UAV's angles. ;

[0060] Constructing HSMS from the perspective of escaped drones:

[0061] (14);

[0062] in, Indicates from the first HSMS constructed from the perspective of an escaped drone. for The One element, ; Defined as , for The One element, , , for The One element, express The The elements are in the integral variable. The value at time, It is a positive definite matrix; It is a lower triangular matrix and , For the system state variables of the escaped drone corresponding to the neighboring vertex;

[0063] definition It is a diagonal matrix. and All are constants; defined and According to the iterative algorithm, we can obtain and Further, we can obtain , If it is a constant, ,but , for The One element, and vice versa. ;

[0064] Based on formulas (7) and (14), the dynamic calculation of the HSMS of the escaped UAV angle is as follows:

[0065] (15);

[0066] in, for The first derivative, , , For the control input signal of the escapee drone corresponding to the neighboring vertex;

[0067] According to formulas (14) and (15), under the Bouc-Wen hysteresis effect, the first The escape drone in its first Control quantities on layer HSMS , and This can be expressed as:

[0068] (16);

[0069] in, , and These represent the effects of the Bouc-Wen hysteresis, specifically the... The escape drone in its first Control parameters for roll angle, pitch angle and yaw angle on the layer HSMS; , and These represent the effects of the Bouc-Wen hysteresis, specifically the... The escape drone in its first The control parameters for roll angle, pitch angle, and yaw angle on the layer HSMS, and they satisfy... ; , and These represent the effects of the Bouc-Wen hysteresis, specifically the... The escape drone in its first The equivalent control values ​​for roll angle, pitch angle, and yaw angle on the layer HSMS, and they satisfy... ; , and These represent the effects of the Bouc-Wen hysteresis, specifically the... The escape drone in its first The switching control variables for roll angle, pitch angle, and yaw angle on the layer HSMS, and they satisfy... ;

[0070] According to the iterative algorithm, formula (16) can be rewritten as:

[0071] (17);

[0072] Based on formula (17), we can further obtain the result in formula (7). for:

[0073] (18);

[0074] in, ; ;

[0075] B3: Based on the control input signal in the nonlinear dynamics model of the hunter drone's motion attitude. Control input signals in the nonlinear dynamics model of the motion attitude of the escape drone Establish cost functions for the pursuer drone and the escapee drone respectively;

[0076] The attitude response of the Chaser drone can be characterized by the following cost function:

[0077] (19);

[0078] in, Indicates the first The cost function of a hunter drone; Defined as ;

[0079] (20);

[0080] in, It is a positive constant; , and All are positive definite matrices; ,in, It is a non-quadratic function. It is a constant. For integration variables;

[0081] The attitude response of the escapee drone can be characterized by the following cost function:

[0082] (twenty one);

[0083] in, Indicates the first The cost function of an escapee drone; Defined as ;

[0084] (twenty two);

[0085] in, It is a positive constant; , and All are positive definite matrices; ,and It is a constant. For integration variables, It is a non-quadratic function.

[0086] Furthermore, the establishment of a pursuit-escape game mechanism with multiple participants based on the HSMS and cost function of the pursuer drone and the escaper drone is specifically as follows: Define the first... The optimal equivalent control law and optimal switching control law for the hunter drone are as follows: and And define the first The optimal equivalent control law and optimal switching control law for the escaped drone under the Bouc-Wen hysteresis effect are as follows: and If the following conditions are met:

[0087] (twenty three);

[0088] (twenty four);

[0089] but , , and The Nash equilibrium constituting the pursuit-escape game, in which, For the first Besides the individual hunter drone, the set of optimal control laws for other hunter drones. For the first The set of optimal control laws for other escaped drones under the Bouc-Wen hysteresis effect, excluding the escaped drone.

[0090] Furthermore, based on the established pursuit-escape game mechanism and utilizing adaptive dynamic programming, a DET-optimized control law is designed for both the pursuer and escape drones to achieve control over them. The specific steps are as follows:

[0091] C1: Based on the established pursuit-escape game mechanism and utilizing adaptive dynamic programming, design the first... Optimal equivalent control law for a hunter drone and optimized switching control law ;

[0092] First, based on the HSMS of the Chaser drone, the first... The Hamiltonian function for the hunter drone is:

[0093] (25);

[0094] in, For the first Hamiltonian functions for a hunter drone for The gradient;

[0095] Combining formulas (10) and (25), we obtain the first... The Hamilton-Jacobi-Bellman equations for the hunter drone are as follows:

[0096] (26);

[0097] in, For the first Hamilton-Jacobi-Bellman equations for a single hunter drone; , Let's define the optimized equivalent control law for the hunter drone corresponding to the neighboring vertex. An optimized switching control law for the hunter drones corresponding to neighboring vertices;

[0098] Based on formula (26) and by solving and ,get and The specific form is as follows:

[0099] (27);

[0100] (28);

[0101] C2: Based on the established pursuit-escape game mechanism and utilizing adaptive dynamic programming, design the first... The optimal equivalent control law of an escaped drone under the Bouc-Wen hysteresis effect and optimized switching control law ;

[0102] Based on the escapee drone, the HSMS defines the first The Hamiltonian function for an escapee drone is:

[0103] (29);

[0104] in, For the first Hamiltonian function of an escaped drone for The gradient;

[0105] According to formulas (15) and (29), we obtain the first... The Hamilton-Jacobi-Bellman equations for an escapee drone are as follows:

[0106] (30);

[0107] in, For the first Hamilton-Jacobi-Bellman equations for an escapee drone; ;

[0108] Based on formula (30) and by solving and

[0109] ,get and The specific form is as follows:

[0110] (31);

[0111] (32);

[0112] C3: Design the difference function between the trigger state variable and the system state variable of the pursuer drone, and the difference function between the trigger state variable and the system state variable of the escapee drone;

[0113] definition and The first Each triggering moment and its corresponding time sequence For the moment of triggering the number and , Represents the set of natural numbers; satisfy , For the first A triggering moment, and when hour Then, the first The pursuer drone and the first The escape drone in the first The trigger state variables at each trigger instant are respectively represented as follows: and ;No. The pursuer drone and the first The time series of the triggering state variables of each escape drone are denoted as follows: and ;

[0114] when At that time, the first The difference function between the trigger state variable and the system state variable of the individual hunter drone is:

[0115] (33);

[0116] in, For the first The difference function between the trigger state variables and the system state variables of a hunter drone;

[0117] when At that time, the first The difference function between the trigger state variable and the system state variable of the escaped drone is:

[0118] (34);

[0119] in, For the first The difference function between the trigger state variables and the system state variables of an escapee drone;

[0120] C4: Design No. The DET-optimized control law of a hunter drone, including the DET-optimized equivalent control law and the DET-optimized switching control law;

[0121] According to formulas (27), (28), and (33), we can obtain:

[0122] (35);

[0123] (36);

[0124] in, and These respectively represent the state variables that are triggered. The next DET-optimized equivalent control law and DET-optimized switching control law for a single hunter drone; ; Indicates the first The first trigger moment HSMS of a hunter drone; Indicates that the state variable is triggered Below, except for the first Optimized control input signals for all other Hunter drones besides the Hunter drone; for The gradient; Indicates that the state variable is triggered The next Control gain of the hunter drone;

[0125] C5: Design No. The DET optimal control law of an escaped UAV under the Bouc-Wen hysteresis effect, including the DET optimal equivalent control law and the DET optimal switching control law;

[0126] According to formulas (31), (32), and (34), the first... An escapee drone under the Bouc-Wen hysteresis effect and They are designed as follows:

[0127] (37);

[0128] (38);

[0129] in, and These respectively represent the state variables that are triggered. The next DET-optimized equivalent control law and DET-optimized switching control law for an escapee drone; ; for The gradient; Indicates that the state variable is triggered The next Control gain of the escapee drone; To indicate the first The first trigger moment HSMS of an escapee drone Indicates that the state variable is triggered Below, except for the first Optimized control input signals for all escape drones except for the escape drone;

[0130] C6: Based on the difference function between the trigger state variables and system state variables of the pursuer drone, the difference function between the trigger state variables and system state variables of the escaper drone, and the DET optimization control law of the pursuer drone and the escaper drone, the control of the pursuer drone and the escaper drone is realized.

[0131] Specifically: When the difference function between the trigger state variable of the Chaser UAV and the system state variable meets the preset trigger condition, the DET optimization control law of the Chaser UAV is used to control the Chaser UAV; otherwise, the current control is maintained.

[0132] When the difference function between the escaped drone's trigger state variable and the system state variable meets the preset trigger condition, the escaped drone's DET optimized control law is used to control the escaped drone; otherwise, the current control is maintained.

[0133] Furthermore, the constructed evaluation network performs online estimation of the DET optimal control law for both the pursuer and the escapee drones, and then updates the DET optimal control law. Specific steps include:

[0134] Approximation is achieved by implementing a judging neural network. and ,get:

[0135] (39);

[0136] (40);

[0137] in, , , The first The evaluation neural network weights, activation function, and approximation error of a single hunter drone; All Elements in; All Elements in; The number of neural nodes; , , The first The evaluation neural network weights, activation function, and approximation error of an escaped drone; All Elements in; All Elements in;

[0138] Based on formulas (39) and (40), we can further obtain:

[0139] (41);

[0140] (42);

[0141] in, , and They are respectively , and The gradient; , and They are respectively , and The gradient;

[0142] According to formulas (35), (36), and (41), we can obtain:

[0143] (43);

[0144] (44);

[0145] in, , and They are respectively , and The estimate;

[0146] According to formulas (37), (38), and (42), we can obtain:

[0147] (45);

[0148] (46);

[0149] in, , and They are respectively , and The estimate;

[0150] According to formula (25), we can obtain the first... The Bellman residuals for each hunter drone are:

[0151] (47);

[0152] in, Indicates the first Bellman residuals of a hunter drone for The estimate; Indicates that the state variable is triggered Below, except for the first Besides the individual hunter drone, the optimized control input signals for all other hunter drones, and for The estimate;

[0153] According to formula (29), we can obtain the first... The Bellman residuals for each escapee drone are:

[0154] (48);

[0155] in, Indicates the first Bellman residuals of an escapee drone for The estimate; Indicates that the state variable is triggered Below, except for the first Besides the escapee drone, the optimized control input signals for all other escapee drones, and for The estimate;

[0156] Based on formula (47) and by using gradient descent and standardization principles, the first... The evaluation network weight update law for the escaped drone: For the first... The hunter drone and the first Each escapee drone is defined as having an experienced recovery status. and Their corresponding time series are respectively and ,in, For the first At the moment of triggering, , , Represents the set of natural numbers;

[0157] Design No. The evaluation network weight update law for each hunter drone is as follows:

[0158] (49);

[0159] in, Indicates the first The evaluation network weight update law for individual hunter drones; and For update rate; and Used for standardization;

[0160] (50);

[0161] in, Indicates the state of experience recycling. The next The DET-optimized equivalent control law for a hunter drone. for The estimate; Indicates the state of experience recycling. The next DET-optimized switching control law for a single hunter drone for The estimate; Indicates the state of experience recycling. Below, except for the first The set consisting of the estimates of the DET optimal control law of all the other hunter drones besides the hunter drone. and They represent the first time. The first trigger moment HSMS and state error of a single hunter drone;

[0162] (51);

[0163] in, Indicates the state of being triggered. The next Control gain of the hunter drone; Indicates the first At the moment of triggering, and the first The adjacent hunter drone The experience recovery status of the hunter drone; Indicates the state of being triggered. Below, with the The adjacent hunter drone Control gain of the hunter drone; Indicates the state of being triggered. Below, with the The adjacent hunter drone The DET-optimized equivalent control law for a hunter drone, and for The estimate; Indicates the state of being triggered. Below, with the The adjacent hunter drone The DET-optimized switching control law for a hunter drone, and for The estimate; Indicates the state of being triggered. The next Control gain of the escapee drone; Indicates the state of experience recycling. The next The DET-optimized equivalent control law for an escapee drone. for The estimate; Indicates the state of experience recycling. The next DET-optimized switching control law for an escapee drone for The estimate;

[0164] (52);

[0165] in, Indicates the state of experience recycling. The next The DET-optimized equivalent control law for an escapee drone. for The estimate; Indicates the state of experience recycling. The next DET-optimized switching control law for an escapee drone for The estimate;

[0166] Based on formula (48) and by using gradient descent and standardization principles, the first... The evaluation network weight update law for each escapee drone is as follows:

[0167] (53);

[0168] in, Indicates the first The evaluation network weight update law for individual escapee drones; and For update rate; and Used for standardization;

[0169] (54);

[0170] in, Indicates the state of experience recycling. The next The DET-optimized equivalent control law for an escapee drone. for The estimate; Indicates the state of experience recycling. The next DET-optimized switching control law for an escapee drone for The estimate; Indicates the state of experience recycling. Below, except for the first The set of DET-optimized control laws for all escape drones except for the escape drone; and They represent the first time. The first trigger moment HSMS and state error of an escaped drone;

[0171] (55);

[0172] in, Indicates the first At the moment of triggering, and the first The escaped drone adjacent to the first The experience recovery status of the escaped drone; Indicates the state of being triggered. Below, with the The escaped drone adjacent to the first Control gain of the escapee drone; Indicates the state of being triggered. Below, with the The escaped drone adjacent to the first The DET-optimized equivalent control law for an escapee drone, and for The estimate; Indicates the state of being triggered. Below, with the The escaped drone adjacent to the first The DET-optimized switching control law for an escapee drone, and for The estimate;

[0173] (56);

[0174] in, and They represent the first time. The first trigger moment HSMS and state error of an escaped drone; Indicates the first At the moment of triggering, and the first The escaped drone adjacent to the first The experience recovery status of the escaped drone; Indicates the state of being triggered. Below, with the The escaped drone adjacent to the first Control gain of the escapee drone; Indicates the state of being triggered. Below, with the The escaped drone adjacent to the first The DET-optimized equivalent control law for an escapee drone, and for The estimate; Indicates the state of being triggered. Below, with the The escaped drone adjacent to the first The DET-optimized switching control law for an escapee drone, and for The estimate; Indicates the state of being triggered. The next Control gain of the escapee drone; Indicates the state of experience recycling. The next The DET-optimized equivalent control law for a hunter drone. for The estimate; Indicates the state of experience recycling. The next DET-optimized switching control law for a single hunter drone for The estimate.

[0175] Secondly, this application proposes an electronic device, including: one or more processors, and a memory, wherein the memory is used to store instructions, and when the instructions are executed by the one or more processors, the one or more processors cause the one or more processors to execute the aforementioned UAV DET optimization control method based on HSMS under the pursuit and escape game.

[0176] Thirdly, this application proposes a computer-readable storage medium storing executable instructions that, when executed, cause a processor to perform the HSMS-based UAV DET optimization control method under the pursuit-escape game.

[0177] Fourthly, this application proposes a computer program product, including a computer program or instructions, which, when executed by a processor, implements the aforementioned HSMS-based UAV DET optimization control method under the pursuit-escape game.

[0178] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0179] This invention employs a divide-and-conquer algebraic graph approach to model the dynamic interaction between pursuer and escaper UAVs, overcoming the limitations of traditional fixed topology descriptions in dynamic environments. To reduce the communication burden between UAVs and improve system response speed, a DET-based optimization control method based on HSMS is proposed. This method determines whether to maintain or update the control law by capturing event-triggered state information, effectively avoiding unnecessary redundant operations. Compared to adaptive dynamic programming methods using an actor-critic network architecture, this invention uses a single-critic network architecture, reducing computational burden. Furthermore, it designs a weight update law using gradient descent and standardization principles, enhancing the adaptability of the UAV system in real-time control tasks. Attached Figure Description

[0180] Figure 1 This is a flowchart of the UAV DET optimization control method based on HSMS under the pursuit and escape game in an embodiment of the present invention;

[0181] Figure 2 This is an undirected topological graph showing the relationship between the pursuer drone and the escapee drone in the simulation of this embodiment of the invention.

[0182] Figure 3 The above are the HSMS response curves of the pursuer drone and the escapee drone in the simulation of the embodiments of the present invention;

[0183] Among them, (a) is the response curve of the HSMS of the first pursuer drone; (b) is the response curve of the HSMS of the second pursuer drone; (c) is the response curve of the HSMS of the first escapee drone; and (d) is the response curve of the HSMS of the second escapee drone.

[0184] Figure 4 This is a DET optimized control law response curve of the first pursuer UAV in the simulation of the embodiment of the present invention for its roll angle, pitch angle and yaw angle;

[0185] Figure 5 This is a DET optimized control law response curve of the second pursuer UAV in the simulation of the embodiment of the present invention for its roll angle, pitch angle and yaw angle;

[0186] Figure 6The graph shows the DET optimized control law response curves of the first escapee UAV in the simulation of the embodiment of the present invention for its roll angle, pitch angle and yaw angle.

[0187] Figure 7 The graph shows the DET optimized control law response curves of the second escapee UAV in the simulation of the embodiment of the present invention for its roll angle, pitch angle and yaw angle. Detailed Implementation

[0188] To enable those skilled in the art to better understand the present invention, the technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. However, it should be understood that, without further description, the present invention can be advantageously incorporated into other embodiments.

[0189] like Figure 1 As shown, this invention provides a DET optimization control method for unmanned aerial vehicles (UAVs) based on HSMS under a pursuit-escape game, comprising the following steps:

[0190] S1: Establish a nonlinear dynamic model of the motion attitude of the pursuer drone and the escaper drone, and define the communication network topology of the pursuer drone and the escaper drone; the pursuer drone is the drone of the pursuer; the escaper drone is the drone of the escaper.

[0191] The specific steps include:

[0192] S1.1: Regarding the pursuer... The motion attitude of the drone was modeled to obtain a nonlinear dynamic model of the motion attitude of the pursuer drone.

[0193] The nonlinear dynamic model of the motion attitude of the pursuer drone is established as follows:

[0194] (1);

[0195] in, The drone is used to represent the pursuer. , , They represent the first as the pursuers. A drone orbiting the body coordinate system axis, shaft and Moment of inertia in the axial direction, This indicates the serial number of the drone acting as the pursuer. The coordinate system is based on the drone's center of mass as the origin, and extends along the longitudinal axis of the drone towards the nose. The axis, along the horizontal axis of the fuselage, points towards the right wing of the UAV. Axis, perpendicular to The plane is downward. axis; , , They represent the first as the pursuers. The roll angle, pitch angle, and yaw angle of each drone; , , They represent , and The first derivative; , , They represent , and The second derivative; Indicating the role of the pursuer The moment of inertia of the motor blades of a drone. Indicating the role of the pursuer Additional perturbations from the drone; , , Indicating the role of the pursuer A drone in the body coordinate system axis, shaft and Axial drag coefficient; , , They represent the first as the pursuers. The control parameters for the roll, pitch, and yaw angles of a single UAV; (collection) Defined as , Indicates the number of drones acting as pursuers;

[0196] Defined as the pursuer The system state variables of a drone And let its state components , , , , , , If we denote the transpose, then the nonlinear dynamic model of the motion attitude of the pursuer UAV is further expressed by formula (1) as:

[0197] (2);

[0198] in, express The first derivative, and , , ; ; To control the input signal;

[0199] S1.2: Regarding the first as an escapee The motion attitude of the drone is modeled to obtain a nonlinear dynamic model of the motion attitude of the escapee drone.

[0200] The nonlinear dynamic model of the escapee drone's motion attitude is constructed as follows:

[0201] (3);

[0202] in, This refers to the drone acting as an escapee; , , These represent the first as an escapee A drone orbiting the body coordinate system axis, shaft and Moment of inertia in the axial direction, The serial number indicating the drone that escaped; , , These represent the first as an escapee The roll angle, pitch angle, and yaw angle of each drone; , , They represent , and The first derivative; , , They represent , and The second derivative; Indicating the first as an escapee The moment of inertia of the motor blades of a drone. Indicating the first as an escapee Additional perturbations from the drone; , , Indicating the first as an escapee A drone in the body coordinate system axis, shaft and Axial drag coefficient; , , These represent the first as an escapee The control parameters for the roll, pitch, and yaw angles of a single UAV; (collection) Defined as , This indicates the number of drones that escaped;

[0203] Defined as the escapee The system state variables of a drone And let its state components , , , , , Then, as the escapee's first The nonlinear dynamic model of the motion attitude of the UAV can be re-expressed by formula (3):

[0204] (4);

[0205] in, express The first derivative, and , , ; ; To control the input signal;

[0206] In the pursuit and escape game, the escaping drone often exhibits Bouc-Wen hysteresis, stemming from multivariate coupling and uncertainties external to the system. Bouc-Wen hysteresis typically manifests as a delay in the response time of the control input signal. In this invention, the escaping drone is considered as the... Control input signal of a drone Due to the Bouc-Wen hysteresis effect, the control input signal... The mathematical model is as follows:

[0207] (5);

[0208] in, and These are the input and output of the mathematical model, respectively. , and They are respectively the first as escapees The input control quantities of a UAV for roll angle, pitch angle and yaw angle under Bouc-Wen hysteresis; , and They are respectively the first as escapees The output of a UAV in terms of roll angle, pitch angle and yaw angle under the Bouc-Wen hysteresis effect; , ,and , , , , and All are constants, and they satisfy , and ; ,and , and All are auxiliary variables;

[0209] According to formula (5), we can obtain The time derivative is:

[0210] (6);

[0211] in, and They are respectively and The first derivative; ,and , and The shape stiffness proportionality coefficient of the hysteresis loop representing the Bouc-Wen hysteresis effect satisfies , and ; , and This represents the magnitude of the hysteresis loop; additionally, auxiliary variables... , and Each satisfies , and ; , and The smoothness index represents the hysteresis loop; , and They are defined as , and ,and , and They are respectively , and The first derivative;

[0212] Based on formulas (4), (5), and (6), the first as the escapee The nonlinear dynamic model of the motion attitude of the UAV can be further expressed as:

[0213] (7);

[0214] S1.3: Define the communication network topology between the pursuer drone and the escapee drone;

[0215] Specifically, the communication network topology between the pursuer drones is defined as an undirected graph. ,in, and They are the vertex set and the undirected edge set, respectively. Indicating the role of the pursuer Each drone corresponds to a vertex, and each edge represents a communication link between the corresponding pursuer drones; define... As vertex The neighborhood group, Indicates the number of the neighboring vertex. Representing neighboring vertices, Represents vertices and The edge between; It is an adjacency matrix. Representing Euclidean space, Representing the adjacency matrix The Line number Column elements, and satisfying and , Representing the adjacency matrix The Line number Column elements;

[0216] Define the communication network topology between escapee drones as an undirected graph. ,in, and They are the vertex set and the undirected edge set, respectively. Indicating the first as an escapee Each drone corresponds to a vertex, and each edge represents a communication link between the escapee drones corresponding to that vertex; Define... As vertex The neighborhood group, Indicates the number of the neighboring vertex. Representing neighboring vertices, Represents vertices and The edge between; It is an adjacency matrix. Representing the adjacency matrix The Line number Column elements, and satisfying and , Adjacency matrix The Line number Column elements;

[0217] Define the communication network topology between the pursuer drone and the escapee drone as an undirected graph. ,in, and These are the vertex set and the undirected edge set, respectively. A hunter drone and The escapee drones are located in two different sectors. and All drones exchange information between two different sectors; for those containing A sector of a hunter drone In other words, if the first The hunter drone was able to capture the first Information about each escapee drone, and the connectivity weights between them. ,on the contrary, For those containing A sector of an escaped drone In other words, if the first The escapee can obtain the first Information about each pursuer, and the connection weights between them. ,on the contrary, ;

[0218] S2: Based on the nonlinear dynamic model of the motion attitude of the pursuer drone and the escaper drone and the communication network topology, construct HSMS from the perspective of the pursuer drone and the escaper drone respectively, and then construct the cost functions of the pursuer drone and the escaper drone.

[0219] Specifically, the following steps are included:

[0220] S2.1: Based on the nonlinear dynamic model of the hunter drone's motion attitude and the communication network topology, construct the HSMS of the hunter drone's angles, and further design the control input signals in the nonlinear dynamic model of the hunter drone's motion attitude based on the HSMS of the hunter drone's angles. ;

[0221] From the perspective of the hunter drone, the HSMS is constructed as follows:

[0222] (8);

[0223] in, Indicates from the first The HSMS is constructed from the perspective of a hunter drone, and for The One element, Indicates element number and ; Defined as , for The One element, and ,in, For the system state variables of the hunter drone corresponding to the neighboring vertex; for The One element, For time variables, For integration variables, express The The elements are in the integral variable. The value at time, It is a positive definite matrix; Let be a diagonal matrix, where, and All are positive constants; Defined as and According to the iterative algorithm, we can obtain and Further, we can obtain , , For number index, for The One element, If it is a constant, ,but , for The One element, and vice versa. ;

[0224] From formula (8), we can see that the first... The HSMS of a single hunter drone can be re-expressed as:

[0225] (9);

[0226] in, It is a lower triangular matrix and ;

[0227] Based on formulas (2) and (9), the dynamic calculation of the HSMS of the hunter drone angle is as follows:

[0228] (10);

[0229] in, for The first derivative, , , For the control input signal of the hunter drone corresponding to the neighboring vertex;

[0230] According to formulas (9) and (10), the first The hunter drone in its first Control quantities on layer HSMS , and This can be expressed as:

[0231] (11);

[0232] in, , , They represent the first The hunter drone in its first Control parameters for roll angle, pitch angle and yaw angle on the layer HSMS; , , They represent the first The hunter drone in its first The control parameters for roll angle, pitch angle, and yaw angle on the layer HSMS, and they satisfy... ; , , They represent the first The hunter drone in its first The equivalent control values ​​for roll angle, pitch angle, and yaw angle on the layer HSMS, and they satisfy... ; , , They represent the first The hunter drone in its first The switching control values ​​for roll angle, pitch angle, and yaw angle on the layer HSMS, and satisfying ;

[0233] According to the iterative algorithm, formula (11) can be rewritten as:

[0234] (12);

[0235] in, Indicates the index number;

[0236] Based on formula (12), we can further obtain the result in formula (2). for:

[0237] (13);

[0238] in, ; ;

[0239] S2.2: Based on the nonlinear dynamic model of the escaped UAV's motion attitude and the communication network topology, construct the HSMS of the escaped UAV's angles, and further design the control input signals in the nonlinear dynamic model of the escaped UAV's motion attitude based on the HSMS of the escaped UAV's angles. ;

[0240] Constructing HSMS from the perspective of escaped drones:

[0241] (14);

[0242] in, Indicates from the first HSMS constructed from the perspective of an escaped drone. for The One element, ; Defined as , for The One element, and , for The One element, express The The elements are in the integral variable. The value at time, It is a positive definite matrix; It is a lower triangular matrix and , For the system state variables of the escaped drone corresponding to the neighboring vertex;

[0243] definition It is a diagonal matrix. and All are constants; defined and According to the iterative algorithm, we can obtain and Further, we can obtain , If it is a constant, ,but , for The One element, and vice versa. ;

[0244] Based on formulas (7) and (14), the dynamic calculation of the HSMS of the escaped UAV angle is as follows:

[0245] (15);

[0246] in, for The first derivative, , , For the control input signal of the escapee drone corresponding to the neighboring vertex;

[0247] According to formulas (14) and (15), under the Bouc-Wen hysteresis effect, the first The escape drone in its first Control quantities on layer HSMS , and This can be expressed as:

[0248] (16);

[0249] in, , and These represent the effects of the Bouc-Wen hysteresis, specifically the... The escape drone in its first Control parameters for roll angle, pitch angle and yaw angle on the layer HSMS; , and These represent the effects of the Bouc-Wen hysteresis, specifically the... The escape drone in its first The control parameters for roll angle, pitch angle, and yaw angle on the layer HSMS, and they satisfy... ; , and These represent the effects of the Bouc-Wen hysteresis, specifically the... The escape drone in its first The equivalent control values ​​for roll angle, pitch angle, and yaw angle on the layer HSMS, and they satisfy... ; , and These represent the effects of the Bouc-Wen hysteresis, specifically the... The escape drone in its first The switching control variables for roll angle, pitch angle, and yaw angle on the layer HSMS, and they satisfy... ;

[0250] According to the iterative algorithm, formula (16) can be rewritten as:

[0251] (17);

[0252] Based on formula (17), we can further obtain the result in formula (7). for:

[0253] (18);

[0254] in, ; ;

[0255] S2.3: Based on the control input signal in the nonlinear dynamic model of the hunter drone's motion attitude. Control input signals in the nonlinear dynamics model of the motion attitude of the escape drone Cost functions for the pursuer drone and the escaper drone are established respectively to characterize their attitude responses.

[0256] The attitude response of the Chaser drone can be characterized by the following cost function:

[0257] (19);

[0258] in, Indicates the first The cost function of a hunter drone; Defined as ;

[0259] (20);

[0260] in, It is a positive constant; , and All are positive definite matrices; ,in, It is a non-quadratic function. It is a constant. For integration variables;

[0261] The attitude response of the escapee drone can be characterized by the following cost function:

[0262] (twenty one);

[0263] in, Indicates the first The cost function of an escapee drone; Defined as ;

[0264] (twenty two);

[0265] in, It is a positive constant; , and All are positive definite matrices; ,and It is a constant. For integration variables, It is a non-quadratic function;

[0266] S3: Based on the HSMS and cost function of the pursuer drone and the escapee drone, establish a pursuit-escape game mechanism with multiple participants.

[0267] Specifically: Define the first The optimal equivalent control law and optimal switching control law for the hunter drone are as follows: and And define the first The optimal equivalent control law and optimal switching control law for the escaped drone under the Bouc-Wen hysteresis effect are as follows: and If the following conditions are met:

[0268] (twenty three);

[0269] (twenty four);

[0270] but , , and The Nash equilibrium constituting the pursuit-escape game, in which, For the first Besides the individual hunter drone, the set of optimal control laws for other hunter drones. For the first The set of optimal control laws for other escaped drones under the Bouc-Wen hysteresis effect, excluding the escaped drone;

[0271] S4: Based on the established pursuit-escape game mechanism and using adaptive dynamic programming, design a DET optimal control law for both the pursuer drone and the escape drone to achieve control of the pursuer drone and the escape drone.

[0272] The specific steps are as follows:

[0273] S4.1: Based on the established pursuit-escape game mechanism and utilizing adaptive dynamic programming, design the first... Optimal equivalent control law for a hunter drone and optimized switching control law ;

[0274] First, based on the HSMS of the Chaser drone, the first... The Hamiltonian function for the hunter drone is:

[0275] (25);

[0276] in, For the first Hamiltonian functions for a hunter drone for The gradient;

[0277] Combining formulas (10) and (25), we obtain the first... The Hamilton-Jacobi-Bellman equations for the hunter drone are as follows:

[0278] (26);

[0279] in, For the first Hamilton-Jacobi-Bellman equations for a single hunter drone; , Let's define the optimized equivalent control law for the hunter drone corresponding to the neighboring vertex. An optimized switching control law for the hunter drones corresponding to neighboring vertices;

[0280] Based on formula (26) and by solving and

[0281] ,get and The specific form is as follows:

[0282] (27);

[0283] (28);

[0284] S4.2: Based on the established pursuit-escape game mechanism and utilizing adaptive dynamic programming, design the first... The optimal equivalent control law of an escaped drone under the Bouc-Wen hysteresis effect and optimized switching control law ;

[0285] Based on the escapee drone, the HSMS defines the first The Hamiltonian function for an escapee drone is:

[0286] (29);

[0287] in, For the first Hamiltonian function of an escaped drone for The gradient;

[0288] According to formulas (15) and (29), we obtain the first... The Hamilton-Jacobi-Bellman equations for an escapee drone are as follows:

[0289] (30);

[0290] in, For the first Hamilton-Jacobi-Bellman equations for an escapee drone; ;

[0291] Based on formula (30) and by solving and

[0292] ,get and The specific form is as follows:

[0293] (31);

[0294] (32);

[0295] S4.3: Design the difference function between the trigger state variable and the system state variable of the pursuer drone, and the difference function between the trigger state variable and the system state variable of the escapee drone;

[0296] To reduce the communication load between UAVs, based on formulas (27), (28), (31) and (32), a difference function is further designed to determine whether the control law is maintained or updated, thereby effectively avoiding unnecessary redundant operations.

[0297] definition and The first Each triggering moment and its corresponding time sequence For the moment of triggering the number and , Represents the set of natural numbers; satisfy , For the first A triggering moment, and when hour Then, the first The pursuer drone and the first The escape drone in the first The trigger state variables at each trigger instant are respectively represented as follows: and ;No. The pursuer drone and the first The time series of the triggering state variables of each escape drone are denoted as follows: and ;

[0298] when At that time, the first The difference function between the trigger state variable and the system state variable of the individual hunter drone is:

[0299] (33);

[0300] in, For the first The difference function between the trigger state variables and the system state variables of a hunter drone;

[0301] when At that time, the first The difference function between the trigger state variable and the system state variable of the escaped drone is:

[0302] (34);

[0303] in, For the first The difference function between the trigger state variables and the system state variables of an escapee drone;

[0304] S4.4: Design Section The DET-optimized control law of a hunter drone, including the DET-optimized equivalent control law and the DET-optimized switching control law;

[0305] According to formulas (27), (28), and (33), we can obtain:

[0306] (35);

[0307] (36);

[0308] in, and These respectively represent the state variables that are triggered. The next DET-optimized equivalent control law and DET-optimized switching control law for a single hunter drone; ; Indicates the first The first trigger moment HSMS of a hunter drone; Indicates that the state variable is triggered Below, except for the first Optimized control input signals for all other Hunter drones besides the Hunter drone; for The gradient; Indicates that the state variable is triggered The next Control gain of the hunter drone;

[0309] S4.5: Design No. The DET optimal control law of an escaped UAV under the Bouc-Wen hysteresis effect, including the DET optimal equivalent control law and the DET optimal switching control law;

[0310] According to formulas (31), (32), and (34), the first... An escapee drone under the Bouc-Wen hysteresis effect and They are designed as follows:

[0311] (37);

[0312] (38);

[0313] in, and These respectively represent the state variables that are triggered. The next DET-optimized equivalent control law and DET-optimized switching control law for an escapee drone; ; for The gradient; Indicates that the state variable is triggered The next Control gain of the escapee drone; To indicate the first The first trigger moment HSMS of an escapee drone Indicates that the state variable is triggered Below, except for the first Optimized control input signals for all escape drones except for the escape drone;

[0314] S4.6: Based on the difference function between the trigger state variables and system state variables of the pursuer drone, the difference function between the trigger state variables and system state variables of the escaper drone, and the DET optimization control law of the pursuer drone and the escaper drone, the control of the pursuer drone and the escaper drone is realized.

[0315] Specifically:

[0316] When the difference function between the trigger state variable and the system state variable of the Chaser UAV meets the preset trigger condition, the DET optimization control law of the Chaser UAV is used to control the Chaser UAV; otherwise, the current control is maintained.

[0317] When the difference function between the escape drone's trigger state variable and the system state variable meets the preset trigger condition, the escape drone's DET optimized control law is used to control the escape drone; otherwise, the current control is maintained.

[0318] S5: Construct a judgment network to estimate the DET optimal control law of the pursuer drone and the escaper drone online, and then update the DET optimal control law;

[0319] The specific steps include:

[0320] Approximation is achieved by implementing a judging neural network. and ,get:

[0321] (39);

[0322] (40);

[0323] in, , , The first The evaluation neural network weights, activation function, and approximation error of a single hunter drone; All Elements in; All Elements in; The number of neural nodes; , , The first The evaluation neural network weights, activation function, and approximation error of an escaped drone; All Elements in; All Elements in;

[0324] Based on formulas (39) and (40), we can further obtain:

[0325] (41);

[0326] (42);

[0327] in, , and They are respectively , and The gradient; , and They are respectively , and The gradient;

[0328] According to formulas (35), (36), and (41), we can obtain:

[0329] (43);

[0330] (44);

[0331] in, , and They are respectively , and The estimate;

[0332] According to formulas (37), (38), and (42), we can obtain:

[0333] (45);

[0334] (46);

[0335] in, , and They are respectively , and The estimate;

[0336] According to formula (25), we can obtain the first... The Bellman residuals for each hunter drone are:

[0337] (47);

[0338] in, Indicates the first Bellman residuals of a hunter drone for The estimate; Indicates that the state variable is triggered Below, except for the first Besides the individual hunter drone, the optimized control input signals for all other hunter drones, and for The estimate;

[0339] According to formula (29), we can obtain the first... The Bellman residuals for each escapee drone are:

[0340] (48);

[0341] in, Indicates the first Bellman residuals of an escapee drone for The estimate; Indicates that the state variable is triggered Below, except for the first Besides the escapee drone, the optimized control input signals for all other escapee drones, and for The estimate;

[0342] Based on formula (47) and by using gradient descent and standardization principles, the first... The evaluation network weight update law for the escaped drone: For the first... The hunter drone and the first Each escapee drone is defined as having an experienced recovery status. and Their corresponding time series are respectively and ,in, For the first At the moment of triggering, , , Represents the set of natural numbers;

[0343] Design No. The evaluation network weight update law for each hunter drone is as follows:

[0344] (49);

[0345] in, Indicates the first The evaluation network weight update law for individual hunter drones; and For update rate; and Used for standardization;

[0346] (50);

[0347] in, Indicates the state of experience recycling. The next The DET-optimized equivalent control law for a hunter drone. for The estimate; Indicates the state of experience recycling. The next DET-optimized switching control law for a single hunter drone for The estimate; Indicates the state of experience recycling. Below, except for the first The set consisting of the estimates of the DET optimal control law of all the other hunter drones besides the hunter drone. and They represent the first time. The first trigger moment HSMS and state error of a single hunter drone;

[0348] (51);

[0349] in, Indicates the state of being triggered. The next Control gain of the hunter drone; Indicates the first At the moment of triggering, and the first The adjacent hunter drone The experience recovery status of the hunter drone; Indicates the state of being triggered. Below, with the The adjacent hunter drone Control gain of the hunter drone; Indicates the state of being triggered. Below, with the The adjacent hunter drone The DET-optimized equivalent control law for a hunter drone, and for The estimate; Indicates the state of being triggered. Below, with the The adjacent hunter drone The DET-optimized switching control law for a hunter drone, and for The estimate; Indicates the state of being triggered. The next Control gain of the escapee drone; Indicates the state of experience recycling. The next The DET-optimized equivalent control law for an escapee drone. for The estimate; Indicates the state of experience recycling. The next DET-optimized switching control law for an escapee drone for The estimate;

[0350] (52);

[0351] in, Indicates the state of experience recycling. The next The DET-optimized equivalent control law for an escapee drone. for The estimate; Indicates the state of experience recycling. The next DET-optimized switching control law for an escapee drone for The estimate;

[0352] Based on formula (48) and by using gradient descent and standardization principles, the first... The evaluation network weight update law for each escapee drone is as follows:

[0353] (53);

[0354] in, Indicates the first The evaluation network weight update law for individual escapee drones; and For update rate; and Used for standardization;

[0355] (54);

[0356] in, Indicates the state of experience recycling. The next The DET-optimized equivalent control law for an escapee drone. for The estimate; Indicates the state of experience recycling. The next DET-optimized switching control law for an escapee drone for The estimate; Indicates the state of experience recycling. Below, except for the first The set of DET-optimized control laws for all escape drones except for the escape drone; and They represent the first time. The first trigger moment HSMS and state error of an escaped drone;

[0357] (55);

[0358] in, Indicates the first At the moment of triggering, and the first The escaped drone adjacent to the first The experience recovery status of the escaped drone; Indicates the state of being triggered. Below, with the The escaped drone adjacent to the first Control gain of the escapee drone; Indicates the state of being triggered. Below, with the The escaped drone adjacent to the first The DET-optimized equivalent control law for an escapee drone, and for The estimate; Indicates the state of being triggered. Below, with the The escaped drone adjacent to the first The DET-optimized switching control law for an escapee drone, and for The estimate;

[0359] (56);

[0360] in, and They represent the first time. The first trigger moment HSMS and state error of an escaped drone; Indicates the first At the moment of triggering, and the first The escaped drone adjacent to the first The experience recovery status of the escaped drone; Indicates the state of being triggered. Below, with the The escaped drone adjacent to the first Control gain of the escapee drone; Indicates the state of being triggered. Below, with the The escaped drone adjacent to the first The DET-optimized equivalent control law for an escapee drone, and for The estimate; Indicates the state of being triggered. Below, with the The escaped drone adjacent to the first The DET-optimized switching control law for an escapee drone, and for The estimate; Indicates the state of being triggered. The next Control gain of the escapee drone; Indicates the state of experience recycling. The next The DET-optimized equivalent control law for a hunter drone. for The estimate; Indicates the state of experience recycling. The next DET-optimized switching control law for a single hunter drone for The estimate.

[0361] To verify the effectiveness of the control method of this invention, two pursuer drones and two escapee drones were selected for simulation verification in the embodiment. The undirected graph topology relationship between them is as follows: Figure 2 As shown, the following adjacency matrix can be obtained: ;

[0362] Select the first The HSMS of the Hunter drone is , The Chaser drone orbiting its body coordinate system axis, shaft and The moments of inertia in the axial direction are respectively , and The moment of inertia of the motor propellers of the Chaser drone is... Additional perturbations from the pursuer drone are and The Chaser drone in the body coordinate system axis, shaft and The drag coefficients in the axial direction are respectively , and Select the first The HSMS of the escapee drone is , The escapee drone's coordinate system around its body. axis, shaft and The moments of inertia in the axial directions are respectively , and The moment of inertia of the electric motor propellers of the escape drone is... Additional perturbations for the escapee drone are and The escapee drone in the body coordinate system axis, shaft and The drag coefficients in the axial direction are respectively , and .

[0363] Simulation results are as follows Figures 3-7 As shown. Figure 3 The figure shows the HSMS response curves of the pursuer drone and the escapee drone in the simulation of the embodiment of the present invention. It can be seen that they oscillate greatly at the beginning and then tend to stabilize. Figure 4 The graph shows the DET optimized control law response curves of the first pursuer UAV in the simulation of the embodiment of the present invention for its roll angle, pitch angle and yaw angle. Figure 5 The graph shows the DET optimized control law response curves of the second pursuer UAV in the simulation of the embodiment of the present invention for its roll angle, pitch angle and yaw angle. Figure 6 The graph shows the DET optimized control law response curves of the first escapee UAV in the simulation of the embodiment of the present invention for its roll angle, pitch angle and yaw angle. Figure 7 This is a graph showing the DET-optimized control law response curves of the second escapee UAV in the simulation of this embodiment of the invention, considering its roll angle, pitch angle, and yaw angle. From... Figures 4-7 As can be seen, the DET optimization control law will converge to the small neighborhood of zero in a finite time.

[0364] Example 4:

[0365] This embodiment proposes an electronic device, including: one or more processors, and a memory, wherein the memory is used to store instructions, and when the instructions are executed by the one or more processors, the one or more processors cause the one or more processors to execute the UAV DET optimization control method based on HSMS under the pursuit and escape game.

[0366] The electronic device can be a mobile phone, computer, or tablet computer, etc., and includes a memory and a processor. The memory stores a computer program, which, when executed by the processor, implements the HSMS-based UAV DET optimization control method under the pursuit-escape game described in the embodiment. It is understood that the electronic device may also include input / output (I / O) interfaces and communication components.

[0367] The processor is used to execute all or part of the steps in the HSMS-based UAV DET optimization control method under the pursuit-escape game described in the above embodiments. The memory is used to store various types of data, which may include, for example, instructions for any application or method in the electronic device, as well as application-related data.

[0368] The processor can be implemented as an Application Specific Integrated Circuit (ASIC), Digital Signal Processor (DSP), Programmable Logic Device (PLD), Field Programmable Gate Array (FPGA), controller, microcontroller, microprocessor, or other electronic components, and is used to execute the UAV DET optimization control method based on HSMS under the pursuit and escape game described in the above embodiments.

[0369] Example 5:

[0370] This embodiment proposes a computer-readable storage medium that stores executable instructions. When these instructions are executed, if they are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium.

[0371] The computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the HSMS-based UAV DET optimization control method under the pursuit and escape game described in the various embodiments of this application.

[0372] The aforementioned storage media include: flash memory, hard disk, multimedia card, card-type memory (e.g., SD (Secure Digital Memory Card) or DX (Memory Data Register, MDR) memory, random access memory (RAM), static random-access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic storage, disk, optical disk, server, APP (Application) application store, and other media capable of storing program verification codes. These media store computer programs, which, when executed by a processor, can implement the various steps of the HSMS-based UAV DET optimization control method under the pursuit-escape game described above.

[0373] Example 6:

[0374] This embodiment proposes a computer program product, including a computer program or instructions, which, when executed by a processor, implements the aforementioned UAV DET optimization control method based on HSMS under the pursuit-escape game.

[0375] Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or part of the technical solution, can be embodied in the form of a computer program product.

[0376] The various embodiments in this application are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0377] The scope of protection of this application is not limited to the embodiments described above. Obviously, those skilled in the art can make various modifications and variations to this disclosure without departing from the scope and spirit of this disclosure. If such modifications and variations fall within the scope of this disclosure and its equivalents, then the intent of this disclosure also includes these modifications and variations.

Claims

1. A DET optimization control method for unmanned aerial vehicles based on HSMS under a pursuit-escape game, characterized in that, Includes the following steps: Establish a nonlinear dynamic model of the motion attitude of the pursuer drone and the escaper drone, and define the communication network topology between the pursuer drone and the escaper drone. Based on the nonlinear dynamic model of the motion attitude of the pursuer drone and the escaper drone and the topology of the communication network, HSMS is constructed from the perspectives of the pursuer drone and the escaper drone respectively, and then the cost functions of the pursuer drone and the escaper drone are constructed. Based on the HSMS and cost function of the pursuer drone and the escapee drone, a pursuit-escape game mechanism with multiple participants is established. Based on the established pursuit-escape game mechanism and using adaptive dynamic programming, a DET optimal control law is designed for both the pursuer drone and the escape drone to achieve control of the pursuer drone and the escape drone. A judgment network is constructed to estimate the DET optimal control law of the pursuer drone and the escaper drone online, and then update the DET optimal control law.

2. The UAV DET optimization control method based on HSMS under the pursuit-escape game as described in claim 1, characterized in that, The establishment of a nonlinear dynamic model of the motion attitude of the pursuer and the escaper UAVs, and the definition of the communication network topology between the pursuer and the escaper UAVs, are as follows: A1: Regarding the pursuer... The motion attitude of the drone was modeled to obtain a nonlinear dynamic model of the motion attitude of the pursuer drone. The nonlinear dynamic model of the motion attitude of the pursuer drone is established as follows: (1)； in, The drone is used to represent the pursuer. , , They represent the first as the pursuers. A drone orbiting the body coordinate system axis, shaft and Moment of inertia in the axial direction, This indicates the serial number of the drone acting as the pursuer. The coordinate system is based on the drone's center of mass as the origin, and extends along the longitudinal axis of the drone towards the nose. The axis, along the horizontal axis of the fuselage, points towards the right wing of the UAV. Axis, perpendicular to The plane is downward. axis; , , They represent the first as the pursuers. The roll angle, pitch angle, and yaw angle of each drone; , , They represent , and The first derivative; , , They represent , and The second derivative; Indicating the role of the pursuer The moment of inertia of the motor blades of a drone. Indicating the role of the pursuer Additional perturbations from the drone; , , Indicating the role of the pursuer A drone in the body coordinate system axis, shaft and Axial drag coefficient; , , They represent the first as the pursuers. The control parameters for the roll, pitch, and yaw angles of a single UAV; (collection) Defined as , Indicates the number of drones acting as pursuers; Defined as the pursuer The system state variables of a drone And let its state components , , , , , , If we denote the transpose, then the nonlinear dynamic model of the motion attitude of the pursuer UAV is further expressed by formula (1) as: (2)； in, express The first derivative, and , , ; ; To control the input signal; A2: Regarding the first as an escapee The motion attitude of the drone is modeled to obtain a nonlinear dynamic model of the motion attitude of the escapee drone. The nonlinear dynamic model of the escapee drone's motion attitude is constructed as follows: (3)； in, This refers to the drone acting as an escapee; , , These represent the first as an escapee A drone orbiting the body coordinate system axis, shaft and Moment of inertia in the axial direction, The serial number indicating the drone that escaped; , , These represent the first as an escapee The roll angle, pitch angle, and yaw angle of each drone; , , They represent , and The first derivative; , , They represent , and The second derivative; Indicating the first as an escapee The moment of inertia of the motor blades of a drone. Indicating the first as an escapee Additional perturbations from the drone; , , Indicating the first as an escapee A drone in the body coordinate system axis, shaft and Axial drag coefficient; , , These represent the first as an escapee The control parameters for the roll, pitch, and yaw angles of a single UAV; (collection) Defined as , This indicates the number of drones that escaped. Defined as the escapee The system state variables of a drone And let its state components , , , , , Then, as the escapee's first The nonlinear dynamic model of the motion attitude of the UAV can be re-expressed by formula (3): (4)； in, express The first derivative, and , , ; ; To control the input signal; Considering the escapee Control input signal of a drone Due to the Bouc-Wen hysteresis effect, the control input signal... The mathematical model is as follows: (5)； in, and These are the input and output of the mathematical model, respectively. , and They are respectively the first as escapees The input control quantities of a UAV for roll angle, pitch angle and yaw angle under Bouc-Wen hysteresis; , and They are respectively the first as escapees The output of a UAV in terms of roll angle, pitch angle and yaw angle under the Bouc-Wen hysteresis effect; , ,and , , , , and All are constants, and they satisfy , and ; ,and , and All are auxiliary variables; According to formula (5), we can obtain The time derivative is: (6)； in, and They are respectively and The first derivative; ,and , and The shape stiffness proportionality coefficient of the hysteresis loop representing the Bouc-Wen hysteresis effect satisfies , and ; , and This represents the magnitude of the hysteresis loop; additionally, auxiliary variables... , and Each satisfies , and ; , and The smoothness index represents the hysteresis loop; , and They are defined as , and ,and , and They are respectively , and The first derivative; Based on formulas (4), (5), and (6), the first as the escapee The nonlinear dynamic model of the motion attitude of the UAV is further expressed as follows: (7)； A3: Define the communication network topology between the pursuer drone and the escapee drone; Specifically, the communication network topology between the pursuer drones is defined as an undirected graph. ,in, and They are the vertex set and the undirected edge set, respectively. Indicating the role of the pursuer Each drone corresponds to a vertex, and each edge represents a communication link between the corresponding pursuer drones; define... As vertices The neighborhood group, Indicates the number of the neighboring vertex. Representing neighboring vertices, Represents vertices and The edge between; It is an adjacency matrix. Representing Euclidean space, Representing the adjacency matrix The Line number Column elements, and satisfying and , Representing the adjacency matrix The Line number Column elements; Define the communication network topology between escapee drones as an undirected graph. ,in, and They are the vertex set and the undirected edge set, respectively. Indicating the first as an escapee Each drone corresponds to a vertex, and each edge represents a communication link between the escapee drones corresponding to that vertex; Define... As vertices The neighborhood group, Indicates the number of the neighboring vertex. Representing neighboring vertices, Represents vertices and The edge between; It is an adjacency matrix. Representing the adjacency matrix The Line number Column elements, and satisfying and , Adjacency matrix The Line number Column elements; Define the communication network topology between the pursuer drone and the escapee drone as an undirected graph. ,in, and These are the vertex set and the undirected edge set, respectively. A hunter drone and The escapee drones are located in two different sectors. and All drones exchange information between two different sectors; for those containing A sector of a hunter drone In other words, if the first The hunter drone was able to capture the first Information about each escapee drone, and the connectivity weights between them. ,on the contrary, For those containing A sector of an escaped drone In other words, if the first The escapee can obtain the first Information about each pursuer, and the connection weights between them. ,on the contrary, .

3. The UAV DET optimization control method based on HSMS under the pursuit-escape game as described in claim 2, characterized in that, Based on the nonlinear dynamic model of the motion attitude of the pursuer UAV and the escaper UAV and the communication network topology, the HSMS is constructed from the perspectives of the pursuer UAV and the escaper UAV respectively, and then the cost functions of the pursuer UAV and the escaper UAV are constructed, specifically including: B1: Based on the nonlinear dynamic model of the hunter drone's motion attitude and the communication network topology, construct the HSMS of the hunter drone's angles, and further design the control input signals in the nonlinear dynamic model of the hunter drone's motion attitude based on the HSMS of the hunter drone's angles. ; From the perspective of the hunter drone, the HSMS is constructed as follows: (8)； in, Indicates from the first The HSMS is constructed from the perspective of a hunter drone, and for The One element, Indicates element number and ; Defined as , for The One element, , ,in, For the system state variables of the hunter drone corresponding to the neighboring vertex; for The One element, For time variables, For integration variables, express The The elements are in the integral variable as The value at time, It is a positive definite matrix; Let be a diagonal matrix, where, and All are positive constants; Defined as and According to the iterative algorithm, we get and Further , , For number index, for The One element, If it is a constant, ,but , for The One element, and vice versa. ; From formula (8), we can see that the first... The HSMS of the individual hunter drone is re-expressed as follows: (9)； in, It is a lower triangular matrix and ; Based on formulas (2) and (9), the dynamic calculation of the HSMS of the hunter drone angle is as follows: (10)； in, for The first derivative, , , For the control input signal of the hunter drone corresponding to the neighboring vertex; According to formulas (9) and (10), the first The hunter drone in its first Control quantities on layer HSMS , and This can be expressed as: (11)； in, , , They represent the first The hunter drone in its first Control parameters for roll angle, pitch angle and yaw angle on the layer HSMS; , , They represent the first The hunter drone in its first The control parameters for roll angle, pitch angle, and yaw angle on the layer HSMS, and they satisfy... ; , , They represent the first The hunter drone in its first The equivalent control values ​​for roll angle, pitch angle, and yaw angle on the layer HSMS, and they satisfy... ; , , They represent the first The hunter drone in its first The switching control values ​​for roll angle, pitch angle, and yaw angle on the layer HSMS, and satisfying ; According to the iterative algorithm, formula (11) can be rewritten as: (12)； in, Indicates the index number; Based on formula (12), we can further obtain the result in formula (2). for: (13)； in, ; ; B2: Based on the nonlinear dynamic model of the escaped UAV's motion attitude and the communication network topology, construct the HSMS of the escaped UAV's angles, and further design the control input signals in the nonlinear dynamic model of the escaped UAV's motion attitude based on the HSMS of the escaped UAV's angles. ; Constructing HSMS from the perspective of escaped drones: (14)； in, Indicates from the first HSMS constructed from the perspective of an escaped drone. for The One element, ; Defined as , for The One element, , , for The One element, express The The elements are in the integral variable as The value at time, It is a positive definite matrix; It is a lower triangular matrix and , For the system state variables of the escaped drone corresponding to the neighboring vertex; definition It is a diagonal matrix. and All are constants; defined and According to the iterative algorithm, we can obtain and Further, we can obtain , If it is a constant, ,but , for The One element, and vice versa. ; Based on formulas (7) and (14), the dynamic calculation of the HSMS of the escaped UAV angle is as follows: (15)； in, for The first derivative, , , For the control input signal of the escapee drone corresponding to the neighboring vertex; According to formulas (14) and (15), under the Bouc-Wen hysteresis effect, the first The escape drone in its first Control quantities on layer HSMS , and This can be expressed as: (16)； in, , and These represent the effects of the Bouc-Wen hysteresis, specifically the... The escape drone in its first Control parameters for roll angle, pitch angle and yaw angle on the layer HSMS; , and These represent the effects of the Bouc-Wen hysteresis, specifically the... The escape drone in its first The control parameters for roll angle, pitch angle, and yaw angle on the layer HSMS, and they satisfy... ; , and These represent the effects of the Bouc-Wen hysteresis, specifically the... The escape drone in its first The equivalent control values ​​for roll angle, pitch angle, and yaw angle on the layer HSMS, and they satisfy... ; , and These represent the effects of the Bouc-Wen hysteresis, specifically the... The escape drone in its first The switching control variables for roll angle, pitch angle, and yaw angle on the layer HSMS, and they satisfy... ; According to the iterative algorithm, formula (16) can be rewritten as: (17)； Based on formula (17), we can further obtain the result in formula (7). for: (18)； in, ; ; B3: Based on the control input signal in the nonlinear dynamics model of the hunter drone's motion attitude. Control input signals in the nonlinear dynamics model of the motion attitude of the escape drone Establish cost functions for the pursuer drone and the escapee drone respectively; The attitude response of the Chaser drone can be characterized by the following cost function: (19)； in, Indicates the first The cost function of a hunter drone; Defined as ; (20)； in, It is a positive constant; , and All are positive definite matrices; ,in, It is a non-quadratic function. It is a constant. For integration variables; The attitude response of the escapee drone can be characterized by the following cost function: (21)； in, Indicates the first The cost function of an escapee drone; Defined as ; (22)； in, It is a positive constant; , and All are positive definite matrices; ,and It is a constant. For integration variables, It is a non-quadratic function.

4. The UAV DET optimization control method based on HSMS under the pursuit-escape game as described in claim 3, characterized in that, The process involves establishing a multi-participant game mechanism for pursuit and escape based on the HSMS and cost functions of the pursuer and escape drones. Specifically, the mechanism is defined as follows: [Define the first...] The optimal equivalent control law and optimal switching control law for the hunter drone are as follows: and And define the first The optimal equivalent control law and optimal switching control law for the escaped drone under the Bouc-Wen hysteresis effect are as follows: and If the following conditions are met: (23)； (24)； but , , and The Nash equilibrium constituting the pursuit-escape game, in which, For the first Besides the individual hunter drone, the set of optimal control laws for other hunter drones. For the first The set of optimal control laws for other escaped drones under the Bouc-Wen hysteresis effect, excluding the escaped drone.

5. The UAV DET optimization control method based on HSMS under the pursuit-escape game as described in claim 4, characterized in that, Based on the established pursuit-escape game mechanism and utilizing adaptive dynamic programming, a DET-optimized control law is designed for both the pursuer and escape drones to achieve control of them. The specific steps are as follows: C1: Based on the established pursuit-escape game mechanism and utilizing adaptive dynamic programming, design the first... Optimal equivalent control law for a hunter drone and optimized switching control law ; First, based on the HSMS of the Chaser drone, the first... The Hamiltonian function for the hunter drone is: (25)； in, For the first Hamiltonian functions for a hunter drone for The gradient; Combining formulas (10) and (25), we obtain the first... The Hamilton-Jacobi-Bellman equations for the hunter drone are as follows: (26)； in, For the first Hamilton-Jacobi-Bellman equations for a single hunter drone; , Let's define the optimized equivalent control law for the hunter drone corresponding to the neighboring vertex. An optimized switching control law for the hunter drones corresponding to neighboring vertices; Based on formula (26) and by solving and ,get and The specific form is as follows: (27)； (28)； C2: Based on the established pursuit-escape game mechanism and utilizing adaptive dynamic programming, design the first... The optimal equivalent control law of an escaped drone under the Bouc-Wen hysteresis effect and optimized switching control law ; Based on the escapee drone, the HSMS defines the first The Hamiltonian function for an escapee drone is: (29)； in, For the first Hamiltonian function of an escaped drone for The gradient; According to formulas (15) and (29), we obtain the first... The Hamilton-Jacobi-Bellman equations for an escapee drone are as follows: (30)； in, For the first Hamilton-Jacobi-Bellman equations for an escapee drone; ; Based on formula (30) and by solving and ,get and The specific form is as follows: (31)； (32)； C3: Design the difference function between the trigger state variable and the system state variable of the pursuer drone, and the difference function between the trigger state variable and the system state variable of the escapee drone; definition and The first Each triggering moment and its corresponding time sequence For the moment of triggering the number and , Represents the set of natural numbers; satisfy , For the first A triggering moment, and when hour Then, the first The pursuer drone and the first The escape drone in the first The trigger state variables at each trigger instant are respectively represented as follows: and ;No. The pursuer drone and the first The time series of the triggering state variables of each escape drone are denoted as follows: and ; when At that time, the first The difference function between the trigger state variable and the system state variable of the individual hunter drone is: (33)； in, For the first The difference function between the trigger state variables and the system state variables of a single hunter drone; when At that time, the first The difference function between the trigger state variable and the system state variable of the escaped drone is: (34)； in, For the first The difference function between the trigger state variables and the system state variables of an escapee drone; C4: Design No. The DET-optimized control law of a hunter drone, including the DET-optimized equivalent control law and the DET-optimized switching control law; According to formulas (27), (28), and (33), we can obtain: (35)； (36)； in, and These respectively represent the state variables that are triggered. The next DET-optimized equivalent control law and DET-optimized switching control law for a single hunter drone; ; Indicates the first The first trigger moment HSMS of a hunter drone; Indicates that the state variable is triggered Below, except for the first Optimized control input signals for all other Hunter drones besides the Hunter drone; for The gradient; Indicates that the state variable is triggered The next Control gain of the hunter drone; C5: Design No. The DET optimal control law of an escaped UAV under the Bouc-Wen hysteresis effect, including the DET optimal equivalent control law and the DET optimal switching control law; According to formulas (31), (32), and (34), the first... An escapee drone under the Bouc-Wen hysteresis effect and They are designed as follows: (37)； (38)； in, and These respectively represent the state variables that are triggered. The next DET-optimized equivalent control law and DET-optimized switching control law for an escapee drone; ; for The gradient; Indicates that the state variable is triggered The next Control gain of the escapee drone; To indicate the first The first trigger moment HSMS of an escapee drone Indicates that the state variable is triggered Below, except for the first Optimized control input signals for all escape drones except for the escape drone; C6: Based on the difference function between the trigger state variables and system state variables of the pursuer drone, the difference function between the trigger state variables and system state variables of the escaper drone, and the DET optimization control law of the pursuer drone and the escaper drone, the control of the pursuer drone and the escaper drone is realized. Specifically: When the difference function between the trigger state variable of the Chaser UAV and the system state variable meets the preset trigger condition, the DET optimization control law of the Chaser UAV is used to control the Chaser UAV; otherwise, the current control is maintained. When the difference function between the escaped drone's trigger state variable and the system state variable meets the preset trigger condition, the escaped drone's DET optimized control law is used to control the escaped drone; otherwise, the current control is maintained.

6. The UAV DET optimization control method based on HSMS under the pursuit-escape game as described in claim 5, characterized in that, The construction of the evaluation network performs online estimation of the DET optimal control law for both the pursuer and the escapee drones, and then updates the DET optimal control law. Specific steps include: Approximation is achieved by implementing a judging neural network. and ,get: (39)； (40)； in, , , The first The evaluation neural network weights, activation function, and approximation error of a single hunter drone; All Elements in; All Elements in; The number of neural nodes; , , The first The evaluation neural network weights, activation function, and approximation error of an escaped drone; All Elements in; All Elements in; Based on formulas (39) and (40), we can further obtain: (41)； (42)； in, , and They are respectively , and The gradient; , and They are respectively , and The gradient; According to formulas (35), (36), and (41), we can obtain: (43)； (44)； in, , and They are respectively , and The estimate; According to formulas (37), (38), and (42), we can obtain: (45)； (46)； in, , and They are respectively , and The estimate; According to formula (25), we can obtain the first... The Bellman residuals for each hunter drone are: (47)； in, Indicates the first Bellman residuals of a hunter drone for The estimate; Indicates that the state variable is triggered Below, except for the first Besides the individual hunter drone, the optimized control input signals for all other hunter drones, and for The estimate; According to formula (29), we can obtain the first... The Bellman residuals for each escapee drone are: (48)； in, Indicates the first Bellman residuals of an escapee drone for The estimate; Indicates that the state variable is triggered Below, except for the first Besides the escapee drone, the optimized control input signals for all other escapee drones, and for The estimate; Based on formula (47) and by using gradient descent and standardization principles, the first... The evaluation network weight update law for the escaped drone: For the first... The hunter drone and the first Each escapee drone is defined as having an experienced recovery status. and Their corresponding time series are respectively and ,in, For the first At the moment of triggering, , , Represents the set of natural numbers; Design No. The evaluation network weight update law for each hunter drone is as follows: (49)； in, Indicates the first The evaluation network weight update law for individual hunter drones; and For update rate; and Used for standardization; (50)； in, Indicates the state of experience recycling. The next The DET-optimized equivalent control law for a hunter drone. for The estimate; Indicates the state of experience recycling. The next DET-optimized switching control law for a single hunter drone for The estimate; Indicates the state of experience recycling. Below, except for the first The set consisting of the estimates of the DET optimal control law of all the other hunter drones besides the hunter drone. and They represent the first time. The first trigger moment HSMS and state error of a single hunter drone; (51)； in, Indicates the state of being triggered. The next Control gain of the hunter drone; Indicates the first At the moment of triggering, and the first The adjacent hunter drone The experience recovery status of the hunter drone; Indicates the state of being triggered. Below, with the The adjacent hunter drone Control gain of the hunter drone; Indicates the state of being triggered. Below, with the The adjacent hunter drone The DET-optimized equivalent control law for a hunter drone, and for The estimate; Indicates the state of being triggered. Below, with the The adjacent hunter drone The DET-optimized switching control law for a hunter drone, and for The estimate; Indicates the state of being triggered. The next Control gain of the escapee drone; Indicates the state of experience recycling. The next The DET-optimized equivalent control law for an escapee drone. for The estimate; Indicates the state of experience recycling. The next DET-optimized switching control law for an escapee drone for The estimate; (52)； in, Indicates the state of experience recycling. The next The DET-optimized equivalent control law for an escapee drone. for The estimate; Indicates the state of experience recycling. The next DET-optimized switching control law for an escapee drone for The estimate; Based on formula (48) and by using gradient descent and standardization principles, the first... The evaluation network weight update law for each escapee drone is as follows: (53)； in, Indicates the first The evaluation network weight update law for individual escapee drones; and For update rate; and Used for standardization; (54)； in, Indicates the state of experience recycling. The next The DET-optimized equivalent control law for an escapee drone. for The estimate; Indicates the state of experience recycling. The next DET-optimized switching control law for an escapee drone for The estimate; Indicates the state of experience recycling. Below, except for the first The set of DET-optimized control laws for all escape drones except for the escape drone; and They represent the first time. The first trigger moment HSMS and state error of an escaped drone; (55)； in, Indicates the first At the moment of triggering, and the first The escape drone adjacent to the first The experience recovery status of the escaped drone; Indicates the state of being triggered. Below, with the The escape drone adjacent to the first Control gain of the escapee drone; Indicates the state of being triggered. Below, with the The escape drone adjacent to the first The DET-optimized equivalent control law for an escapee drone, and for The estimate; Indicates the state of being triggered. Below, with the The escape drone adjacent to the first The DET-optimized switching control law for an escapee drone, and for The estimate; (56)； in, and They represent the first time. The first trigger moment HSMS and state error of an escaped drone; Indicates the first At the moment of triggering, and the first The escape drone adjacent to the first The experience recovery status of the escaped drone; Indicates the state of being triggered. Below, with the The escape drone adjacent to the first Control gain of the escapee drone; Indicates the state of being triggered. Below, with the The escape drone adjacent to the first The DET-optimized equivalent control law for an escapee drone, and for The estimate; Indicates the state of being triggered. Below, with the The escape drone adjacent to the first The DET-optimized switching control law for an escapee drone, and for The estimate; Indicates the state of being triggered. The next Control gain of the escapee drone; Indicates the state of experience recycling. The next The DET-optimized equivalent control law for a hunter drone. for The estimate; Indicates the state of experience recycling. The next DET-optimized switching control law for a single hunter drone for The estimate.

7. An electronic device, characterized in that, include: One or more processors, and a memory for storing instructions that, when executed by the one or more processors, cause the one or more processors to perform the HSMS-based UAV DET optimization control method under a pursuit-escape game as described in any one of claims 1-6.

8. A computer-readable storage medium, characterized in that, It stores executable instructions that, when executed, cause the processor to perform the HSMS-based UAV DET optimization control method under the pursuit-escape game as described in any one of claims 1-6.

9. A computer program product, characterized in that, Includes a computer program or instructions that, when executed by a processor, implement the HSMS-based UAV DET optimization control method under the pursuit-escape game as described in any one of claims 1-6.

Images