Multi-unmanned aerial vehicle cluster random prediction control method and device based on anti-saturation and packet loss compensation

Abstract

The invention discloses a multi-unmanned aerial vehicle cluster stochastic prediction control method and device based on anti-saturation and packet loss compensation, and the method comprises the steps: building a closed-loop control model of a stochastic system based on a pilot-following multi-unmanned aerial vehicle cluster architecture; through a Markov chain modeling data transmission packet loss process, a forgetting factor is introduced to design a packet loss compensation strategy after random data is randomly lost so as to correct a closed-loop control model of a random system; aiming at the corrected closed-loop control model of the following stochastic system, constructing an anti-actuator saturation control law through convex representation of a saturation function, and further constructing a saturation-limited following stochastic augmentation prediction model; solving a predictive control problem on line based on a random augmentation predictive model; and the control performance index is controlled through rolling optimization, the gain of the feedback controller is updated, and anti-saturation random prediction control following with random following in a packet loss environment is realized. According to the method, the coupling influence of data transmission packet loss and actuator saturation can be effectively solved, and the system performance weakness caused by single problem processing is avoided.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) control technology, and in particular to a method, apparatus, terminal equipment, and computer-readable storage medium for stochastic predictive control of multi-UAV swarms based on anti-saturation and packet loss compensation. Background Technology

[0002] With the rapid development of UAV technology, its applications in reconnaissance, surveying and mapping, logistics, and agricultural plant protection are becoming increasingly widespread. Multi-UAV collaborative operations have become an important development direction due to their efficiency advantages. Among them, the navigator-follower architecture, with its clear control logic and high reliability, is widely used in multi-UAV swarming tasks—by designating a navigator to plan the path, the other follower UAVs track the status of the navigator, thus achieving multi-UAV swarm collaboration.

[0003] However, existing multi-UAV leader-follower systems face two major problems in practical applications, severely impacting control performance. First, there's the data transmission packet loss problem: commands are transmitted wirelessly between the controller and the follower's actuators. This wireless communication is susceptible to environmental interference (such as electromagnetic noise and obstructions), leading to data loss, which is also random. Traditional control methods lack compensation mechanisms for packet loss, easily causing control command interruptions, decreased following accuracy, and even tracking instability. Second, there's the actuator saturation problem: when the leader UAV suddenly accelerates or turns, or when the initial state deviation between the two UAVs is too large, the follower needs to quickly correct the deviation, potentially causing control commands to exceed the motor speed limit. The actual motor input is clamped to the maximum speed, resulting in slow convergence of tracking errors and even system oscillation and instability.

[0004] Furthermore, traditional control methods are mostly designed for single packet loss or saturation problems, without fully considering the coupling effect between the two; and they do not combine predictive control with online optimization capabilities to improve system robustness, making it difficult to adapt to complex dynamic scenarios. Therefore, there is an urgent need for a robust control method that can simultaneously solve the problems of data packet loss and actuator saturation, ensuring the stability and following accuracy of the UAV pilot-follow system. Summary of the Invention

[0005] To address the shortcomings of the prior art, this invention provides a method, apparatus, terminal device, and computer-readable storage medium for stochastic predictive control of multi-UAV swarms based on anti-saturation and packet loss compensation. This effectively solves the coupling effect of data transmission packet loss and actuator saturation, and avoids system performance bottlenecks caused by handling a single problem.

[0006] The first objective of this invention is to provide a stochastic predictive control method for multi-UAV swarms based on anti-saturation and packet loss compensation.

[0007] The second objective of this invention is to provide a multi-UAV swarm random prediction control device based on anti-saturation and packet loss compensation.

[0008] A third objective of this invention is to provide a terminal device.

[0009] A fourth objective of this invention is to provide a computer-readable storage medium.

[0010] The first objective of this invention can be achieved by adopting the following technical solution: A stochastic predictive control method for multi-UAV swarms based on anti-saturation and packet loss compensation, the method comprising: Based on the leader-follower multi-UAV swarm architecture, and combining the translational and rotational dynamics of UAVs, a closed-loop control model for the follower system is constructed. By modeling the data transmission packet loss process using Markov chains, a forgetting factor is introduced to design a packet loss compensation strategy after random loss of follower data, so as to correct the closed-loop control model of the follower system. For the modified closed-loop control model of the follower system, an anti-actuator saturation control law is constructed by using the convex representation of the saturation function; based on the anti-actuator saturation control law, a saturation-constrained augmented prediction model for the follower system is constructed. Based on the augmented predictive model of the follower, the predictive control problem is solved online; by rolling optimization of the control performance index and updating the feedback controller gain, anti-saturation stochastic predictive control of the follower in a packet loss environment is achieved.

[0011] Preferably, the step of modeling the data transmission packet loss process using Markov chains and introducing a forgetting factor to design a packet loss compensation strategy after random data loss includes: Because random data loss may occur during data transmission from the controller to the follower actuator, a random variable is introduced. Represent the data state at time k; assume the data loss process takes values ​​from the set. For a discrete-time homogeneous Markov chain, the transition probability matrix is:

[0012] In the formula, , , and Let Pr(A|B) represent the probability of failure and the probability of recovery, respectively, and let Pr(A|B) represent the probability of event A occurring given that event B has occurred. Let the actual input of the actuator be If the data transmission is successful, then , Let be the system input vector at time k; if data is lost, then ;but The expression is:

[0013] In the formula, Forgetting factor, For standard units, nonlinear saturation function; Assuming the maximum number of data loss instances does not exceed When the system has lost h data points consecutively, the system will use the latest control input stored in the actuator cache. ; .

[0014] Preferably, the construction of the anti-actuator saturation control law through the convex representation of the saturation function includes: Define the state feedback matrix as F, and define the state feedback rate as... ,but:

[0015] In the formula, For standard units, nonlinear saturation function, This represents the control input predicted at time k+i from time k. Let F represent the predicted system state at time k+i, and let F be the state feedback matrix. set up Define a polyhedron for the i-th row of matrix F. ;set up It is a positive definite matrix. ellipsoid Contained in a polyhedron The sufficient condition is: Let V be all A set of diagonal matrices, where each diagonal element is either 1 or 0. Each element in V is labeled as... E i ,definition ; Based on the given matrix ,like ,but:

[0016] In the formula, Represents the convex hull; Can Represented as:

[0017] In the formula, For parameters that depend on the follower state and satisfy... , .

[0018] Preferably, the saturated constrained follower augmentation prediction model is:

[0019]

[0020] In the formula, The state at time k+1 is the randomly augmented state. Forgetting factor, This is the updated state feedback matrix. , z ( k () represents the random augmented state at time k. For discrete scheduling variables, For discrete scheduling variables The system matrix changes, B is the input matrix, and I is the identity matrix. Let k be the data transmission state variable at time k. x ( k Let k be the system state at time k. This is the actual input to the actuator at time k-1.

[0021] Preferably, the closed-loop control model of the follower system is as follows:

[0022] In the formula, for x ( t The first derivative with respect to time t, the state vector , , , These are the roll angle, pitch angle, and yaw angle of the following aircraft, respectively. , , They are respectively , , Corresponding angular velocity; input vector , , , These are the roll moment, pitch moment, and yaw moment of the follower aircraft, respectively; scheduling variables. , The total propeller speed of the following aircraft. For scheduling variables The system matrix changes, and B is the input matrix; Discretizing the above closed-loop control model yields the final closed-loop control model as follows:

[0023] In the formula, x ( k +1) x (k The system state vectors at times k+1 and k are respectively. u ( k ) is the system input vector at time k.

[0024] Preferably, the final closed-loop control model considers inherent actuator saturation physical constraints, including: Because the actuators of drones exhibit saturation characteristics, a standard unit nonlinear saturation function is introduced. Therefore, the final closed-loop control model is reshaped as follows: .

[0025] Preferably, when solving the predictive control problem online, a convex optimization problem with the objective of minimizing the expected upper bound of the performance index is solved on a rolling basis at each sampling time, according to... Update controller gain F ( k Design control law This signal, acting as the control input signal at time k, is transmitted to the follower actuator to achieve stochastic stability and optimal predictive control of the system, ensuring its anti-saturation and anti-packet-loss properties. Y ( k ), M 1( k All of these are matrices obtained through convex optimization solutions. x ( k Let be the system state of the follower system at time k.

[0026] The second objective of this invention can be achieved by adopting the following technical solution: A stochastic prediction control device for multi-UAV swarms based on anti-saturation and packet loss compensation, the device comprising: The first building module is used to construct a closed-loop control model for the follower system based on the leader-follower multi-UAV swarm architecture and by combining the translational and rotational dynamics of UAVs. The design module is used to model the data transmission packet loss process through Markov chains, introduce a forgetting factor to design a packet loss compensation strategy after random loss of follower data, so as to correct the closed-loop control model of the follower system. The second construction module is used to construct an anti-actuator saturation control law by using the convex representation of the saturation function for the closed-loop control model of the modified follower system; and to construct a saturated-constrained follower augmented prediction model based on the anti-actuator saturation control law. The online solver module is used to solve predictive control problems online based on the augmented predictive model of the follower machine. By rolling optimization of control performance indicators and updating the feedback controller gain, it realizes anti-saturation stochastic predictive control for follower machine following under packet loss environment.

[0027] The third objective of this invention can be achieved by adopting the following technical solution: A terminal device includes a processor and a memory for storing processor-executable programs. When the processor executes the program stored in the memory, it implements the above-described stochastic predictive control method for multi-UAV swarms based on anti-saturation and packet loss compensation.

[0028] The fourth objective of this invention can be achieved by adopting the following technical solution: A computer-readable storage medium storing a program that, when executed by a processor, implements the above-described stochastic predictive control method for multi-UAV swarms based on anti-saturation and packet loss compensation.

[0029] The present invention has the following advantages over the prior art: (1) Balancing random packet loss compensation and anti-saturation control: The packet loss process is modeled by Markov chain and packet loss compensation is achieved by combining the forgetting factor; the anti-saturation control law is constructed by the convex representation of the saturation function, which effectively solves the coupling effect between the two and avoids the system performance bottleneck caused by single problem handling.

[0030] (2) Mechanical decoupling and robustness enhancement: The cascaded structure is used to decouple the translational and rotational dynamics of the follower, reducing the control complexity of the follower; combined with the online optimization capability of stochastic predictive control, the controller gain is updated in a rolling manner to enhance the system's resistance to attitude parameter perturbations and environmental disturbances, ensuring the closed-loop stability of the follower.

[0031] (3) High tracking accuracy: By continuously optimizing the infinite time domain performance index and updating the controller gain in real time, the follower can still quickly converge the tracking error under packet loss and saturation conditions, thereby improving the navigation-following accuracy and operational safety.

[0032] (4) Strong engineering practicality: The model and strategy are designed based on the actual hardware characteristics of the UAV (such as the upper limit of motor speed and the packet loss rate of wireless communication). The optimization problem can be solved efficiently through the existing toolbox without complex calculations. The sampling period can be adjusted according to the UAV response speed to meet the real-time control requirements. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0034] Figure 1This is a flowchart of the stochastic predictive control method for multi-UAV swarms based on anti-saturation and packet loss compensation according to Embodiment 1 of the present invention; Figure 2 This is a schematic diagram of the pilot-follower multi-drone swarm architecture of Embodiment 1 of the present invention; Figure 3 This is a schematic diagram of the cascaded architecture of the pilot-follower system in Embodiment 1 of the present invention; Figure 4 This is a schematic diagram of the state transition of the Markov packet loss process in Embodiment 1 of the present invention; Figure 5 This is an architecture diagram of the follower process under packet loss compensation and predictive control in Embodiment 1 of the present invention; Figure 6 This is a structural block diagram of the multi-UAV swarm stochastic prediction control device based on anti-saturation and packet loss compensation according to Embodiment 2 of the present invention; Figure 7 This is a structural block diagram of the terminal device according to Embodiment 3 of the present invention. Detailed Implementation

[0035] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. It should be understood that the specific embodiments described are merely used to explain this application and are not intended to limit this application.

[0036] Example 1: like Figure 1 As shown, this embodiment provides a stochastic predictive control method for multi-UAV swarms based on anti-saturation and packet loss compensation, including the following steps: S101. Based on the dynamic model of a quadcopter UAV, a leader-follower multi-UAV swarm architecture is constructed, which is then transformed into a closed-loop control model of the follower system, taking into account the inherent physical limitations of actuator saturation.

[0037] Specifically, a dynamic model of the UAV navigation-following system is constructed, defining the ENU ground coordinate system (E-Frame): the origin is set at the initial takeoff position of the UAV, the E-axis (east) is aligned with the geographic direction, the N-axis (north) is aligned with the geographic longitude direction, and the U-axis (sky) is perpendicular to the ground and upwards, following the right-hand rule; this coordinate system is used to describe the position and attitude of the UAV in global space. A body coordinate system (B-Frame) is defined: the origin is located at the center of gravity of the UAV, the X-axis points in the direction of motor 1, the Z-axis is perpendicular to the fuselage and upwards, and the Y-axis is determined by the right-hand rule (i.e., the X-axis rotated 90° counterclockwise to the direction of the Z-axis); this coordinate system is used to describe the translational and rotational velocities of the UAV.

[0038] The follow system adopts a "leader + follower" multi-drone swarm architecture, such as... Figure 2 As shown, the cluster consists of one navigator (L) and n follower aircraft (L, L ... The navigator serves as the reference source, assigning differentiated reference signals to each follower aircraft.

[0039] To improve control safety and smoothness, the follower is divided into a position control subsystem (translation subsystem) and an attitude control subsystem (rotation subsystem). Let the current position vector of the follower in the ENU coordinate system be... The attitude vector is Using the current attitude of the navigator as the "base anchor point," and superimposing the "differentiated requirement correction amount of the follower aircraft," this serves as the reference attitude angle for each follower aircraft, making the follower aircraft's reference attitude angle as follows: .like Figure 3 As shown, the cascaded structure of the two-stage subsystem decouples translational and rotational dynamics: the inner loop tracks the reference attitude angle. The outer loop assumes that the navigator's three attitude angles are perfectly tracked and the follower's state is stable. Calculate the required control inputs for the system. U 1 and follower reference attitude angle .in, , With path planner generated Together they serve as the tracking targets of the attitude control system.

[0040] The translational velocity vector in the body coordinate system is The rotational velocity vector is According to the XYZ Euler angle rotation rule, the translational velocity relationship between the ENU coordinate system and the body coordinate system is described by the rotation matrix R: (1) In the formula, The specific form is as follows: (2) The relationship between rotational speed and attitude angular velocity can be expressed as: (3) In the formula, T The transformation matrix has the following specific form: (4) Furthermore, assume the follower is a rigidly symmetric body with its geometric center coinciding with its center of gravity; let its mass be m, and its rotational inertia matrix be... Furthermore, the mass and rotational mass matrices are constant, and dynamic changes caused by mass variations or uneven distribution are not considered; only gravity and rotor lift are considered, ignoring secondary environmental factors. Lift modeling is performed for the follower aircraft. The rotation of the follower's propeller generates a downward airflow, which in turn produces an upward lift force of equal magnitude and opposite direction. The lift expression is: (5) In the formula, The lift coefficient, Let ω be the rotational angular velocity of the i-th rotor (i = 1, 2, 3, 4, corresponding to the four rotors).

[0041] Therefore, the expression for the total lift of the aircraft is: (6) because U 1. The tension vector in the body coordinate system acting only along the Z-axis. F B It can be represented as: (7) through R Transformation, Lift in ENU Coordinates F E It can be represented as: (8) Next, the control torque of the follower aircraft is modeled. The roll torque (9) and pitch torque (10) are achieved by controlling the thrust difference between the rotors of the follower aircraft, and their expressions are as follows: (9) (10) In the formula, l This refers to the length of the follower arm.

[0042] As the four rotors of the aircraft rotate, they beat against the air, generating opposing forces and torques. The expression for the yaw moment is: (11) In the formula, This is the inverse torque coefficient.

[0043] Furthermore, since the follower motor has a stable and fixed axis when rotating at high speed, and the gyroscopic torque will resist the rotation axis deviation to maintain the initial state when the attitude changes, the total moment of inertia of a single rotor follower motor rotor and rotor blade about the rotation axis is: J TP , The total propeller speed is Defined as: (12) Combining Newton's second law and the theorem of angular momentum, the equations of motion for translational movement and the equations of motion for rotational movement of a following aircraft are as follows: (13) (14) In the formula, The external force vector acting on the follower. This is the torque vector.

[0044] By combining equations (13) and (14) and substituting them into the force and torque model, and ignoring external disturbances, the motion equations of the attitude subsystem after combining the equations are transformed into a linear parameter variation system model. The transformed model is as follows: (15) Using the parametric nonlinear embedding method, the attitude dynamics equation (15) is rewritten in state-space form, that is, the state-space equation of the multicellular uncertain system model is as follows: (16) in:

[0045] The state-space equation (closed-loop control model) can be expressed as: (17)

[0046] In the formula, the state vector Input vector scheduling variables .

[0047] Discretizing model (17) yields the closed-loop equation (closed-loop control model) of the follower system as follows: (18) In the formula, x ( k +1) x ( k The system state vectors at times k+1 and k are respectively. u( k Let ) be the system input vector at time k. For the discretized scheduling variables, This is the discretized system matrix.

[0048] Considering that in practical systems, due to physical component limitations and safety requirements, the actuators of UAVs exhibit saturation characteristics, a standard unit nonlinear saturation function is introduced. Then (18) is reshaped as: (19) S102. The packet loss process is modeled by Markov chain, and a forgetting factor is introduced to design a packet loss compensation strategy after random loss of follower data, so as to correct the closed-loop control model of the follower system.

[0049] Considering the problem of data transmission packet loss, a packet loss compensation strategy is designed after random loss of random data. A forgetting factor is introduced to deal with the adverse effects of multiple consecutive data losses and to modify the closed-loop control model.

[0050] Specifically, such as Figure 4 As shown, random data loss may occur when data is transmitted from the controller to the follower actuator; therefore, a random variable is introduced. Indicates the data state at time k ( This indicates that the data transmission was successful. (This indicates data loss). Assume the data loss process involves taking values ​​from a set. For a discrete-time homogeneous Markov chain, the transition probability matrix is: (20) In the formula, , , and Let Pr(A|B) represent the probability of failure and the probability of recovery, respectively. Pr(A|B) represents the probability of event A occurring given that event B has occurred.

[0051] To mitigate the adverse effects of repeated random data loss, a compensation strategy based on the executor's cache is introduced. Let the actual input to the executor be... If the data transmission is successful, then If data is lost, then ;at this time The expression is: (twenty one) In the formula, It is a forgetting factor.

[0052] Considering the practical implementation of the compensation strategy, it is assumed that the maximum number of data loss incidents does not exceed [a certain threshold]. When the system has continuously lost h ( The system will use the latest control input stored in the actuator cache for this second data input. .

[0053] Furthermore, due to the control input applied before time k is known... Given the current state x(k), the UAV closed-loop dynamic model (19) can be rewritten as: (twenty two) S103. For the closed-loop control model of the modified follower system, construct an anti-actuator saturation control law to realize the convex representation of the saturated state feedback; based on the anti-actuator saturation control law, establish a saturated-constrained follower augmented prediction model.

[0054] Specifically, when the lead aircraft suddenly accelerates, or the initial state deviation between the two aircraft becomes too large, the follower aircraft needs to quickly correct this large deviation. The control command may exceed the motor speed limit. In this case, the actual motor input is clamped at the motor's maximum speed, causing the tracking error to converge more slowly or even become unstable. Therefore, an anti-saturation control law needs to be constructed to mitigate the negative impact of actuator saturation on the follower's performance. To solve this problem, the state feedback matrix is ​​defined as F, and the state feedback rate is defined as... If we use the standard unit nonlinear saturation function again, then: (twenty three) In the formula, This represents the control input predicted at time k+i from time k. This represents the system state predicted at time k+i from time k.

[0055] set up Define a polyhedron for the i-th row of matrix F. ,set up It is a positive definite matrix. ellipsoid Contained in a polyhedron The sufficient condition is: Let V be all A set of diagonal matrices, where each diagonal element is either 1 or 0, has 2^m elements. Each element in V is labeled as... and define Obviously, if ,but Given a matrix ,like ,but: (twenty four) In the formula, This represents the convex hull.

[0056] This means that it is possible to Represented as: (25) In the formula, For parameters that depend on the follower state and satisfy... , .

[0057] According to (25), for The UAV closed-loop dynamic system (22) can be expanded as follows: (26) In the formula, Representing a symmetrical polyhedron , This represents the r-th row of matrix H(k).

[0058] Define the augmented state of the follower machine at time k as The augmented closed-loop model is as follows: (27) In the formula, This is the updated state feedback matrix. z(k+1) represents the random augmented state at time k+1, and I is the identity matrix. Let k be the data transmission state variable at time k. This is the actual input to the actuator at time k-1.

[0059] S104. Based on the augmented predictive model of the follower, the predictive control problem is solved online; by rolling optimization of the control performance index and updating the feedback controller gain, anti-saturation stochastic predictive control of the follower in a packet loss environment is achieved.

[0060] Specifically, under the combination of packet loss compensation and anti-saturation control, the inner-loop controlled process of the pilot-follower system is as follows: Figure 5 As shown. For the feedback control section, model predictive control technology is used to continuously update the feedback control gain and optimize control performance.

[0061] Based on the aforementioned augmented predictive model (27), the predictive control algorithm is designed to solve for the following minimization of control performance objective: (28) Among them, the infinite time domain index is: (29) In the formula, This represents the random augmented state predicted at time k+i based on the time at time k. This indicates that the actual input of the actuator at time k+1 is predicted at time k. (S1>0 is the state weight matrix), R>0 is the input weight matrix. This is the conditional expectation based on the current augmented state.

[0062] Furthermore, the stability method based on Lyapunov functions is used to solve the aforementioned saturation-resistant stochastic robust predictive control problem.

[0063] Set a quadratic Lyapunov function: (30) In the formula, This represents the data transmission state vector predicted at time k+i from time k. , and It is a positive definite matrix. When... make , The anti-saturation stochastic model predictive control method is reshaped into a convex optimization problem with LMI constraints: considering the pilot-follower model (27) at sampling time k, given parameters If there exist matrices Fnew, H(k), Q(k) > 0, then the symmetric positive definite matrix... , , , This minimizes the following problem for all Solvable: (31) Upper bound constraints on performance metrics:

[0064] (32) In the formula, ; Input saturation constraints: (33) Feasibility constraints: (34) (35) In the formula, ; ; .

[0065] Stochastic stability constraints: (36) (37) (38) (39) In the formula, ( (is a symmetric positive definite matrix) ; ; ; ; ; ; ; ; ; .

[0066] At this point, at each sampling time, a convex optimization problem with the objective of minimizing the expected upper bound of the performance index is solved on a rolling basis, according to... Update controller gain F ( k Design control law This signal, acting as the control input signal at time k, is transmitted to the follower actuator, ultimately achieving stochastic stability and optimal predictive control of the system, which is resistant to saturation and packet loss. Y ( k ), M 1 ( k All of these are matrices obtained through convex optimization solutions.

[0067] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware, and the corresponding program can be stored in a computer-readable storage medium.

[0068] It should be noted that although the method operations of the above embodiments are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. On the contrary, the order of execution of the described steps may be changed. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0069] Example 2: like Figure 6 As shown, this embodiment provides a stochastic prediction control device for multi-UAV swarms based on anti-saturation and packet loss compensation. The device includes a first construction module 601, a design module 602, a second construction module 603, and an online solution module 604, wherein: The first building module 601 is used to build a closed-loop control model of the follower system based on the leader-follower multi-UAV cluster architecture and combining the translational and rotational dynamics of UAVs. Design module 602 is used to model the data transmission packet loss process through Markov chain, introduce a forgetting factor to design a packet loss compensation strategy after random loss of follower data, so as to correct the closed-loop control model of the follower system. The second construction module 603 is used to construct an anti-actuator saturation control law by using the convex representation of the saturation function for the closed-loop control model of the modified follower system; and to construct a saturated-constrained follower augmented prediction model based on the anti-actuator saturation control law. The online solver module 604 is used to solve predictive control problems online based on the augmented predictive model of the follower machine; by rolling optimization of control performance indicators and updating the feedback controller gain, it realizes anti-saturation stochastic predictive control for follower machine following under packet loss environment.

[0070] The specific implementation of each module in this embodiment can be found in Embodiment 1 above, and will not be repeated here. It should be noted that the device provided in this embodiment is only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure can be divided into different functional modules to complete all or part of the functions described above.

[0071] Example 3: This embodiment provides a terminal device, which can be a computer, such as... Figure 7 As shown, the processor 702, memory, input device 703, display 704, and network interface 705 are connected via system bus 701. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium 706 and an internal memory 707. The non-volatile storage medium 706 stores the operating system, computer programs, and database. The internal memory 707 provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. When the processor 702 executes the computer programs stored in the memory, it implements the multi-UAV swarm stochastic predictive control method based on anti-saturation and packet loss compensation described in Embodiment 1 above.

[0072] Example 4: This embodiment provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the multi-UAV swarm stochastic predictive control method based on anti-saturation and packet loss compensation described in Embodiment 1 above.

[0073] It should be noted that the computer-readable storage medium in this embodiment can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof.

[0074] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope disclosed in the present invention, based on the technical solution and inventive concept of the present invention, shall fall within the scope of protection of the present invention.

Claims

1. A stochastic predictive control method for multi-UAV swarms based on anti-saturation and packet loss compensation, characterized in that, The method includes: Based on the leader-follower multi-UAV swarm architecture, and combining the translational and rotational dynamics of UAVs, a closed-loop control model for the follower system is constructed. By modeling the data transmission packet loss process using Markov chains, a forgetting factor is introduced to design a packet loss compensation strategy after random loss of follower data, so as to correct the closed-loop control model of the follower system. For the modified closed-loop control model of the follower system, an anti-actuator saturation control law is constructed by using the convex representation of the saturation function; based on the anti-actuator saturation control law, a saturation-constrained augmented prediction model for the follower system is constructed. Based on the augmented predictive model of the follower, the predictive control problem is solved online; by rolling optimization of the control performance index and updating the feedback controller gain, anti-saturation stochastic predictive control of the follower in a packet loss environment is achieved.

2. The multi-UAV swarm stochastic predictive control method according to claim 1, characterized in that, The process of data transmission packet loss is modeled using Markov chains, and a packet loss compensation strategy is designed to follow random data loss after introducing a forgetting factor, including: Because random data loss may occur during data transmission from the controller to the follower actuator, a random variable is introduced. Represent the data state at time k; assume the data loss process takes values ​​from the set. For a discrete-time homogeneous Markov chain, the transition probability matrix is: In the formula, , , and Let Pr(A|B) represent the probability of failure and the probability of recovery, respectively, and let Pr(A|B) represent the probability of event A occurring given that event B has occurred. Let the actual input of the actuator be If the data transmission is successful, then , Let be the system input vector at time k; if data is lost, then ;but The expression is: In the formula, Forgetting factor, For standard units, nonlinear saturation function; Assuming the maximum number of data loss instances does not exceed When the system has lost h data points consecutively, the system will use the latest control input stored in the actuator cache. ; .

3. The multi-UAV swarm stochastic predictive control method according to claim 1, characterized in that, The construction of an anti-actuator saturation control law through the convex representation of the saturation function includes: Define the state feedback matrix as F, and define the state feedback rate as... ,but: In the formula, For standard units, nonlinear saturation function, This represents the control input predicted at time k+i from time k. Let F represent the predicted system state at time k+i, and let F be the state feedback matrix. set up Define a polyhedron for the i-th row of matrix F. ;set up It is a positive definite matrix. ellipsoid Contained in a polyhedron The sufficient condition is: Let V be all A set of diagonal matrices, where each diagonal element is either 1 or 0. Each element in V is labeled as... E i ,definition ; Based on the given matrix ,like ,but: In the formula, Represents the convex hull; Can Represented as: In the formula, For parameters that depend on the follower state and satisfy... , .

4. The multi-UAV swarm stochastic predictive control method according to claim 3, characterized in that, The saturated and constrained follower augmentation prediction model is as follows: In the formula, The state at time k+1 is the randomly augmented state. Forgetting factor, This is the updated state feedback matrix. , z ( k () represents the random augmented state at time k. For discrete scheduling variables, For discrete scheduling variables The system matrix changes, B is the input matrix, and I is the identity matrix. Let k be the data transmission state variable at time k. x ( k Let k be the system state at time k. This is the actual input to the actuator at time k-1.

5. The multi-UAV swarm stochastic predictive control method according to any one of claims 1 to 4, characterized in that, The closed-loop control model of the follower system is as follows: In the formula, for x ( t The first derivative with respect to time t, the state vector , , , These are the roll angle, pitch angle, and yaw angle of the following aircraft, respectively. , , They are respectively , , Corresponding angular velocity; input vector , , , These are the roll moment, pitch moment, and yaw moment of the follower aircraft, respectively; scheduling variables. , The total propeller speed of the following aircraft. For scheduling variables The system matrix changes, and B is the input matrix; Discretizing the above closed-loop control model yields the final closed-loop control model as follows: In the formula, x ( k +1) x ( k The system state vectors at times k+1 and k are respectively. u ( k ) is the system input vector at time k.

6. The multi-UAV swarm stochastic predictive control method according to claim 5, characterized in that, The final closed-loop control model considers inherent actuator saturation physical constraints, including: Because the actuators of drones exhibit saturation characteristics, a standard unit nonlinear saturation function is introduced. Therefore, the final closed-loop control model is reshaped as follows: 。 7. The multi-UAV swarm stochastic predictive control method according to any one of claims 1 to 4, characterized in that, When solving predictive control problems online, a convex optimization problem with the objective of minimizing the expected upper bound of the performance index is solved on a rolling basis at each sampling time. Update controller gain F ( k Design control law This signal, acting as the control input signal at time k, is transmitted to the follower actuator to achieve stochastic stability and optimal predictive control of the system, ensuring its anti-saturation and anti-packet-loss properties. Y ( k ), M 1( k All of these are matrices obtained through convex optimization solutions. x ( k ) represents the system state of the follower system at time k.

8. A stochastic predictive control device for multi-UAV swarms based on anti-saturation and packet loss compensation, characterized in that, The device includes: The first building module is used to construct a closed-loop control model for the follower system based on the leader-follower multi-UAV swarm architecture and by combining the translational and rotational dynamics of UAVs. The design module is used to model the data transmission packet loss process through Markov chains, introduce a forgetting factor to design a packet loss compensation strategy after random loss of follower data, so as to correct the closed-loop control model of the follower system. The second construction module is used to construct an anti-actuator saturation control law by using the convex representation of the saturation function for the closed-loop control model of the modified follower system; and to construct a saturated-constrained follower augmented prediction model based on the anti-actuator saturation control law. The online solver module is used to solve predictive control problems online based on the augmented predictive model of the follower machine. By rolling optimization of control performance indicators and updating the feedback controller gain, it realizes anti-saturation stochastic predictive control for follower machine following under packet loss environment.

9. A terminal device, comprising a processor and a memory for storing a processor-executable program, characterized in that, When the processor executes the program stored in the memory, it implements the multi-UAV swarm stochastic prediction control method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the multi-UAV swarm stochastic predictive control method according to any one of claims 1 to 7.

Images