A distributed dynamic event-triggered cooperative control method for unmanned aerial vehicle cluster

Abstract

This application relates to the field of distributed cooperative control technology, and in particular to a distributed dynamic event-triggered cooperative control method for UAV swarms. This method does not require continuously receiving the real flight status of neighbors, nor does it require pre-obtaining global topology information such as the total number of nodes, Laplacian matrix eigenvalues, and left eigenvectors. It only relies on the UAV's own state, the most recent triggered broadcast value of its neighbors, and the local communication gain to achieve navigation state estimation and updates. Compared to existing methods that rely on continuous communication or global topology information, the method designed in this application can reduce the communication burden between UAVs and improve distributed deployment capabilities under directed communication topology and limited communication resources.

Description

Technical Field

[0001] This application relates to the field of distributed collaborative control technology, specifically to a distributed dynamic event-triggered collaborative control method for unmanned aerial vehicle (UAV) swarms. Background Technology

[0002] Unmanned aerial vehicle (UAV) swarms are widely used in scenarios such as formation flying, collaborative reconnaissance, area inspection, and emergency communication due to their wide range of collaborative operations, high mission efficiency, and strong system redundancy. To enable multiple UAVs to complete collaborative tasks under conditions of limited communication and environmental uncertainty, technologies such as distributed consensus control, leader-follower control, event-triggered communication, and adaptive observers have been extensively studied. Among these, distributed consensus control allows UAVs to achieve swarm collaboration relying solely on neighbor information; leader-follower control enables following UAVs to track the movement state of the leader UAV; event-triggered communication reduces unnecessary continuous information interaction between UAVs; and adaptive observers can be used to estimate unknown parameters and state information that cannot be directly obtained.

[0003] In practical applications, drones typically communicate with each other via directed links. The dynamic parameters and nonlinear characteristics of each drone may not be the same, and some parameters of the lead drone may also be unknown. At the same time, drones have limited onboard communication and computing resources.

[0004] Therefore, it is still necessary to further study and improve how to design a fully distributed cooperative control method for heterogeneous uncertain UAV swarms that is dynamically triggered without relying on global topology information, so that the following UAV can stably track the state of the lead UAV and reduce the communication burden. Summary of the Invention

[0005] In view of this, this application discloses a distributed dynamic event-triggered cooperative control method for UAV swarms, to address the shortcomings of existing UAV swarm technologies in complex scenarios such as fully distributed implementation, dynamic event-triggered communication, estimation of unknown navigation parameters, and heterogeneous uncertain UAV dynamics. The distributed dynamic event-triggered cooperative control method for UAV swarms is applied to a UAV swarm including one lead UAV and several follower UAVs, comprising:

[0006] S1. Establish a directed leader-follower drone topology model;

[0007] S2. Establish state equations for the lead drone and the heterogeneous uncertain second-order nonlinear follower drone based on the drone topology model; the state equations require that there is at least one directed path from the lead drone to each follower drone; the state equations include the drone's position state, velocity state, and control input;

[0008] S3. Set a trigger time. Each follower drone performs state estimation on the lead drone, maintains the lead drone's estimation parameters, and broadcasts the local lead drone's estimation parameters at the trigger time. The lead drone's estimation parameters include: lead position estimation. Navigation speed estimation Navigation maneuver parameter estimation ;

[0009] S4. The following drone constructs a local inconsistency quantity based on the difference between the navigation estimate most recently broadcast and maintained by the current node and the neighboring drone node, combined with the difference between the current node's broadcast maintained value and the latest broadcast value from the navigation end.

[0010] S5. Construct the communication gain and update the communication gain of each channel using a projection-type local adaptive law;

[0011] S6. Construct a state observer and use local inconsistency and communication gain to update the local maintenance of the pilot position estimate, pilot speed estimate, and pilot maneuver parameter estimate of the following UAV; the parameter observer is set at the pilot end and is used to obtain the pilot end estimate of the pilot maneuver parameter.

[0012] S7. The following drone calculates adaptive compensation control input based on its own state and the estimated parameters of the local lead drone to achieve coordinated tracking of the lead drone's state.

[0013] The beneficial effects of this application include:

[0014] In the method designed in this application, the control input, distributed observer, dynamic triggering function, and local gain update law all do not contain the total number of nodes, eigenvalues ​​of the Laplacian matrix, or left eigenvectors. Alternatively, based on other global topology thresholds, each UAV only needs to broadcast information about its local state and the triggering time of its neighbors to operate, achieving fully distributed operation. Based on the problem of fixed feedback gain and stability conditions depending on global topology quantities, a UAV controller, observer, triggering function, and gain update law that do not contain global information such as the total number of nodes, Laplacian matrix eigenvalues, and left eigenvectors are provided to achieve fully distributed consistent collaborative control without relying on global information.

[0015] To address the issue of insufficient communication adjustment capability of static trigger thresholds, this application introduces dynamic trigger envelope and bounded dynamic scale to reduce unnecessary communication while ensuring that trigger errors gradually disappear. The following UAV only broadcasts navigation state estimation and navigation parameter estimation when the dynamic trigger conditions are met, and remains silent in non-trigger intervals. This can significantly reduce bandwidth occupation and airborne energy consumption caused by continuous communication or high-frequency periodic communication, resulting in low communication costs.

[0016] Existing technologies only apply to problems involving undirected graphs, integrators, or linear systems where some fully distributed dynamic triggering methods are applicable. This application extends the fully distributed dynamic event triggering concept to general directed leader-follower drone swarms and heterogeneous uncertain second-order nonlinear flight channels. It is applicable to general directed leader-follower topologies, requiring only that there is a directed path from the lead drone to each follower drone, without requiring the communication graph to be undirected, balanced, or that the follower drone subgraphs to be strongly connected. It is suitable for common unidirectional links and temporary network structures in drone swarms, providing a new design approach for those skilled in the art.

[0017] To address the difficulty in handling asymmetric coupling terms in directed graphs caused by directly reconstructing the actual flight states of neighbors, this application establishes a distributed leader state and leader parameter observation layer, enabling each following UAV to track only the leader state estimate it maintains.

[0018] Different follower UAVs may have different unknown parameters and different nonlinear regression terms. The local adaptive controller compensates for its own dynamic uncertainties, improves the adaptability and robustness to load differences, aerodynamic parameter changes and model errors, and adapts to heterogeneous uncertain UAV systems.

[0019] It can handle unknown navigation maneuver parameters. Through the parameter observer at the navigator end and the distributed parameter estimator of the following UAV, the navigation parameter estimation error and the navigation state observation error can be converged, thereby supporting cooperative tracking under unknown maneuver reference;

[0020] With bounded local gain, the projection-type adaptive communication gain can adjust the transient coupling strength according to the local error, avoid the unbounded growth of ordinary adaptive gain, and does not require pre-selection of a fixed large gain based on unknown global topology quantities.

[0021] Dynamic envelope and bounded scaling factor can adjust the trigger threshold according to local error, improving the trade-off between communication times and tracking performance while ensuring that the trigger error asymptotically disappears, making dynamic triggering more flexible;

[0022] To address the issue of mutual dependence between global existence and Zeno-free proofs in dynamically event-triggered drone swarms, a concise analysis process for finite-time boundedness and Zeno-free proofs is presented. The stability analysis proceeds in the order of "local existence—finite-time boundedness—Zeno-free—global existence—asymptotic convergence," thus avoiding the common circular proofs of existence and Zeno-free proofs in event-triggered drone swarm analysis. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of a distributed dynamic event-triggered collaborative control method for unmanned aerial vehicle (UAV) swarms in an embodiment of this application;

[0024] Figure 2 This is a schematic diagram of the directed leader-follower drone swarm topology in an embodiment of this application;

[0025] Figure 3 These are the initial simulation conditions in the embodiments of this application;

[0026] Figure 4 This is a schematic diagram showing the position and speed trajectories of the four follower drones in an embodiment of this application;

[0027] Figure 5 This is a schematic diagram of the leader parameter estimation error and cascade error in the embodiments of this application;

[0028] Figure 6 This is a schematic diagram illustrating the changes in control inputs for each follower drone in the embodiments of this application;

[0029] Figure 7 This is a schematic diagram illustrating the adaptive communication gain evolution in an embodiment of this application;

[0030] Figure 8 This is a schematic diagram illustrating the event triggering time in an embodiment of this application;

[0031] Figure 9 This is a statistical representation of event triggering in the embodiments of this application;

[0032] Figure 10 This is a schematic diagram comparing the triggering error and the dynamic threshold in an embodiment of this application;

[0033] Figure 11 This is a schematic diagram comparing the total tracking error under different triggering parameters in the embodiments of this application;

[0034] Figure 12 These are performance metrics under different triggering parameters in the embodiments of this application;

[0035] Figure 13 This is a schematic diagram comparing the total tracking error of different control communication schemes in the embodiments of this application;

[0036] Figure 14 This application provides a performance comparison of different control communication schemes in its embodiments.

[0037] Figure 15 This is a schematic diagram comparing the total tracking error under different directed topologies in the embodiments of this application;

[0038] Figure 16 These are performance metrics for different directed topologies in the embodiments of this application. Detailed Implementation

[0039] To make the objectives, technical solutions, features, and advantages of this application clearer and to facilitate a better understanding of the technical solutions of this application by those skilled in the art, the following detailed description of this application is provided in conjunction with the accompanying drawings and embodiments.

[0040] Example 1:

[0041] This embodiment includes a distributed dynamic event-triggered cooperative control method for drone swarms, applied to a drone swarm comprising a lead drone and several follower drones, including:

[0042] S1. Establish a directed leader-follower drone topology model and define adjacency weight, access weight, and grounding Laplace matrix.

[0043] For a drone swarm consisting of 1 lead drone and N follower drones:

[0044] use This represents the adjacency weight between following drones. If following drone i can receive broadcast information from following drone j, then... >0, otherwise =0; Establish the adjacency matrix of the subgraph following the drone. This indicates the directed communication relationship between the drones following each other;

[0045] Let D represent the in-degree matrix of the following drone. ;

[0046] use This represents the Laplacian matrix of the subgraph following the drone. ;

[0047] use This represents the access weight of the lead drone to the i-th follower drone. If follower drone i can directly receive information from the lead drone, then... ,otherwise =0, establish access matrix ;

[0048] Furthermore, establish a grounded Laplace matrix. , represents the graph relationships between drone swarms; the inverse matrix and eigenvalues ​​of H are only used for theoretical proof and are not included in any online drone implementation formulas.

[0049] S2. Establish state equations for the lead drone and the heterogeneous uncertain second-order nonlinear follower drone based on the drone topology model; the state equations require that there is at least one directed path from the lead drone to each follower drone; the state equations include the drone's position state, velocity state, and control input.

[0050] Specifically, the position / attitude channel of the i-th following UAV is represented as a heterogeneous uncertain second-order nonlinear system, as shown in the formula:

[0051]

[0052]

[0053] in, These represent the position, speed, and control input of the i-th following UAV, respectively. Indicates an unknown constant parameter. Indicates the first The nonlinear regression vector that follows the dynamics of the UAV itself.

[0054] In the aforementioned heterogeneous uncertain second-order nonlinear system, different following drones and The dimensions can differ; multi-dimensional positions or three-dimensional flight scenes can be expanded using coordinate axes or vectors. The formula for establishing a navigation drone is:

[0055]

[0056]

[0057] in, This indicates unknown navigation maneuver parameters. The navigation regression vector can be obtained from a common mission clock or reference generator.

[0058] S3. Set the trigger time, each follower drone estimates the state of the lead drone, maintains the lead drone's estimated parameters, and broadcasts the local lead drone's estimated parameters at the trigger time.

[0059] Each follower drone i maintains three types of local estimation variables: leader position estimation. Navigation speed estimation Navigation maneuver parameter estimation Node i broadcasts at its own trigger time. The neighbor uses the most recently received value between two triggers. .

[0060] Construct broadcast conditions based on event metrics, determine whether to broadcast local estimates based on dynamic event triggering conditions, and remain silent in non-triggering intervals, including:

[0061] Step 1: Define the local broadcast error using the following formula:

[0062]

[0063] in, This indicates the position estimation error. This indicates the speed estimation error. This indicates the error in estimating the maneuver parameters.

[0064] Step 2: Follow the drone i to construct the event measure based on the local broadcast error, using the following formula:

[0065]

[0066] in, , , Represents position, speed, and maneuver parameters. , , These represent the weighting coefficients of the position estimation broadcast error, velocity estimation broadcast error, and navigation maneuver parameter estimation broadcast error in the event measure, respectively, and are all positive numbers.

[0067] Step 3: Set the fading envelope and bounded dynamic scale :

[0068]

[0069]

[0070]

[0071] Step 4: Define the broadcast trigger time based on the fading envelope and bounded dynamic scale, using the following formula:

[0072]

[0073] When the trigger condition is met, the drone broadcasts the current status. And reset the local broadcast error; remain silent during non-trigger intervals. Navigator endpoint... Similar triggering mechanisms can be configured.

[0074] S4. The following drone constructs a local inconsistency value based on the difference between the navigation estimate most recently broadcast and maintained by the current node and the neighboring drone nodes, combined with the difference between the current node's broadcast maintained value and the most recent broadcast value from the navigation node. .

[0075] Specifically, the formula for constructing the local inconsistency quantity is:

[0076]

[0077]

[0078]

[0079]

[0080] The core difference in the above design lies in propagating the pilot state estimate and pilot parameter estimate, rather than continuously propagating the actual flight states of neighbors. This allows the network coupling of a general directed graph to be processed at the observation layer, avoiding the formation of an asymmetric second-order relative error term directly at the physical tracking layer.

[0081] S5, Constructing Communication Gain The communication gain of each channel is updated using a projection-type local adaptive law.

[0082] Set local communication gain for each follower drone The gain is driven by the local error, and the communication gain is limited to a given local range by the projection operator. The formula is:

[0083]

[0084]

[0085]

[0086] in, The update rate represents the local communication gain. This represents the projection operator, used to maintain a positive lower bound and a finite upper bound on the gain. Represents the vector projection function. Indicates communication gain The projection range, Let represent the communication gain update rate coefficient of the i-th following UAV on the ℓ-th channel, where ℓ∈{p,q,θ}, used to adjust the communication gain on the error driving term. Response speed The error-driven term is used to adjust the coupling strength according to the degree of local inconsistency, while the fading suppression term reduces gain drift during small error phases. Indicates a quantity of local inconsistency. This represents the fading suppression term, used to reduce communication gain drift during periods of small errors. This represents the initial strength of the suppression term. The decay rate of the suppression term is represented by these parameters, all of which are locally configurable positive parameters.

[0087] S6. Construct a state observer and use local inconsistencies and communication gain to estimate the parameters of the lead drone, which is locally maintained by the follower drone. Updates will be made, including:

[0088] Set up a parameter observer at the navigator end to obtain the navigator end estimate of the navigator maneuver parameters. :

[0089]

[0090]

[0091] in, This indicates the estimated speed of the pilot drone. The time derivative, i.e., the update rate of the velocity estimator. Represents the estimated maneuver parameter vector transpose, Represents the navigation regression vector. The feedback gain represents the error in the pilot speed estimation. Indicates the speed status of the pilot drone. The estimated value, This indicates the actual speed status of the pilot drone. Indicates the estimated values ​​of navigation maneuver parameters The derivative, This represents the adaptive gain matrix for estimating navigation parameters.

[0092] Using local inconsistency quantities With communication gain The estimated parameters of the local lead drone for each following drone are updated using the following formula:

[0093]

[0094]

[0095]

[0096] in, , , These represent the estimated parameters of the navigation maneuver. Navigation position estimation Navigation speed estimation The derivative of , , Indicates the corresponding communication gain. , , This represents the corresponding locally inconsistent variable. This represents the transpose of the estimated navigation maneuver parameters. Let represent the navigation regression vector. It can be seen that the aforementioned state observer only uses its own variables, the most recently triggered broadcast value of a neighbor, the most recently broadcast value of the navigator, and the local gain; it does not require continuous neighbor states or global topology information. Therefore, the method designed in this application no longer needs to continuously receive the real flight states of neighbors following the UAV, nor does it need to obtain global topology information such as the total number of nodes, the eigenvalues ​​of the Laplacian matrix, and the left eigenvector in advance. It only relies on its own state, the most recently triggered broadcast value of a neighbor, and the local communication gain to achieve the navigation state estimation update. Compared with existing methods that rely on continuous communication or global topology information, this method can reduce the communication burden between UAVs and improve the distributed deployment capability under directed communication topology and limited communication resources.

[0097] S7. The following drone calculates adaptive compensation control input based on its own state and the estimated parameters of the local lead drone to achieve coordinated tracking of the lead drone's state.

[0098] Follow the drone to only track its own maintenance navigation status estimation ,definition:

[0099]

[0100]

[0101]

[0102] in, Indicates the first The position status of the drone being followed. This represents the speed state of the i-th following drone. This represents the combined error consisting of position tracking error and velocity tracking error. This represents the estimated position of the i-th following drone relative to the local leader. Position tracking error, This represents the estimated speed of the i-th following drone relative to the local leader. Speed ​​tracking error, Positive error combination coefficients are used to adjust the position tracking error in the combined error. The weights in the equation. It should be noted that... Used to estimate the unknown dynamic parameters of the i-th following UAV, distinct from... The latter is used to estimate the maneuver parameters of the lead drone. Because the distributed position observer is... When constructing local sliding mode error and compensation control laws, it is necessary to consider .

[0103] To ensure strict consistency between the local sliding mode error derivation and the distributed position observer, the control law for the following UAV is expressed in a compensated form, as follows:

[0104]

[0105] in, This indicates the control law for compensated unmanned aerial vehicles (UAVs). This represents an estimate of the unknown parameters of the drone itself. Indicates the first A nonlinear regression vector of the following UAV's own dynamics is used to describe and compensate for the dynamic uncertainties of the following UAV. Indicates local pilot speed estimate Time derivative, Represents the error combination coefficient. Indicates speed tracking error. Indicates communication gain. Representing variables with local inconsistencies, Indicates the combined error feedback gain. This represents the combined error consisting of position tracking error and velocity tracking error.

[0106] Calculated in real time by a distributed leader velocity observer, it does not require numerical differentiation of the measured signal; the formula is as follows:

[0107]

[0108] in, Represents the navigation regression vector. Indicates the local communication gain following the speed of the drone. This indicates the local inconsistency in the speed of the drone.

[0109] Adaptive laws are applied based on the unknown parameters of the drone itself.

[0110]

[0111]

[0112] in, This indicates the update rate of the estimated values ​​of the unknown parameters of the drone itself. Indicates the first The adaptive gain positive definite matrix of the drone following the aircraft. This represents the estimation error of the drone's own dynamic parameters. Indicates the first The unknown constant parameters of the drone itself.

[0113] The local sliding mode error equation can be transformed into a standard adaptive control form, avoiding the influence of local sliding mode error equations. In The omission of a certain item caused incoherence in the derivation.

[0114] Example 2:

[0115] This embodiment includes a distributed dynamic event-triggered collaborative control method for UAV swarms. The difference from Embodiment 1 is that this embodiment explains the stability and Zeno-free characteristics of the method, proving that the scheme in Embodiment 1 can ensure the boundedness of the closed-loop signal, eliminate finite-time Zeno behavior, and achieve state convergence in the forward control design.

[0116] Step A: Finite-time boundedness.

[0117] Since each regression vector is locally Lipschitz on the closed-loop compact set, has finite initial values, and the projection operator guarantees... Since the system always lies within a locally given positive bounded interval, a Carathéodory solution exists between any two trigger times. Furthermore, the dynamic threshold... The local gain and adaptive parameters are bounded over any finite time interval.

[0118] Step B: Analyze map observation errors.

[0119] Substitute broadcast error The terms can be uniformly written in the form of "grounding Laplace matrix consistency term + event sampling perturbation + pilot-end estimation perturbation":

[0120]

[0121]

[0122]

[0123]

[0124]

[0125]

[0126] in, , It consists of trigger sampling error and navigation estimation error.

[0127] Step C: Weighted maximum norm of the M-matrix. Since there are directed paths from the lead drone to each follower drone, H is a non-singular M-matrix. In the proof, we take... and define .

[0128] right We can obtain the comparison inequalities:

[0129]

[0130] in, This represents the upper right Dini derivative.

[0131] If d(t) is bounded and tends to zero, then It is bounded and tends to zero. This conclusion is applied, in turn, to the errors in leadership parameter estimation, leadership velocity observation, and leadership position observation. It is used only in the proof and does not enter the online protocol of any drone.

[0132] Step D: Local flight status tracking error analysis. Select a Lyapunov function for each following UAV:

[0133]

[0134] The unknown parameter cross terms can be canceled out by the compensating control law and the parameter update law, resulting in:

[0135]

[0136] Depend on Bounded, And by combining the boundedness of the closed-loop signal, we can obtain Bounded, and further obtained by Barbalat's lemma. .

[0137] Depend on , combined , can be obtained .

[0138] Step E: No Zeno behavior. Taking any finite time interval [0, T], as shown in Step A, the derivatives of the closed-loop signal and the event error are bounded, and thus exist... make ;at the same time There exists a positive lower bound in the region [0,T], and After each trigger Event measurement An event metric requires at least one positive time to grow from zero to a positive threshold. Therefore, there is a positive lower bound between adjacent triggering times, and an infinite number of triggers cannot occur within any finite time interval.

[0139] Step F: Global Existence and Asymptotic Convergence. The finite-time boundedness and absence of Zeno behavior eliminate finite escape, extending the local solution to a global solution. Further analysis using the leader parameter observer, the M-matrix ISS (Interface Settlement Map), and local tracking convergence reveals that all following UAVs ultimately achieve convergence, expressed by the formula:

[0140]

[0141]

[0142] Example 3

[0143] This embodiment includes a distributed dynamic event-triggered collaborative control method for drone swarms. The difference from Embodiment 1 is that this embodiment uses a drone swarm of 1 lead drone and 4 follower drones as an example to illustrate the method.

[0144] The directed leader-follower drone swarm topology diagram in this embodiment is as follows: Figure 2 As shown, the lead drone is numbered 0, and the follower drones are numbered 1, 2, 3, and 4. The arrows indicate the direction of information transmission. Lead drone 0 provides information to follower drone 1, follower drone 1 provides information to follower drones 2 and 3, and follower drone 3 provides information to follower drone 4. This topology satisfies the requirement that there are directed paths from the lead drone to each follower drone, but does not require undirected, balanced, or strongly connected paths. This means "following drone i receives information from following drone j". In this embodiment, the following is selected:

[0145]

[0146]

[0147] Furthermore, models of heterogeneous uncertain follower drones and unknown parameter pilot drones are established.

[0148] A simplified model is adopted, isomorphic to the second-order nonlinear channel commonly used in UAV attitude / position control. A heterogeneous, undamped, pendulum-type second-order nonlinear model is used for the follower UAV.

[0149]

[0150] in Therefore, the unknown parameters of each following drone The differences reflect the heterogeneity of UAV dynamics. The navigation UAV uses a reference model with unknown parameters:

[0151]

[0152]

[0153] The simulation step size is set to 0.005 s, and the simulation time is set to 20 s. (The following is a list of parameters / steps, not part of the simulation description): Figure 3 The initial conditions shown give the curves a clear initial difference while avoiding excessive control shocks.

[0154] Initialize the trigger broadcast variables and local estimators. At time t=0, the lead drone and all follower drones perform an initial broadcast, enabling... It is clearly defined. Afterwards, unless the dynamic event triggering conditions are met, each drone will no longer continuously transmit information. Each follower drone stores its own state parameters locally: .

[0155] Within each integration step, the drone first calculates using the most recently triggered broadcast value from its neighbors. Update according to the projective adaptive law Then update according to the distributed leader parameters and state observer. and These updates use only local variables, the most recently triggered broadcast value from a neighbor, and local parameters, meeting the requirements of a fully distributed implementation.

[0156] Furthermore, the compensation-type local adaptive cooperative control input is calculated, and the following UAV i is adjusted according to its own state. and local navigation status estimation Calculate sliding mode error The control input is calculated using a compensated control law. The regression term of the simple pendulum is taken as... The control law is denoted as follows in the simulation:

[0157]

[0158] The control law provides explicit compensation. In Item, make It can form a standard adaptive control structure.

[0159] Execute dynamic event-triggered communication, each following drone according to and The size relationship determines whether to trigger. If If the threshold is reached, the current status will be broadcast immediately. and The system will reset the local broadcast error; otherwise, it will remain silent. In this embodiment, the navigation drone also uses a similar triggering mechanism for broadcasting. and .

[0160] Furthermore, Figure 4The positions and trajectories of four follower drones are shown. It can be seen that, under the control method of this invention, the positions and velocities of all follower drones gradually track the trajectory of the lead drone.

[0161] Figure 5 The study demonstrates the navigation parameter estimation error, distributed leader state observation error, local tracking error, and overall leader-follower error. The navigation parameter estimation error gradually decreases, and the observation error and local tracking error converge in a cascade, indicating the effectiveness of the cascaded design of the network observation layer and the UAV physical tracking layer proposed in this application.

[0162] Figure 6 This indicates that the control inputs of each following drone are bounded and do not exhibit violent high-frequency oscillations; Figure 7 This indicates that the adaptive communication gains of the three channels p, q, and θ all remain within the projection range and are locally adjusted according to the error changes. Figure 8 The event trigger times for the lead drone and each follower drone are given. The triggers do not occur periodically, but are determined by the local error and dynamic threshold of each drone. Figure 9 Trigger statistics are provided, and the minimum trigger interval for all nodes is greater than 0. Simulations verify that there is no Zeno behavior. Figure 10 The triggering errors of follower drone 1 and follower drone 3 were further demonstrated. With dynamic threshold The relationship is as follows: Each time the error reaches the threshold, a broadcast is triggered and the error is reset. The threshold gradually decays over time, eventually causing the sampling error to disappear.

[0163] Furthermore, to verify the impact of dynamic event triggering parameters on the number of communications and convergence performance, this embodiment sets three sets of triggering parameters: tight threshold, nominal parameter, and large / slow threshold. Figure 11 and Figure 12 This indicates that: tight threshold communication is more frequent but converges faster; large / slow threshold communication significantly reduces the number of communication attempts but the error disappears more slowly; nominal parameters achieve a good trade-off between the number of communication attempts and tracking performance.

[0164] To illustrate the function of each technical module of the present invention, this embodiment also sets up a comparison of four schemes: the scheme of the present invention, the continuous communication scheme, the static triggering scheme, and the fixed local gain scheme. From Figure 13 and Figure 14 As can be seen, continuous communication achieves faster convergence, but it requires continuous information exchange, which is not suitable for UAV swarm scenarios with limited communication resources; the static triggering scheme triggers slightly more times and lacks dynamic threshold adjustment; although the fixed local gain scheme has fewer communication times, the error integral and adjustment time deteriorate significantly. The proposed solution achieves a good balance between tracking performance and communication without requiring continuous communication or global topology information.

[0165] To verify the adaptability of this invention to different directed leader-follower UAV topologies, chain, tree, directed loop, and star topologies were tested while keeping the control parameters basically consistent. Figure 15 and Figure 16 It can be seen that under these unbalanced and non-strongly connected directed topologies, as long as there is a directed path from the lead drone to each follower drone, the system can achieve a reduction in total tracking error and state convergence.

[0166] Finally, it should be noted that the above description only depicts some embodiments of this application. For those skilled in the art, various changes, modifications, substitutions, and variations can be conceived of these embodiments without departing from the principles and spirit of this application. The scope of protection of this application is defined by the appended claims and their equivalents, and all the above-mentioned behaviors should be covered within the scope of protection of this application.

Claims

1. A distributed dynamic event-triggered cooperative control method for drone swarms, applied to a drone swarm including one lead drone and several follower drones, characterized in that, include: S1. Establish a directed leader-follower drone topology model; S2. Establish state equations for the lead drone and the heterogeneous uncertain second-order nonlinear follower drone based on the drone topology model; the state equations require that there is at least one directed path from the lead drone to each follower drone; the state equations include the drone's position state, velocity state, and control input; S3. Set the trigger time, each follower drone estimates the state of the lead drone, maintains the estimated parameters of the lead drone, and broadcasts the estimated parameters of the local lead drone at the trigger time. The estimated parameters for the navigation drone include: navigation position estimation. Navigation speed estimation Navigation maneuver parameter estimation ; S4. The following drone constructs a local inconsistency quantity based on the difference between the navigation estimate most recently broadcast and maintained by the current node and the neighboring drone node, combined with the difference between the current node's broadcast maintained value and the latest broadcast value from the navigation end. S5. Construct the communication gain and update the communication gain of each channel using a projection-type local adaptive law; S6. Construct a state observer and use local inconsistency and communication gain to update the local maintenance of the pilot position estimate, pilot speed estimate, and pilot maneuver parameter estimate of the following UAV; the parameter observer is set at the pilot end and is used to obtain the pilot end estimate of the pilot maneuver parameter. S7. The following drone calculates adaptive compensation control input based on its own state and the estimated parameters of the local lead drone to achieve coordinated tracking of the lead drone's state.

2. The distributed dynamic event-triggered collaborative control method for UAV swarms according to claim 1, characterized in that, The state equations for the lead drone and the heterogeneous uncertain second-order nonlinear follower drone are established based on the drone topology model. A certain position / attitude channel of the i-th follower drone is represented as a heterogeneous uncertain second-order nonlinear system, as shown in the following formula: ； ； in, These represent the position, speed, and control input of the i-th following UAV, respectively. Indicates an unknown constant parameter. Indicates the first The nonlinear regression vector that follows the dynamics of the drone itself; The formula for piloting a drone is: ； ； in, This indicates unknown navigation maneuver parameters. This represents the navigation regression vector.

3. The distributed dynamic event-triggered cooperative control method for UAV swarms according to claim 1, characterized in that, The setting of the trigger time, constructing broadcast conditions based on event metrics, determining whether to broadcast the local estimated value according to the dynamic event trigger conditions, and remaining silent in non-trigger intervals, includes: Step 1: Define local broadcast error; Step 2: Construct event measures based on local broadcast errors using the drone. ; Step 3: Set the fading envelope and bounded dynamic scale ; Step 4: Define the broadcast trigger time based on the fading envelope and bounded dynamic scale, using the following formula: ； When the trigger condition is met, follow the drone's broadcast of the current moment. And reset the local broadcast error.

4. The distributed dynamic event-triggered cooperative control method for UAV swarms according to claim 1, characterized in that, The local lead drone estimation parameters are broadcast at the trigger time, and each follower drone i maintains three types of local estimation variables: lead position estimation... Navigation speed estimation Navigation maneuver parameter estimation Node i broadcasts at its own trigger time. The neighboring node uses the most recently received value between two triggers. .

5. The distributed dynamic event-triggered cooperative control method for UAV swarms according to claim 1, characterized in that, The construction of communication gain involves setting a local communication gain for each follower drone. The gain is driven by the local error, and the communication gain is limited to a given local range by the projection operator. The formula is: ； ； ； in, The update rate represents the local communication gain. This represents the projection operator, used to maintain a positive lower bound and a finite upper bound on the gain. Represents the vector projection function. Indicates communication gain The projection range, Let represent the communication gain update rate coefficient of the i-th following UAV on the ℓ-th channel, where ℓ∈{p,q,θ}, used to adjust the communication gain on the error driving term. Response speed The error-driven term is used to adjust the coupling strength according to the degree of local inconsistency, while the fading suppression term reduces gain drift during small error phases. Indicates a quantity of local inconsistency. This represents the fading suppression term, used to reduce communication gain drift during periods of small errors. This represents the initial strength of the suppression term. This indicates the decay rate of the suppression term.

6. The distributed dynamic event-triggered cooperative control method for UAV swarms according to claim 1, characterized in that, The method employs local inconsistency and communication gain to update the locally maintained pilot position estimate, pilot speed estimate, and pilot maneuver parameter estimate of the following UAV, as shown in the following formula: ； ； ； in, , , These represent the estimated parameters of the navigation maneuver. Navigation position estimation Navigation speed estimation The derivative of , , Indicates the corresponding communication gain. , , This represents the corresponding locally inconsistent variable. This represents the transpose of the estimated navigation maneuver parameters. This represents the navigation regression vector.

7. The distributed dynamic event-triggered cooperative control method for UAV swarms according to claim 1, characterized in that, The following drone calculates adaptive compensation control input based on its own state and estimated parameters from the local lead drone. The control law of the following drone is expressed in a compensation form, and the formula is as follows: ； in, This indicates the control law for compensated unmanned aerial vehicles (UAVs). This represents an estimate of the unknown parameters of the drone itself. Indicates the first A nonlinear regression vector of the following UAV's own dynamics is used to describe and compensate for the dynamic uncertainties of the following UAV. Indicates local pilot speed estimate Time derivative, Represents the error combination coefficient. Indicates speed tracking error. Indicates communication gain. Representing variables with local inconsistencies, Indicates the combined error feedback gain. This represents the combined error consisting of position tracking error and velocity tracking error.

8. The distributed dynamic event-triggered cooperative control method for UAV swarms according to claim 7, characterized in that, The formula is calculated in real time by the distributed leader velocity observer: ； in, Represents the navigation regression vector. Indicates the local communication gain following the speed of the drone. This indicates the local inconsistency in the speed of the drone.

9. The distributed dynamic event-triggered cooperative control method for UAV swarms according to claim 7, characterized in that, An adaptive law is applied based on the unknown parameters of the drone itself, and the formula is: ； ； in, This indicates the update rate of the estimated values ​​of the unknown parameters of the drone itself. Indicates the first The adaptive gain positive definite matrix of the drone following the aircraft. This represents the estimation error of the drone's own dynamic parameters. Indicates the first The unknown constant parameters of the drone itself.

Images