Nonlinear multi-unmanned vehicle system self-triggering time-varying formation optimization control method

Abstract

The invention relates to a non-linear multi-unmanned vehicle system self-triggering time-varying formation control method, which comprises the following steps: constructing a second-order dynamic model for a non-linear multi-unmanned vehicle system, and describing communication between unmanned systems through a directed topological structure; a backstepping method is adopted, a virtual controller is introduced, and a leader-follower formation error system under the triggering condition is constructed; designing influence weights of communication resources and control performance, and respectively constructing an optimization model based on virtual input and formation distributed position errors and an optimization model based on real input and tracking formation speed errors; solving an approximate solution of a corresponding Hamiltonian-Jacobi-Bellman equation through a simplified reinforcement learning method, and sequentially obtaining an optimal virtual control protocol and an optimal reality control protocol; similarly, determining a self-triggering prediction condition; and finally, under the condition of self-triggering prediction, the multi-unmanned vehicle system not only realizes time-varying formation, but also balances communication resources and control performance.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned vehicle formation, and in particular to a self-triggered time-varying formation optimization control method for a nonlinear multi-unmanned vehicle system. Background Art

[0002] Heterogeneous unmanned systems can accurately encompass diverse swarms, including drones, unmanned vehicles, and land robots. In existing heterogeneous unmanned vehicle systems, control methods based on homogeneous output information are commonly used to achieve platoon formation. However, this homogeneous output control method often suffers from poor performance when applied to intelligent agents with complex dynamics due to the uncertainty of the output matrix.

[0003] Chinese patent application publication number CN116610119A discloses a method for controlling an unmanned vehicle formation based on a new time-varying spacing strategy. By adopting the new time-varying spacing strategy and sliding mode function in the unmanned vehicle formation, combined with a neural network, the method solves the instability problem caused by the initial spacing error in the traditional time-varying spacing strategy and achieves stable control of the unmanned vehicle formation. However, it does not take into account the limitations of communication resources in large-scale formation mission scenarios.

[0004] In a system's communication network, a fixed topology is common. However, as communication networks scale and dynamically change, fixed topologies often fail to effectively address challenges such as load fluctuations and fault recovery. Regarding system states, observers used in heterogeneous unmanned systems are mostly applicable only to linear systems or systems subject to linear constraints. These observers struggle to observe systems with uncertain nonlinear characteristics. For large-scale formation missions, relying solely on distributed formation control results in high energy consumption and communication resource usage.

[0005] Therefore, in the formation control process, optimal control that simultaneously reduces energy consumption and communication resource consumption has become an urgent problem to be solved. Summary of the Invention

[0006] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and to provide a nonlinear multi-unmanned vehicle system self-triggered time-varying formation optimization control method to solve or partially solve the contradiction between communication resource utilization and control performance, and the problem that a good balance between the two cannot be achieved.

[0007] The purpose of the present invention can be achieved by the following technical solutions:

[0008] The present invention provides a self-triggered time-varying formation optimization control method for a nonlinear multi-unmanned vehicle system, which is applied to a nonlinear unmanned vehicle system including a leading unmanned vehicle and multiple following unmanned vehicles. The method includes the following steps:

[0009] Execute formation tasks based on preset self-trigger optimization strategies;

[0010] During the execution of the formation mission, based on the position and speed information of each unmanned vehicle and the real control input information based on virtual control, the distributed position error of each following unmanned vehicle and the speed error of the following formation are obtained, and the next triggering moment is predicted in combination with the optimal self-triggering condition in the self-triggering optimization strategy;

[0011] In response to reaching the next departure time, updating the actual control input information of each unmanned vehicle and the optimal actual control input information of the actual controller;

[0012] The controller following the unmanned vehicle performs optimal real-world control based on virtual control, updates the control input moment through the predicted self-triggering conditions, and controls the unmanned vehicle. Based on the acquired virtual and real-world control input information of each unmanned vehicle, the weights of communication resources and control performance are calculated and updated through the weight prediction function, and the self-triggering optimization strategy is updated.

[0013] Based on the updated self-triggered optimization strategy, the formation task continues to be executed, completing the self-triggered time-varying formation task of the nonlinear multi-unmanned vehicle system.

[0014] As a preferred technical solution, based on the virtual and real control input information obtained from each unmanned vehicle, the weights of communication resources and control performance are calculated and updated through a weight prediction function. The process of obtaining the updated weights includes the following steps:

[0015] According to the formation task, a directed communication topology structure of a nonlinear multi-unmanned vehicle system described by graph theory is constructed;

[0016] Construct a nonlinear dynamic model of the multi-unmanned vehicle system based on the optimal control input information of each unmanned vehicle's real controller, the position information of the unmanned vehicle, and the speed information of the unmanned vehicle;

[0017] According to the dynamic model, the state information of the position and speed of each unmanned vehicle is obtained, and the formation offset is used to calculate the following formation position error of the nonlinear multi-unmanned vehicle system, and a virtual control is set on the following formation speed error to obtain the following formation position error and the following formation speed error under triggering. According to the following formation position error of the nonlinear multi-unmanned vehicle system, the distributed position error of the formation under triggering of the nonlinear multi-unmanned vehicle system is obtained in combination with the directed communication topology structure;

[0018] According to the formation distributed position error and following formation speed error obtained in the triggering situation, the weights of communication resources and control performance are calculated and updated through the weight prediction function.

[0019] As a preferred technical solution, the dynamic model of the following unmanned vehicle cluster is modeled as follows:

[0020]

[0021] in, is the position and speed state of the i-th unmanned vehicle, is the acceleration control input of the i-th unmanned vehicle, f i (x i,1 ,t) and f i (x i,1 ,x i,2 ,t) represents a continuous bounded nonlinear function and satisfies is a positive constant, n is the dimension of the vector,

[0022] The dynamic model of the pilot unmanned vehicle is:

[0023]

[0024] in, is the dynamic state of the pilot unmanned vehicle x0(t), f0(x 0,1 ,x 0,2 ,t) is expressed as a nonlinear function embedded in the dynamic equation of the pilot unmanned vehicle.

[0025] As a preferred technical solution, the following formation position error and following formation speed error when the nonlinear multi-unmanned vehicle system is triggered are:

[0026]

[0027] in, are the tracking formation position and speed errors of the i-th unmanned vehicle, h i is the preset formation offset vector and h i , It is bounded. h i The derivative of the trigger signal Indicates following the unmanned vehicle m i ,i∈{1,2,…,M} in k i The state within the cycle, M is the number of following unmanned vehicles, is the position and speed state of the i-th unmanned vehicle, n is the dimension of the vector, α i For virtual controllers,

[0028] The distributed position error of the formation under the triggering condition of the nonlinear multi-unmanned vehicle system is:

[0029]

[0030] in, is the distributed formation position error of the i-th unmanned vehicle, a ij is the adjacency matrix weight of the directed communication topology, indicating whether there is a communication connection between the i-th unmanned vehicle and the j-th unmanned vehicle.

[0031] As a preferred technical solution, the weight prediction function is:

[0032]

[0033] in, They are the i-th unmanned vehicle at k i The weights of communication resources and control performance within the cycle and satisfy They represent the partial derivatives of communication resources and formation performance under slight perturbations respectively.

[0034] As a preferred technical solution, the process of updating the self-triggered optimization strategy includes the following steps:

[0035] According to k i The weights of communication resources and control performance at the moment and k i -1 time period self-triggering interval, combined with the formation distributed position error under the triggering of multiple unmanned vehicle systems, to build an optimization model based on virtual input and formation distributed position error;

[0036] A virtual control protocol is obtained through reinforcement learning based on the Hamilton-Jacobi-Bellman equation based on virtual inputs and the formation's distributed position errors;

[0037] According to the virtual control protocol, the speed error of the following formation is updated under triggering conditions, and an optimization model based on the real input and the speed error of the following formation is constructed;

[0038] A realistic control protocol is obtained through reinforcement learning using the Hamilton-Jacobi-Bellman equation based on realistic inputs and velocity errors of the following formation;

[0039] According to k i The Hamilton-Jacobi-Bellman equation at the triggering moment is used to obtain the self-triggering prediction condition to predict the triggering interval and realize the automatic prediction triggering in the self-triggering strategy.

[0040] As a preferred technical solution, the optimization model of the virtual input and formation distributed position error is modeled as follows:

[0041]

[0042] in, is an implicit function, and Represents k i-1 time period self-triggering interval and optimal virtual controller, represents the optimization objective function at the position level of the i-th unmanned vehicle, Represents the formation distributed position error.

[0043] As a preferred technical solution, the optimization model based on real input and following formation speed error is modeled as follows:

[0044]

[0045] in, is an implicit function, represents the optimal real controller based on virtual control, represents the optimization objective function at the speed level of the i-th unmanned vehicle, and u i Represents k i -1 time period self-triggering interval and optimal true controller, r i,2 is the speed error of following formation, u i Enter information for acceleration.

[0046] As a preferred technical solution, the Hamilton-Jacobi-Bellman equation includes the Hamilton-Jacobi-Bellman equation based on virtual input and formation distributed position error, the Hamilton-Jacobi-Bellman equation based on real input and tracking formation velocity error, and the k i Hamilton-Jacobi-Bellman equation at the triggering moment.

[0047] As a preferred technical solution, the self-triggering strategy includes a virtual control protocol, a real control protocol and a self-triggering prediction condition.

[0048] Compared with the prior art, the present invention has at least one of the following beneficial effects:

[0049] (1) Realizing formations in complex environments: The present invention addresses the nonlinear uncertainty of unmanned system clusters and realizes control of unmanned vehicles by updating the control input moments based on predicted self-triggering conditions. Based on the virtual and real control input information obtained from each unmanned vehicle, the weights of communication resources and control performance are calculated and updated through a weight prediction function. The self-triggering optimization strategy is updated, and a trade-off is first made between the formation's distributed position error and the virtual input, and then a trade-off is made between the formation's tracking speed error and the real input, thereby achieving an optimal balance between the control performance and communication resource consumption of multiple unmanned vehicle systems.

[0050] (2) Dual-objective optimal control of communication resource utilization and control performance is achieved: The present invention embeds the influence weights of communication and performance into the objective function, and then derives the prediction function of the weights through small disturbances. The influence function of the weights is derived from the small disturbances, so that the weights are dynamically adjusted according to the performance and communication conditions during the formation process, thereby achieving quantitative optimization of communication and performance. The present invention considers the dual-objective optimal problem of communication resource utilization and control performance, rather than simply pursuing the optimal solution of one of the objectives. It proposes a self-triggering prediction condition that balances communication resources and control performance, and establishes a novel self-triggering strategy. The trade-off strategy embedded in the self-triggering control can avoid high energy consumption caused by the reduction of communication frequency, can effectively balance the contradiction between communication resource utilization and control performance, and at the same time ensure the stability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 Flowchart of the self-triggered time-varying formation optimization control method for a nonlinear multi-unmanned vehicle system in an embodiment;

[0052] Figure 2 : This is a directed communication topology diagram of a heterogeneous multi-unmanned vehicle system in an embodiment;

[0053] Figure 3 Schematic diagram of the state of a multi-unmanned vehicle system in an embodiment;

[0054] Figure 4 This is a formation diagram of the unmanned vehicle system in the embodiment;

[0055] Figure 5 This is an impact function diagram for the unmanned vehicle system in the embodiment to balance communication resources and control performance;

[0056] Figure 6 A trigger diagram of communication of a multi-unmanned vehicle system in an embodiment;

[0057] Figure 7 Schematic diagram of the control performance of the multi-unmanned vehicle system in the embodiment;

[0058] Figure 8 Schematic diagram of distributed position error of formation in the embodiment;

[0059] Figure 9 This is a formation tracking error diagram in the embodiment;

[0060] Figure 10 Schematic diagram of an electronic device in an embodiment. DETAILED DESCRIPTION

[0061] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0062] Example 1

[0063] To address the problems of the aforementioned prior art, this embodiment provides a nonlinear multi-unmanned vehicle system self-triggered time-varying formation optimization control method, which mainly includes the following steps:

[0064] 1. The nonlinear multi-unmanned vehicle system performs platooning tasks according to a preset self-triggering optimization strategy;

[0065] 2. During the formation process, based on the position and speed information of each unmanned vehicle and the real control input information based on virtual control, the distributed error position information and formation error speed information of each following unmanned vehicle formation are obtained, and the optimal self-trigger condition in the self-trigger optimization strategy is combined to predict the next trigger time;

[0066] 3. At the next trigger moment, obtain the virtual control input information of each unmanned vehicle and the optimal control input information of the actual controller in real time;

[0067] 4. Execute the optimal real-world control protocol based on virtual control on the controller following the unmanned vehicle, and update the control input time through self-triggered prediction conditions to achieve control of the unmanned vehicle. Based on the virtual and real control input information obtained from each unmanned vehicle, the weight prediction function is used to calculate and update the weights of communication resources and control performance. The updated weights are used to adjust the self-triggered optimization strategy.

[0068] 5. Execute formation tasks according to the updated self-trigger optimization strategy;

[0069] 6. Repeat the above process to realize the self-triggered time-varying formation task of nonlinear multi-unmanned vehicle system.

[0070] Among them, see Figure 1 ,The preset self-triggering strategy acquisition process includes the following steps:

[0071] Step S1: Construct a dynamic model of a nonlinear multi-unmanned vehicle system. The heterogeneous nonlinear multi-unmanned vehicle system includes a cluster of M follower unmanned vehicles and a leading unmanned vehicle. The state information refers to the motion state of the vehicle at a certain moment, such as position, speed, acceleration, heading angle, etc.

[0072] The dynamic model of the following unmanned vehicle cluster is:

[0073]

[0074] in, and are the system state and control input respectively. In addition, f i (x i,1 ,t) and f i (x i,1 ,x i,2 ,t) represents a continuous bounded nonlinear function and satisfies

[0075] The dynamic model of the pilot unmanned vehicle is:

[0076]

[0077] in, is the dynamics of the pilot autonomous vehicle state x0(t), f0(x 0,1 ,x 0,2 ,t) is expressed as a nonlinear function embedded in the dynamic equation of the pilot unmanned vehicle;

[0078] Step S2, based on graph theory, describes the directed communication topology of the nonlinear multi-unmanned vehicle system, obtains the state information of the position and speed of each unmanned vehicle and the formation offset according to the dynamic model, and obtains the following formation position error of the unmanned vehicle system. Sets virtual control on the following formation speed error, that is, the speed error in the following formation. After adjustment of the virtual control amount, the purpose is to adjust the speed of the intelligent agent i so that it can effectively follow the leader and eventually make this speed error approach zero. Through such configuration, the following formation position error and the following formation speed error under triggering conditions are obtained. According to the following formation position error of the unmanned vehicle system and the directed communication topology structure information, the distributed position error of the formation under triggering conditions of the multi-unmanned vehicle system is obtained.

[0079] This step aims to address the position and velocity formation problem in nonlinear unmanned vehicle systems, achieving efficient and stable formation following control through backstepping. The introduction of virtual control effectively manages and compensates for formation errors caused by various factors, such as sensor noise, environmental interference, and communication delays. Through virtual control, the system can compensate for measurement errors in real time, reducing position and velocity errors between formation members and ensuring the formation follows the intended trajectory.

[0080] Furthermore, since platooning control typically involves extensive communication and computing resources, directly adopting centralized control can increase system overhead and potentially reduce real-time performance. Therefore, this step introduces a distributed control approach, calculating the platooning estimation error based on a neighboring communication topology. This allows each platooning vehicle to complete estimation and control solely based on information from its local and neighboring vehicles, thereby improving system scalability and computational efficiency.

[0081] Furthermore, to improve computational and communication efficiency, this step introduces a self-triggering control mechanism, rather than traditional periodic communication. Specifically, this triggering mechanism ensures that the autonomous vehicle updates information only when necessary, rather than continuously exchanging data. This reduces unnecessary communication overhead and improves system energy efficiency.

[0082] The position error and speed error of the following formation when the multi-unmanned vehicle system is triggered are:

[0083]

[0084] in, are the tracking formation position and speed errors of the i-th unmanned vehicle, h i is the preset formation offset vector and h i , is bounded, triggering the signal Indicates following the unmanned vehicle m i ,i∈{1,2,…,M} in k i Status within the cycle;

[0085] The distributed position error of the formation when multiple unmanned vehicle systems are triggered is:

[0086]

[0087] in, is the distributed formation position error of the i-th unmanned vehicle, a ij is the adjacency matrix weight of the directed communication topology, indicating whether there is a communication connection between the i-th unmanned vehicle and the j-th unmanned vehicle;

[0088] Step S3, based on the formation distributed position error and formation speed error obtained under the triggering condition, calculate and update the weights of communication resources and control performance through the weight prediction function. i The weights of communication resources and control performance at the moment and k i-1 time period, combined with the distributed position error of the formation under the triggering condition of multiple unmanned vehicle systems, an optimization model based on virtual input and formation distributed position error is constructed; through the Hamilton-Jacobi-Bellman equation based on virtual input and formation distributed position error, a virtual control protocol is obtained using a simplified reinforcement learning method.

[0089] The weight prediction function is:

[0090]

[0091] in, represents the comprehensive formation error following unmanned vehicle i, represents the control input of the following unmanned vehicle i, ∈ represents a fixed parameter, ι represents a small perturbation, They represent the partial derivatives of communication resources and formation performance under slight disturbances, Q = blkdiag {Q i,j},R=blkdiag{R i,j},j=1,2,Q i,j ,R i,j are positive definite matrices, are the weights of communication resources and control performance respectively and satisfy

[0092] Step S4: According to the trigger time, the first trigger interval is obtained, and the weight prediction function of the communication resources and control performance is designed according to the distributed formation estimation error under the triggering condition. The weights of the communication resources and control performance are set by the weight prediction function. The distributed cross-dimensional formation error and the first trigger interval are combined to establish the objective function and construct the optimization model. Similarly, the Hamilton-Jacobi-Bellman equation based on the real input and the tracking formation speed error is used to obtain the real control protocol using the simplified reinforcement learning method.

[0093] The core purpose of steps S3 and S4 is to balance and optimize the communication resource allocation and control performance of the unmanned vehicle system by designing a weighted prediction function for communication resources and control performance based on the previous trigger interval. Using backstepping, they first explored the tradeoff between the formation's distributed position error and virtual input, and then the tradeoff between the formation's tracking velocity error and real input, thereby achieving an optimal balance between control performance and communication resource consumption for the multi-unmanned vehicle system.

[0094] First, based on the system's integration of formation errors and virtual reality inputs, combined with the trigger interval, a weight prediction function is designed using a micro-perturbation method to dynamically balance control performance and communication resource consumption. This approach prevents over-optimization of a single objective at the expense of another, ensuring efficient operation.

[0095] Secondly, step S3 establishes an optimization model based on virtual input and formation distributed position error to achieve the minimization of the objective function involving comprehensive consideration of the distributed position error of each unmanned vehicle formation and the virtual control strategy, and uses a simplified reinforcement learning method to solve the approximate solution of the Hamilton-Jacobi-Bellman equation based on virtual input and formation distributed position error, and further obtains the optimal virtual self-triggered control protocol.

[0096] Finally, the optimal virtual input obtained in step S3 influences the formation tracking speed error. Step S4 establishes an optimization model based on the real input and the formation tracking speed error, minimizing the objective function that comprehensively considers the tracking speed errors of each unmanned vehicle formation and the actual control strategy, thereby achieving optimized communication and control. A simplified reinforcement learning method is used to approximate the Hamilton-Jacobi-Bellman equation based on the real input and the formation tracking speed error, further obtaining the optimal real-world self-triggering control protocol and self-triggering prediction conditions.

[0097] Specifically, according to the updated weights of communication resources and control performance, the self-triggering strategy is adjusted according to k i The weights of communication resources and control performance at the moment and k i -1 time period, combined with the distributed position error of the formation under the triggering condition of multiple unmanned vehicle systems, an optimization model based on virtual input and distributed position error of the formation is constructed, and the virtual control protocol is obtained by using the Hamilton-Jacobi-Bellman equation based on virtual input and distributed position error of the formation; according to the virtual control protocol, the tracking speed error of the formation under the triggering condition is updated, and an optimization model based on real input and tracking speed error of the formation is constructed. Similarly, the real control protocol is obtained by using the Hamilton-Jacobi-Bellman equation based on real input and tracking speed error of the formation; according to the k i The Hamilton-Jacobi-Bellman equation at the trigger moment is used to obtain the self-triggering prediction condition to predict the trigger interval, thus realizing the automatic predictive triggering in the self-triggering strategy. The control protocol is derived using reinforcement learning (i.e., Actor-Critic Neural Network approximation) using existing technology.

[0098] The optimization model based on virtual input and formation distributed position error is:

[0099]

[0100] in, is an implicit function, and Represents k i -1 time period self-triggering interval and optimal virtual controller, represents the optimization objective function at the position level of the i-th unmanned vehicle;

[0101] The optimization model based on realistic input and tracking formation velocity error is:

[0102]

[0103] in, is an implicit function, represents the optimal real controller based on virtual control, Represents the optimization objective function at the speed level of the i-th unmanned vehicle.

[0104] The Hamilton-Jacobi-Bellman equations include the Hamilton-Jacobi-Bellman equation based on virtual input and formation distributed position error, the Hamilton-Jacobi-Bellman equation based on real input and tracking formation velocity error, and the Hamilton-Jacobi-Bellman equation at the triggering time of k_i;

[0105] Hamilton-Jacobi-Bellman equation based on virtual input and formation distributed position error:

[0106]

[0107] in parameter Needs to be satisfied is the reward function.

[0108] Hamilton-Jacobi-Bellman equation based on real input and tracking formation velocity error:

[0109]

[0110] in parameter Is a positive definite matrix and needs to satisfy is the reward function.

[0111] k i The Hamilton-Jacobi-Bellman equation at the triggering moment:

[0112]

[0113] The self-triggering strategy includes virtual control protocols, real control protocols, and self-triggering prediction conditions;

[0114] The virtual control protocol is:

[0115]

[0116] in, as well as Represents the ideal weights in the neural network Estimates.

[0117] The actual control protocol is:

[0118]

[0119] in, as well as Represents the ideal weights in the critic-action neural network Estimates.

[0120] Self-trigger prediction conditions:

[0121]

[0122] Step S5, according to the optimization model, solve the problem based on k by simplifying the reinforcement learning method i The approximate solution of the Hamilton-Jacobi-Bellman equation at the triggering moment is used to obtain the self-triggering prediction condition to predict the triggering interval, thereby realizing automatic predictive triggering in the self-triggering strategy; the self-triggering strategy is formed by combining the optimal virtual control input and the optimal real control input.

[0123] In order to better illustrate the effectiveness of the time-varying cross-dimensional formation control method for heterogeneous nonlinear multi-unmanned systems described in the present invention, the method proposed in the present invention is simulated and verified, and the details are as follows:

[0124] In the simulation, it is assumed that the unmanned system cluster consists of a leading unmanned vehicle (node ​​0) and five following unmanned vehicles (nodes 1 to 5). Its directed communication topology is as follows: Figure 2 shown.

[0125] The dynamics of the unmanned vehicle satisfies the above dynamics formula, where the nonlinear function is:

[0126] f i (x i.1 ,t)=0.05[sin(x i.1 (1)),cos(x i.1 (2)))],

[0127]

[0128] x i,1 =[x i,1 (1),x i,1 (2)] T ,

[0129] x i,2 =[x i,2 (1),x i,2 (2)]T

[0130] Set the state offset between the following unmanned vehicle and the leading unmanned vehicle to The remaining parameters are selected as

[0131] Running MATLAB software at a computing frequency of 1 GHz, we can obtain the following results. Figure 3 A schematic diagram showing the status of a multi-unmanned vehicle system. Figure 4 A diagram showing the formation of a multi-unmanned vehicle system, showing a pentagon and within the range of 1. Figure 5 Shows the impact function diagram of the multi-unmanned vehicle system in terms of communication resources and control performance. tends to 1 and Approaching 0 is consistent with minimizing the objective function. Figure 6 A trigger diagram showing the communication of a multi-route autonomous vehicle system. Figure 7 A schematic diagram showing control performance is shown, where the shaded area of ​​each agent represents the performance indicator. Figure 8 The distributed position error graph of the formation is displayed. It can be seen from the graph that the position error converges to a bounded interval. Figure 9 The formation tracking velocity error graph is displayed, and it can be seen from the graph that the velocity error converges to a bounded interval.

[0132] This method has the following beneficial effects:

[0133] (1) This method considers the nonlinear uncertainties inherent in unmanned system swarms and employs backstepping to ensure the stability of the control system. Furthermore, the control strategy is dynamically adjusted through RL to accurately compensate for nonlinearities. The designed algorithm can be applied to individuals with uncertain states and can accomplish formation tasks in complex environments.

[0134] (2) This method embeds the influence weights of communication and performance into the objective function, and then derives a prediction function of the weights through small perturbations. The influence function of the weights is derived from the small perturbations, thereby dynamically adjusting the weights based on the performance and communication conditions during the formation process, thereby achieving quantitative optimization of communication and performance. Compared with the existing technology, this method has the advantages of wider coverage, higher security, and long-term operation.

[0135] (3) This method explores the dual-objective optimization problem of communication resource utilization and control performance, rather than simply pursuing the optimal solution of one of the objectives. It proposes a self-triggering prediction condition that balances communication resources and control performance, and establishes a novel self-triggering strategy. The trade-off strategy embedded in the self-triggering control can avoid the high energy consumption caused by the reduction of communication frequency, effectively balance the contradiction between communication resource utilization and control performance, and ensure the stability of the system.

[0136] Example 2

[0137] Based on Example 1, this embodiment provides an electronic device, including: one or more processors and a memory, wherein the memory stores one or more programs, and the one or more programs include instructions for executing the self-triggered time-varying formation optimization control method of the nonlinear multi-unmanned vehicle system as described in Example 1.

[0138] like Figure 10 As mentioned above, at the hardware level, the electronic device includes a processor, an internal bus, a network interface, a memory and a non-volatile memory, and may also include other hardware required for the business. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it to achieve the above Figure 1 Of course, in addition to software implementation, the present invention does not exclude other implementation methods, such as logic devices or a combination of software and hardware, etc., that is, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.

[0139] Memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0140] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory media such as modulated data signals and carrier waves.

[0141] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and such modifications or substitutions are intended to be within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.

Claims

1. A nonlinear multi-unmanned vehicle system self-triggered time-varying formation optimization control method, characterized by: Applied to a nonlinear unmanned vehicle system including a pilot unmanned vehicle and multiple following unmanned vehicles, the method includes the following steps: Execute formation tasks based on preset self-trigger optimization strategies; During the execution of the formation mission, based on the position and speed information of each unmanned vehicle and the real control input information based on virtual control, the distributed position error of each following unmanned vehicle and the speed error of the following formation are obtained, and the next triggering moment is predicted in combination with the optimal self-triggering condition in the self-triggering optimization strategy; In response to reaching the next departure time, updating the actual control input information of each unmanned vehicle and the optimal actual control input information of the actual controller; The controller following the unmanned vehicle performs optimal real-world control based on virtual control, updates the control input moment through the predicted self-triggering conditions, and controls the unmanned vehicle. Based on the acquired virtual and real-world control input information of each unmanned vehicle, the weights of communication resources and control performance are calculated and updated through the weight prediction function, and the self-triggering optimization strategy is updated. Based on the updated self-triggered optimization strategy, the formation task continues to be executed, completing the self-triggered time-varying formation task of the nonlinear multi-unmanned vehicle system.

2. The method for optimizing the self-triggered time-varying formation control of a nonlinear multi-unmanned vehicle system according to claim 1, characterized in that: Based on the virtual and real control input information obtained for each unmanned vehicle, the weights of communication resources and control performance are calculated and updated through the weight prediction function. The process of obtaining the updated weights includes the following steps: According to the formation task, a directed communication topology structure of a nonlinear multi-unmanned vehicle system described by graph theory is constructed; Construct a nonlinear dynamic model of the multi-unmanned vehicle system based on the optimal control input information of each unmanned vehicle's real controller, the position information of the unmanned vehicle, and the speed information of the unmanned vehicle; According to the dynamic model, the state information of the position and speed of each unmanned vehicle is obtained, and the formation offset is used to calculate the following formation position error of the nonlinear multi-unmanned vehicle system, and a virtual control is set on the following formation speed error to obtain the following formation position error and the following formation speed error under triggering. According to the following formation position error of the nonlinear multi-unmanned vehicle system, the distributed position error of the formation under triggering of the nonlinear multi-unmanned vehicle system is obtained in combination with the directed communication topology structure; According to the formation distributed position error and following formation speed error obtained in the triggering situation, the weights of communication resources and control performance are calculated and updated through the weight prediction function.

3. The method for self-triggered time-varying formation optimization control of a nonlinear multi-unmanned vehicle system according to claim 2, characterized in that: The dynamic model of the following unmanned vehicle cluster is modeled as follows: in, is the position and speed state of the i-th unmanned vehicle, is the acceleration control input of the i-th unmanned vehicle, f i (x i,1 ,t) and f i (x i,1 ,x i,2 ,t) represents a continuous bounded nonlinear function and satisfies is a positive constant, n is the dimension of the vector, The dynamic model of the pilot unmanned vehicle is: in, is the dynamic state of the pilot unmanned vehicle x0(t), f0(x 0,1 ,x 0,2 ,t) is expressed as a nonlinear function embedded in the dynamic equation of the pilot unmanned vehicle.

4. The method for optimizing and controlling a nonlinear multi-unmanned vehicle system self-triggered time-varying formation according to claim 2, characterized in that: The following formation position error and following formation speed error when the nonlinear multi-unmanned vehicle system is triggered are: in, are the tracking formation position and speed errors of the i-th unmanned vehicle, h i is the preset formation offset vector and h i , It is bounded. h i The derivative of the trigger signal Indicates following the unmanned vehicle m i ,i∈{1,2,…,M} in k i The state within the cycle, M is the number of following unmanned vehicles, is the position and speed state of the i-th unmanned vehicle, n is the dimension of the vector, α i For virtual controllers, The distributed position error of the formation under the triggering condition of the nonlinear multi-unmanned vehicle system is: in, is the distributed formation position error of the i-th unmanned vehicle, a ij is the adjacency matrix weight of the directed communication topology, indicating whether there is a communication connection between the i-th unmanned vehicle and the j-th unmanned vehicle.

5. The method for self-triggered time-varying formation optimization control of a nonlinear multi-unmanned vehicle system according to claim 2, characterized in that: The weight prediction function is: in, They are the i-th unmanned vehicle at k i The weight of communication resources and control performance within the cycle and satisfy They represent the partial derivatives of communication resources and formation performance under slight perturbations respectively.

6. The method for optimizing and controlling a nonlinear multi-unmanned vehicle system self-triggered time-varying formation according to claim 1, characterized in that: The process of updating the self-triggered optimization strategy includes the following steps: According to k i The weights of communication resources and control performance at the moment and k i -1 time period self-triggering interval, combined with the formation distributed position error under the triggering of multiple unmanned vehicle systems, to build an optimization model based on virtual input and formation distributed position error; A virtual control protocol is obtained through reinforcement learning based on the Hamilton-Jacobi-Bellman equation based on virtual inputs and the formation's distributed position errors; According to the virtual control protocol, the speed error of the following formation is updated under triggering conditions, and an optimization model based on the real input and the speed error of the following formation is constructed; A realistic control protocol is obtained through reinforcement learning using the Hamilton-Jacobi-Bellman equation based on realistic inputs and velocity errors of the following formation; According to k i The Hamilton-Jacobi-Bellman equation at the triggering moment is used to obtain the self-triggering prediction condition to predict the triggering interval and realize the automatic prediction triggering in the self-triggering strategy.

7. The method for optimizing and controlling a nonlinear multi-unmanned vehicle system self-triggered time-varying formation according to claim 6, characterized in that: The optimization model of the virtual input and formation distributed position error is modeled as: in, is an implicit function, and Represents k i -1 time period self-triggering interval and optimal virtual controller, represents the optimization objective function at the position level of the i-th unmanned vehicle, Represents the formation distributed position error.

8. The method for self-triggered time-varying formation optimization control of a nonlinear multi-unmanned vehicle system according to claim 6, characterized in that: The optimization model based on the actual input and the speed error of the following formation is modeled as: in, is an implicit function, represents the optimal real controller based on virtual control, represents the optimization objective function at the speed level of the i-th unmanned vehicle, and u i Represents k i -1 time period self-triggering interval and optimal true controller, r i,2 is the speed error of following formation, u i Enter information for acceleration.

9. The method for optimizing and controlling a nonlinear multi-unmanned vehicle system self-triggered time-varying formation according to claim 6, characterized in that: The Hamilton-Jacobi-Bellman equation includes the Hamilton-Jacobi-Bellman equation based on virtual input and formation distributed position error, the Hamilton-Jacobi-Bellman equation based on real input and tracking formation velocity error, and the k i Hamilton-Jacobi-Bellman equation at the triggering moment.

10. The method for self-triggered time-varying formation optimization control of a nonlinear multi-unmanned vehicle system according to claim 6, characterized in that: The self-triggering strategy includes a virtual control protocol, a real control protocol and a self-triggering prediction condition.

Images