A collaborative control method for heterogeneous unmanned system clusters based on event triggering

Abstract

The present invention discloses a method for cluster collaborative control of heterogeneous unmanned systems based on event triggering. Cluster control of unmanned systems is a typical application of the coordinated control of the consistency of multi-agent systems. Therefore, this patent uses a multi-agent system for specific description. Taking into account the cooperative and competitive interactive relationships between agents, a new cluster consistency protocol is designed, and a fully distributed event triggering condition that does not rely on global information is proposed. Based on the Lyapunov stability theorem, sufficient and conditions for heterogeneous multi-agent systems to achieve multi-group consistency are obtained, and the containment strategy of agents in heterogeneous cooperative-competitive systems is discussed, thereby reducing the number of contained agents and reducing economic control costs. Finally, the effectiveness of the proposed event triggering is verified by examples.

Description

Technical Field

[0001] The present invention belongs to the field of multi-agent system control and relates to a heterogeneous unmanned system cluster collaborative control method based on event triggering. Background Art

[0002] In recent decades, multi-group consistency, as an extension of consistency, has been widely used in the fields of drones, mobile robots, and drone formation control. Multi-group consistency is a typical and critical problem in the collaborative control of multi-agent systems. It means that the agents in the same group in the system converge to consistency, while agents in different groups reach different convergence values. Multi-agent multi-group consistency is to better solve the parallel processing of complex tasks. Therefore, multi-group consistency for multi-agent systems has become an important scientific research issue, which is also very challenging and has great application value.

[0003] At present, most of the research on multi-agent cooperative control is based on homogeneous multi-agent systems, that is, all the agents in the system have the same dynamic behavior. At the same time, the topology requirements of the multi-agent system are high, such as the need to contain directed spanning trees, in-degree balance, strong connectivity, symmetry, etc. These special conditions cannot be generally applied to complex real systems. In practical applications, multi-agent systems require different types of agents to divide the work to complete complex tasks, such as the division of labor of bees in nesting. In addition, in homogeneous multi-agent systems, a single convergence value limits the processing capacity of the system, and it is impossible to achieve multi-task parallel processing, resulting in low work efficiency.

[0004] For complex multi-agent systems, it is difficult to achieve consistency through the coupling effect within the system, and external effects are usually required. Compared with controlling all agents, the specified consistency state can be achieved by controlling some of the agents. This control method not only reduces the workload, but also reduces the difficulty of work and control costs.

[0005] Most of the research works mentioned above are based on a single cooperative or competitive relationship. However, the limited resources in reality trigger competitive behaviors among agents as well as cooperative behaviors of joint collaboration, making the cooperative-competitive relationship among agents common in complex systems. The co-existing cooperative-competitive relationship is more in line with actual needs, such as the railway system.

[0006] The premise for agents to achieve multi-group consistency is that they need to communicate with each other and update their own states and control protocols. However, real-time communication between agents will cause some disadvantages, such as network congestion and data loss. In response to the above problems, periodic communication is proposed. However, for each agent, the time and amount of tasks to be performed are different, and periodic communication will cause different burdens on different agents, and the above problems will also occur. Therefore, event-triggered control is proposed, that is, a suitable event trigger condition is designed for each agent in the system. When the agent meets this condition, it communicates with the adjacent agent and updates the corresponding control protocol and state information. At present, many studies are based on static event-triggered control, but its triggering conditions are not completely distributed, that is, the triggering conditions contain global information, such as the eigenvalues ​​of the Laplace matrix, the total number of systems, etc. Due to the variability of the system, it is generally difficult to directly obtain the global information of the system.

[0007] In summary, it is an urgent problem to study how to achieve multi-group consistency in heterogeneous multi-agent systems with cooperative-competitive relationships based on an event-triggered approach. Summary of the invention

[0008] The present invention aims to solve the above problems of the prior art. A method for coordinated control of heterogeneous unmanned system clusters based on event triggering is proposed, and the method comprises the following steps:

[0009] Step 1: Based on the premise that the multi-group consistency between agents is information interaction, determine the topological structure of the heterogeneous multi-agent system, group the agents in the heterogeneous multi-agent system, and set the expected convergence state value for each agent group. The heterogeneous multi-agent system contains first-order agents and second-order agents, where the first-order agents only have position state information and the second-order agents have position state information and speed state information.

[0010] Step 2: Consider the cooperative-competitive interaction relationship between agents in the system, set up a multi-group consistency control protocol for the agents in the system, and consider the measurement error and system error of the agents; the agents in the same group have not only cooperative interaction relationships but also competitive interaction relationships, and the agents in different groups also have not only cooperative interaction relationships but also competitive interaction relationships. This type of dual interaction relationship is more in line with practical applications, and

[0011] Step 3: Design appropriate event trigger conditions for each agent in the system. When an agent meets the current event trigger conditions, it communicates with adjacent agents and updates the control protocol. This event trigger does not contain global information, which is consistent with the simple and flexible characteristics of the agent.

[0012] Step 4: Construct the agent's containment strategy based on the Lyapunov method, and derive the conditions for the agent to meet multi-group consistency.

[0013] Step 5: Based on the control input value of the multi-group consistency control protocol, the position state information and speed state information of the agent at the next moment are calculated through the system model, and it is determined whether the current position state value and speed state value meet the conditions for achieving multi-group consistency, and the measurement error and global error of the current agent are further updated to prepare for the next trigger.

[0014] Beneficial effects of the present invention:

[0015] 1. The present invention considers a heterogeneous multi-agent system in steps 1 and 2, which includes multi-agents in a cooperative-competitive relationship. The agents are divided into two groups according to the competitive and cooperative relationship between the agents in the heterogeneous multi-agent system. Each group is heterogeneous, that is, each group includes first-order and second-order agents. Compared with a single cooperative relationship or competitive relationship, this internal relationship of competition and cooperation is more in line with the internal relationship of a real complex system, and ultimately makes the agents in the same group converge to the same value, while the agents in different groups converge to opposite values.

[0016] 2. In step 3, fully distributed event triggering conditions are considered, in which event triggering conditions do not contain global information. Generally, for the continuous update of multi-agent systems, its global information is difficult to obtain. At the same time, higher requirements are placed on the storage capacity of a single agent. Since the information storage and computing capabilities of a single agent are limited, this conflicts with the flexibility, simplicity, and low cost of the agent system itself. Therefore, the fully distributed event triggering conditions of the invention that do not contain any global information put forward higher requirements for the multi-group consistency and optimization of agents in a non-continuous communication environment.

[0017] 3. In step 4, the present invention uses the Lyapunov method to obtain a containment control strategy, and by containing some agents, the system converges to the expected value, which not only reduces the control cost, but also reduces the workload. The present invention studies a general topological structure, which can contain isolated nodes or connected branches, releases some strict requirements on the system topology, and the system has a wide range of applicability.

[0018] 4. Step 5 of the present invention is the prerequisite for the system to achieve multi-group consistency, and the intelligent agent in the system continuously updates the position state and speed state. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 is a flow chart of the present invention;

[0020] Figure 2 is a system topology diagram of the present invention;

[0021] Figure 3 is a position state evolution diagram of the intelligent agent of the present invention;

[0022] Figure 4 is a velocity state evolution diagram of the intelligent agent of the present invention;

[0023] Figure 5 It is the event triggering time diagram of the intelligent agent of this method. DETAILED DESCRIPTION

[0024] The following will describe the technical solutions in the embodiments of the present invention in detail in conjunction with the accompanying drawings in the embodiments of the present invention. The described embodiments are only part of the embodiments of the present invention.

[0025] See also Figure 1 The technical solution of the present invention comprises the following steps:

[0026] Step 1: Based on the premise that the multi-group consistency between agents is information interaction, determine the topological structure of the heterogeneous multi-agent system and fully consider the cooperative-competitive interaction relationship between agents. Group the agents in the heterogeneous multi-agent system and set the expected convergence state value for each agent group.

[0027] Step 2: The heterogeneous multi-agent system considered in the present invention mainly includes first-order agents and second-order agents, wherein the first-order agents only have position state information, and the second-order agents have position state information and speed state information. Its system model can be expressed by a dynamic equation.

[0028]

[0029] Among them, x i (t),v i (t) and They are the position state, velocity state and control input of the agent respectively; Represents a real number. m and I n-m They are respectively a set of first-order agents and a combination of second-order agents. Represent the derivatives of the agent's position state and velocity state respectively.

[0030] Step 3: Consider the cooperative-competitive interaction relationship between agents in the system, and set up a multi-group consistency control protocol for the agents in the system. And consider the measurement error and system error of the agents; agents in the same group can have not only cooperative interaction relationships but also competitive interaction relationships, and agents in different groups can have not only cooperative interaction relationships but also competitive interaction relationships. Based on the above interaction relationship between agents, the multi-group consistency control protocol is designed as follows:

[0031]

[0032] Among them, k1, k2 and k3∈R + is the coupling strength of the agent control protocol, R + represents a real number; d i represents the containment gain of the agent. The interaction relationship between agents in the system is represented by c ij To indicate that if c ij =1, indicating that the agents are in a cooperative interaction relationship; if the agents are in a competitive interaction relationship, then c ij =-1. x j and x i denote the jth and ith agents respectively. ij Represents the topological relationship between the i-th agent and the j-th agent. Indicates the most recent triggering time of the neighboring node of agent i; represents the current triggering moment of agent i; The next triggering moment of agent i. represents the current grouping final state of agent j; Represents the current grouping final state of agent i. i represents the speed state of the ith agent, t is the current time. N i The set of neighbor nodes of the i-th agent.

[0033] In the process of multi-group consistency analysis, measurement error and global error are used to judge whether the agents in the system have achieved group consistency. Among them, the measurement error refers to the state value of the agent at the previous moment and the current state value. Its position measurement error is:

[0034]

[0035] The speed measurement error is:

[0036]

[0037] The measurement error about the neighboring agents is:

[0038]

[0039] The system global error is the criterion for judging whether the current agent has reached the final state value, and it is set as:

[0040]

[0041] Based on the above system model, multi-group consistency control protocol, measurement error and global error, the multi-group control protocol can be modified as follows:

[0042]

[0043] Step 4: In order to avoid real-time and continuous communication between agents, design appropriate event triggering conditions for each agent in the system. When an agent meets the current event triggering conditions, it communicates with adjacent agents and updates the control protocol.

[0044] The purpose of using event triggering between agents is to avoid network congestion, data loss and other problems, and it is also conducive to saving economic costs. Considering a heterogeneous multi-agent system, the dynamic behaviors of the agents in the system are different, so different agents need to be considered when designing event triggering. Therefore, the fully distributed event triggering design of first-order agents and second-order agents is:

[0045]

[0046] and

[0047]

[0048] Where K1 = k1l min +d min , and the constant ε i ∈[0,1). l represents a constant, l ii The in-degree of the ith agent.

[0049] It can be seen from the event trigger conditions proposed above that the event trigger conditions do not contain global information, such as the eigenvalues ​​of the Laplace matrix and the total number of agents in the system. Because the system is constantly changing, global information is generally difficult to obtain. At the same time, if the event trigger contains global information, it puts forward higher requirements on the computing power and storage capacity of the agent.

[0050] Step 5: From the above analysis, step 4 and step 3 are transformed into the designed Lyapunov function:

[0051] W(t)=W1(t)+W2(t)

[0052] in, and n is the total number of agents, m is the number of first-order agents. Analyzing the Lyapunov method, we can derive sufficient conditions to ensure that the agents in the system achieve multi-group consistency:

[0053] k1l ii +d i >0,i∈I m ,

[0054] k2l ii +d i>0,i∈I n-m ,

[0055] k3-1>0,i∈I n-m .

[0056] At the same time, a containment strategy is proposed based on the Lyapunov method, that is, agents with zero in-degree must be contained. Therefore, to ensure that the agents in the system achieve multi-group consistency, it is necessary to meet the above conditions and the containment strategy.

[0057] Step 6: Based on the control input value of the multi-group consistency control protocol, the position state information and speed state information of the agent at the next moment are calculated through the system model. And it is determined whether the current position state value and speed state value meet the conditions for achieving multi-group consistency. Further update the measurement error and global error of the current agent to prepare for the next trigger.

[0058] The process of judging the position and speed status of the agents in the system includes two aspects: on the one hand, judging whether the position status information of the same group in the system is the same; on the other hand, judging whether the speed status of the second-order agent is zero. It can be described by the following conditions:

[0059] Condition one:

[0060] Among them, x i (t) represents the position status information of the agent. is the group to which the ith agent belongs. Indicates that agent i and agent j are in the same group; when Indicates that agents i and j are in different groups.

[0061] Condition two:

[0062] Among them, v i (t) represents the speed state information of agent i.

[0063] like Figure 2 As shown. The system contains ten agents, of which the first-order agents are nodes 1, 2, 3, 4, 5, 6 and the second-order agents are nodes 7, 8, 9, 10. The system is divided into three groups, of which 1, 5, 6, 10 are the first group, 2, 3, 8 are the second group, and 4, 7, 9 are the third group, and the final state values ​​of the three groups are set to 2, 6, and -8 respectively. Among them, the initial position state of the agent is set to x(0) = (-2, -6, 7, 1, -4, 7, 1, -4, 5, 8) T , the initial velocity state of the second-order agent is set to v(0)=(5,-3,-2,3)T According to the containment strategy proposed in step 5, nodes 1, 2, and 7 must be contained. Therefore, the containment gains of the designed agent are: d1 = 1.2, d2 = 1.5, d7 = 0.8 and d i =0, i∈{3,4,5,6,8,9,10}. The present invention comprehensively considers the cooperative-competitive interaction between intelligent agents and Figure 1 The “+” between the agents in the equation represents a competitive relationship between the agents, and the “-” represents a cooperative interaction relationship between the agents. The coupling gains in the system are set to: k1 = 0.5, k2 = 0.2, and k3 = 1.2.

[0064] Figure 3 and Figure 4 It shows the evolution of the position state of all nodes in the system and the velocity state of the second-order nodes. Figure 3 It can be seen that the agents in the agent system finally reach the expected state value and achieve multi-group consistency. Figure 4 It can be seen that the velocity states of the second-order agents in the system all converge to zero.

[0065] Figure 5 It is a diagram showing the triggering time of the intelligent agents in the system. If the periodic communication of the intelligent agents is adopted, the average number of intelligent agent communications is 200 times; if the event-triggered control method is adopted, the average number of intelligent agent communications is only 54 times, which is reduced by 73%.

[0066] The above embodiments further illustrate the purpose, technical solutions and advantages of the present invention in detail. It should be understood that the above embodiments are only preferred implementation modes of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made to the present invention within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for coordinated control of heterogeneous unmanned system clusters based on event triggering, characterized in that: The following steps are involved: Step 1: Based on the premise that the multi-group consistency between agents is information interaction, determine the topological structure of the heterogeneous multi-agent system, and comprehensively consider the cooperation-competition interaction relationship between agents, group the agents in the heterogeneous multi-agent system, and set the expected convergence state value for each agent group; Step 2: The heterogeneous multi-agent system includes first-order agents and second-order agents, where the first-order agents only have position state information and the second-order agents have position state information and speed state information; Step 3: Consider the cooperative-competitive interaction relationship between agents in the system, set up a multi-group consistency control protocol for the agents in the system, and consider the measurement error and system error of the agents; the agents in the same group have not only cooperative interaction relationships but also competitive interaction relationships, and the agents in different groups also have not only cooperative interaction relationships but also competitive interaction relationships; The multi-packet consistency control protocol is: Where k1, k2 and k3 are the coupling strengths of the agent control protocol; d i represents the agent's containment gain; c ij Represents the interaction relationship between agents in the system. If c ij =1, indicating that there is a cooperative interaction relationship between the agents; if there is a competitive interaction relationship between the agents, then c ij =-1; x j and x i denote the jth and ith agents respectively, a ij represents the topological relationship between the i-th agent and the j-th agent, Indicates the most recent triggering time of the neighboring node of agent i; represents the current triggering moment of agent i; The next triggering moment of agent i, represents the current grouping final state of agent j; represents the final state of the current group of agent i, v i represents the speed state of the ith agent, t is the current time, N i The set of neighbor nodes of the i-th agent; Step 4: Design appropriate event triggering conditions for each agent in the system. When an agent meets the current event triggering conditions, it communicates with adjacent agents and updates the control protocol. The event triggering conditions include fully distributed event triggering of first-order agents and second-order agents, respectively: and Where K1 = k1l min +d min , and the constant ε i ∈[0,1); l represents a constant, l ii The in-degree of the ith agent; Step 5: Construct the agent's containment strategy based on the Lyapunov method, and derive the conditions for the agent to meet multi-group consistency; Step 6: Based on the control input value of the multi-group consistency control protocol, the position state information and speed state information of the agent at the next moment are calculated through the system model, and it is determined whether the current position state value and speed state value meet the conditions for achieving multi-group consistency, and the measurement error and global error of the current agent are further updated to prepare for the next trigger.

2. The event-triggered heterogeneous unmanned system cluster collaborative control method according to claim 1 is characterized by: The heterogeneous multi-agent system is expressed by the dynamic equation: Among them, x i (t),v i (t) and μ i (t) are the position state, velocity state and control input of the agent respectively; I m and I n-m They are the first-order agent set and the second-order agent combination respectively; represents the derivative of the agent's position state, Represents the derivative of the agent's velocity state.

3. The event-triggered heterogeneous unmanned system cluster collaborative control method according to claim 1 is characterized by: The measurement error and the global error are used to judge whether the agents in the system have achieved group consistency; wherein the position measurement error is: I n represents a collection of agents; The speed measurement error is: The measurement error about the neighboring agents is: x j (t) the jth agent; The system global error is the criterion for judging whether the current agent has reached the final state value, and it is set as:

4. The event-triggered heterogeneous unmanned system cluster collaborative control method according to claim 3 is characterized by: Modify the multi-packet control protocol to:

5. The event-triggered heterogeneous unmanned system cluster collaborative control method according to claim 1 is characterized by: The containment strategy of the intelligent agent based on the Lyapunov method described in step 5 is: The designed Lyapunov function is: W(t)=W1(t)+W2(t) in, and n represents the total number of agents, m represents the number of first-order agents; the sufficient condition to ensure that the agents in the system achieve multi-group consistency is: k1l ii +d i >0,i∈I m , k2l ii +d i >0,i∈I n-m , k3-1>0,i∈I n-m 。 6. The event-triggered heterogeneous unmanned system cluster collaborative control method according to claim 1, characterized in that: The process of determining whether the position state value and the speed state value meet the multi-group consistency condition described in step 6 includes two aspects: on the one hand, determining whether the position state information of the same group in the system is the same; on the other hand, determining whether the speed state of the second-order agent is zero; Describe it using the following conditions: Condition 1: Among them, x i (t) represents the position state information of the agent, l i is the group to which the ith agent belongs. When l i = l j Indicates that agent i and agent j are in the same group, when l i ≠l j Indicates that agent i and agent j are in different groups; Condition 2: Among them, v i (t) represents the speed state information of agent i.

Images