Cluster intelligence system protection method based on specified time distribution aggregation optimization

Abstract

The application discloses a cluster intelligence system protection method based on specified time distributed aggregation optimization, and belongs to the technical field of cluster intelligence systems, and comprises the following steps: designing a local cost function for an intelligent agent, converting a cluster intelligence system cooperative protection problem into a distributed aggregation optimization problem based on an undirected connected graph based on the local cost function; setting a predetermined convergence time and a sampling time sequence for the cluster intelligence system; designing a specified time distributed aggregation optimization algorithm based on the sampling time sequence, and obtaining an optimal solution of the distributed aggregation optimization problem at the predetermined convergence time, while controlling the intelligent agent to reach an optimal position of the cooperative protection task. The method combines the distributed aggregation optimization framework with the specified time convergence technology, solves the cluster intelligence system cooperative protection problem, enables the intelligent agent to obtain the optimal solution of the distributed aggregation optimization problem at the specified time, and is beneficial to the cluster intelligence system to more accurately complete the cooperative protection task in the time layer.

Description

Technical Field

[0001] This application relates to the field of cluster intelligent system technology, and in particular to a cluster intelligent system protection method based on time-distributed aggregation optimization. Background Technology

[0002] Swarm intelligence systems are network systems composed of multiple intelligent agents that collaborate to complete designated tasks. These agents can be neural networks, sensors, drones, robots, and other devices. In recent years, practical scenarios such as marine environmental monitoring, unmanned vehicle ground patrols, and drone-based collaborative offense and defense have required sensors, unmanned vehicles, and drones to collaborate to find optimal positions to efficiently perform protection tasks in each scenario. Therefore, research on collaborative protection using swarm intelligence systems has significant practical implications.

[0003] In practical applications, each agent needs to construct its own private cost function, modeling the task to be executed as a distributed optimization problem. Agents collaborate and exchange information to find the global optimum of this distributed optimization problem, thereby achieving efficient task execution. However, in the scenario of collaborative protection in a swarm intelligence system, each agent's private cost function depends not only on its own decision variables but also on the aggregated information of the decision variables of all agents in the system. Under existing distributed optimization frameworks, each agent's private cost function only depends on its own decision variables, making it difficult to solve within the current framework. Furthermore, in the scenario of collaborative protection in a swarm intelligence system, each agent needs to reach the globally optimal protection position within a pre-specified time. Therefore, researching a collaborative protection method for swarm intelligence systems based on time-specified distributed aggregation optimization is of great practical significance and warrants further investigation. Summary of the Invention

[0004] To address the aforementioned shortcomings in existing technologies, this application provides a cluster intelligent system protection method based on distributed aggregation optimization at a specified time. This method enables intelligent agents to obtain the optimal solution to the distributed aggregation optimization problem at a specified time, which is beneficial for cluster intelligent systems to complete collaborative protection tasks more accurately at the time level.

[0005] To achieve the aforementioned objectives, the technical solution adopted in this application is as follows: This application provides a cluster intelligent system protection method based on time-distributed aggregation optimization, including: S1: For the collaborative protection problem of cluster intelligent systems, a local cost function is designed for the agents, and based on the local cost function, the collaborative protection problem of cluster intelligent systems is transformed into a distributed aggregation optimization problem based on an undirected connected graph. The expression for the local cost function is:

[0006] In the formula, For the first Local cost function of an agent Describing the 2-norm, Indicates the first The location of the intruder corresponding to each agent. and They represent the first The location of the intruder corresponding to each agent is on the horizontal and vertical components of the coordinate system, with superscripts indicating the location. For transpose, Indicates the location of the protected target. and These represent the components of the protected target's position on the horizontal and vertical axes of the coordinate system, respectively. and Represents positive numbers. This represents the cluster center of the swarm intelligence system. A vector representing the stacked positions of all agents. Indicates the first The location of each agent. and They represent the first The position of an agent is on the components of the horizontal and vertical axes of the coordinate system. The number of agents. Indicates time, It is the set of real numbers; S2: Set the predetermined convergence time and sampling time series for the swarm intelligence system; S3: Based on the sampled time series, design a distributed aggregation optimization algorithm for a specified time, obtain the optimal solution of the distributed aggregation optimization problem within a predetermined convergence time, and control the agent to reach the optimal position of the collaborative protection task.

[0007] Further, S1 includes: S101: Establish a communication topology diagram of a swarm intelligence system consisting of multiple intelligent agents, wherein the communication topology diagram The expression is:

[0008] in, Represents a set of intelligent agents. The number of agents. Indicates the first An intelligent agent. The edge set representing communication between intelligent agents; S102: Construct the model of the intelligent agent, expressed as:

[0009] in, Indicates time, superscript Represents the derivative with respect to time. Indicates the first The location of each agent. and They represent the first The position of an agent is on the components of the horizontal and vertical axes of the coordinate system. Indicates the first Control input for an intelligent agent and Indicates the first The components of the control input of an agent on the horizontal and vertical axes of the coordinate system. It is the set of real numbers; S103: Construct the local cost function of the agent, and under the condition that the communication topology of the swarm intelligence system is an undirected graph, based on the local cost function of the agent, abstract the collaborative protection problem of the swarm intelligence system into a distributed aggregation optimization problem, expressed as: .

[0010] Furthermore, the communication topology diagram includes: A Laplace matrix is ​​constructed to represent the connection relationships in the communication topology. The expression for the Laplace matrix is ​​as follows:

[0011]

[0012]

[0013] in, It is an adjacency matrix. For the first The first intelligent agent and the first Communication weights between agents Let be the in-degree matrix. For the first The in-degree of an agent, It is a diagonal matrix. It is a Laplace matrix.

[0014] Furthermore, the expression for the sampling time series is:

[0015] in, It is a sampled time series. For the first Each sampling time, For the first Each sampling time, The initial sampling time, Indicates the first The sampling time and the first The time interval between each sampling moment Represents the Euler sequence. The predetermined convergence time.

[0016] Furthermore, the expression for the specified time distributed aggregation optimization algorithm is:

[0017] in, For the first Each sampling time, and It is the algorithm's constant gain. Indicates the first Local cost function of an agent The gradient relative to the first variable, , Indicates the first Local cost function of an agent The gradient relative to the second variable, , Indicates the first Local cost function of an agent The gradient relative to the second variable, , For the first The cluster center of a cluster intelligent system for local estimation of physical abilities. For the first The intelligent agent can locally estimate global gradient aggregation information. , and They represent the first A single agent , and exist Sampling information at any given time , and For the first An intelligent agent in Sampling information at any given time.

[0018] The beneficial effects of this application are: This application proposes a protection method for swarm intelligent systems based on specified-time distributed aggregation optimization. By combining the distributed aggregation optimization framework with specified-time convergence technology, it solves the problem of collaborative protection of swarm intelligent systems. The local cost function of the agent in this invention is related not only to its own decision variables but also to the aggregated information of the decision variables of all agents in the network. Furthermore, the agent in this invention can reach the globally optimal position for collaborative protection of the swarm intelligent system within a specified time, which is beneficial for better completing tasks such as marine environmental monitoring and unmanned vehicle ground patrol. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other embodiments can be obtained based on these drawings.

[0020] Figure 1 This is a flowchart illustrating a cluster intelligent system protection method based on time-distributed aggregation optimization provided in an embodiment of this application.

[0021] Figure 2 This is a communication topology diagram between intelligent agents provided in an embodiment of this application.

[0022] Figure 3 This is a schematic diagram of the motion trajectory of an intelligent agent in a collaborative protection task of a cluster intelligent system based on distributed aggregation optimization at a specified time, provided as an embodiment of this application.

[0023] Figure 4 This is a schematic diagram illustrating the distributed estimation of the cluster center trajectory by agents in a collaborative protection task of a cluster intelligent system based on distributed aggregation optimization at a specified time, as provided in an embodiment of this application.

[0024] Figure 5 This is a schematic diagram illustrating the trajectory of an agent in a collaborative protection task of a cluster intelligent system based on time-distributed aggregation optimization, which is provided in an embodiment of this application, to distribute and estimate the global aggregated gradient information in a distributed manner.

[0025] Figure 6 This is a schematic diagram illustrating the change of the global cost function in a collaborative protection task of a cluster intelligent system based on distributed aggregation optimization at a specified time, as provided in an embodiment of this application. Detailed Implementation

[0026] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art based on this application are within the scope of protection of this application.

[0027] Example 1: Based on this, embodiments of this application provide a cluster intelligent system protection method based on time-distributed aggregation optimization, which can be found in [reference needed]. Figure 1 , Figure 1 The diagram shown is a flowchart illustrating a cluster intelligent system protection method based on distributed aggregation optimization at a specified time, according to an embodiment of this application, including: S1: For the collaborative protection problem of swarm intelligent systems, a local cost function is designed for each agent, transforming the collaborative protection problem of swarm intelligent systems into a distributed aggregation optimization problem based on an undirected connected graph.

[0028] In one embodiment of this application, S1 specifically includes: S101: Establish by Communication topology diagram of a swarm intelligence system composed of intelligent agents :

[0029] in, Represents a set of intelligent agents. Indicates the first An intelligent agent. The number of agents. The edge set representing communication between intelligent agents. Indicates the first The agent can receive the first... Messages sent by an agent, communication topology diagram The adjacency matrix is ,if , then for the first The first intelligent agent and the first Communication weights between agents ,otherwise Communication topology diagram The in-degree matrix is , representing the in-degree of each agent The diagonal matrix formed Since it is a diagonal matrix, the Laplace matrix representing the connection relationships in the communication topology is: .

[0030] S102: Construct the model of the intelligent agent, expressed as:

[0031] in, Indicates time, superscript Represents the derivative with respect to time. Indicates the first The location of each agent. and They represent the first The position of an agent is on the components of the horizontal and vertical axes of the coordinate system. Indicates the first Control input for an intelligent agent and Indicates the first The components of the control input of an agent on the horizontal and vertical axes of the coordinate system. It is the set of real numbers.

[0032] S103: In a collaborative protection task, each agent's task consists of three parts: to better protect the intruder, the agent's position should be as close to the intruder as possible; to better monitor the state of the protected target, the agent's position should be as close to the protected target as possible; to better achieve protection for the agent cluster, the cluster center of the cluster intelligence system should track the position of the protected target as closely as possible, expressing it mathematically, for the... For each agent, the local cost function is as follows:

[0033] in, For the first Local cost function of an agent Describing the 2-norm, Indicates the first The location of the intruder corresponding to each agent. and They represent the first The location of the intruder corresponding to each agent is on the horizontal and vertical components of the coordinate system, with superscripts indicating the location. For transpose, Indicates the location of the protected target. and These represent the components of the protected target's position on the horizontal and vertical axes of the coordinate system, respectively. and Represents positive numbers. This represents the cluster center of the swarm intelligence system. Indicates the number of agents. A vector representing the stacked positions of all agents.

[0034] S104: Given that the communication topology of a swarm intelligence system is undirected, the collaborative protection problem of a swarm intelligence system is abstracted into a distributed aggregation optimization problem, expressed as: .

[0035] S2: Set the predetermined convergence time and sampling time series for the swarm intelligence system.

[0036] In an optional embodiment of the present invention, this embodiment sets a predetermined convergence time for the swarm intelligence system. Set the sampling time series It was divided into multiple time periods, and the time series was sampled. Specifically:

[0037] in, Indicates the first Each sampling time, For the first Each sampling time, The initial sampling time, Indicates the first The sampling time and the first The time interval between each sampling moment Represents the Euler sequence. The predetermined convergence time.

[0038] S3: Based on the sampled time series, design a distributed aggregation optimization algorithm for a specified time to obtain the optimal solution of the distributed aggregation optimization problem within a predetermined convergence time, while controlling the agent to reach the optimal position of the collaborative protection task.

[0039] In an optional embodiment of the present invention, the time-specific distributed aggregation optimization algorithm designed in this embodiment is specifically as follows:

[0040] in, Indicates the current sampling time. This indicates the next sampling time, specifically given by the sampling time series. and It is the algorithm's constant gain. Indicates the first The location of each agent. Indicates the first Local cost function of an agent The gradient relative to the first variable is specifically: , Indicates the first The location of the intruder corresponding to each agent. Indicates the location of the protected target. Indicates the first Local cost function of an agent The gradient relative to the second variable is specifically: , and Represents auxiliary variables. Indicates the first Local cost function of an agent The gradient relative to the second variable is specifically: , For the first The cluster center of a cluster intelligent system for local estimation of physical abilities. For the first The intelligent agent can locally estimate global gradient aggregation information. , and Representing variables respectively , and exist Sampling information at any given time The number of agents. For the first The first intelligent agent and the first Communication weights between agents , and For the first An intelligent agent in Sampling information at any given time.

[0041] Example 2: This application provides a detailed analysis and explanation of the method provided in Embodiment 1, using specific examples. This embodiment considers a swarm intelligence system composed of six agents, with the communication topology between the agents as follows: Figure 2 As shown, the Laplace matrix of the corresponding swarm intelligence system is:

[0042] The initial positions of each agent are shown in Table 1 below: Table 1 Initial positions of each agent

[0043] The locations of the various intruders are shown in Table 2 below: Table 2 Locations of each intruder

[0044] The protected target is located at [6.5, 5], and the local cost function for each agent is:

[0045] in, , , .

[0046] For a specified time distributed aggregation optimization algorithm, you can choose... , The algorithm specifies the convergence time. .

[0047] Figure 3 This describes the motion trajectory of agents in a collaborative protection task of a swarm intelligence system based on distributed aggregation optimization at a specified time. Here, x(m) represents the agent's position information in the x-direction, y(m) represents the agent's position information in the y-direction, triangles represent the intruder's position, squares represent the protected target's position, hollow circles represent the agent's initial position, solid circles represent the optimal protection position reached by the agent, and asterisks represent the swarm center of the swarm intelligence system. Figure 3 It can be seen that the agents achieved the collaborative protection task and reached the optimal position through cooperation. The agents' position was closer to the intruder, which allowed for better protection of the intruder; the agents' position was closer to the target, which allowed for better monitoring of the target's status; and the agents' cluster center tracked the target, which enabled better protection of the target.

[0048] Figure 4 and Figure 5 The diagram illustrates the trajectory of agents in a collaborative protection task of a swarm intelligence system based on time-distributed aggregation optimization, where agents distribute and estimate the cluster center and global aggregation gradient information. Each curve represents the estimation information of one agent. Figure 4 It can be seen that the agents reach a consensus on the distributed estimation of the cluster center location of the swarm intelligence system within 3 seconds. Figure 5 It can be seen that the agents reach a consensus on the distributed estimation of the global aggregated gradient information of the swarm intelligence system within 3 seconds. Furthermore, by Figure 4 and Figure 5 It can be seen that the distributed estimation information of the agents remains unchanged after 3 seconds. This indicates that at 3 seconds, each agent in the swarm intelligence system reaches the optimal position for the collaborative protection task. The positions of the agents remain unchanged, the cluster center remains unchanged, and the global aggregated gradient information also remains unchanged.

[0049] Figure 6 This describes the variation of the global cost function in a collaborative protection task of a cluster intelligent system based on distributed aggregation optimization at a specified time, where the solid line represents the global cost function. Figure 6 It can be seen that the global cost function reaches its minimum value at 3 seconds.

[0050] The results above show that the cluster intelligent system protection method based on distributed aggregation optimization at a specified time proposed in this invention is effective. The intelligent agent can reach the globally optimal position for collaborative protection of the cluster intelligent system at a specified time, which is conducive to better completing tasks such as marine environmental monitoring and unmanned vehicle ground patrol.

[0051] It should be noted that those skilled in the art will recognize that the embodiments described herein are for the purpose of helping readers understand the principles of this application, and should be understood as not limiting the scope of protection of this application to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this application without departing from the essence of this application, and these modifications and combinations are still within the scope of protection of this application.

Claims

1. A cluster intelligent system protection method based on distributed aggregation optimization at a specified time, characterized in that, include: S1: For the collaborative protection problem of cluster intelligent systems, a local cost function is designed for the agents, and based on the local cost function, the collaborative protection problem of cluster intelligent systems is transformed into a distributed aggregation optimization problem based on an undirected connected graph. The expression for the local cost function is: In the formula, For the first Local cost function of an agent Describing the 2-norm, Indicates the first The location of the intruder corresponding to each agent. and They represent the first The location of the intruder corresponding to each agent is on the horizontal and vertical components of the coordinate system, with superscripts indicating the location. For transpose, Indicates the location of the protected target. and These represent the components of the protected target's position on the horizontal and vertical axes of the coordinate system, respectively. and Represents positive numbers. This represents the cluster center of the swarm intelligence system. A vector representing the stacked positions of all agents. Indicates the first The location of each agent. and They represent the first The position of an agent is on the components of the horizontal and vertical axes of the coordinate system. The number of agents. Indicates time, It is the set of real numbers; S2: Set the predetermined convergence time and sampling time series for the swarm intelligence system; S3: Based on the sampled time series, design a distributed aggregation optimization algorithm for a specified time, obtain the optimal solution of the distributed aggregation optimization problem within a predetermined convergence time, and control the agent to reach the optimal position of the collaborative protection task. The expression for the specified time distributed aggregation optimization algorithm is: Among them, superscript Represents the derivative with respect to time. For the first Each sampling time, For the first Each sampling time, and It is the algorithm's constant gain. Indicates the first Local cost function of an agent The gradient relative to the first variable, , Indicates the first Local cost function of an agent The gradient relative to the second variable, , Indicates the first Local cost function of an agent The gradient relative to the second variable, , For the first The cluster center of a cluster intelligent system for local estimation of physical abilities. For the first The intelligent agent can locally estimate global gradient aggregation information. , and They represent the first A single agent , and exist Sampling information at any given time , and For the first An intelligent agent in Sampling information at any given time For the first The first intelligent agent and the first Communication weights between agents The number of agents.

2. The cluster intelligent system protection method based on time-distributed aggregation optimization according to claim 1, characterized in that, S1 includes: S101: Establish a communication topology diagram of a swarm intelligence system consisting of multiple intelligent agents, wherein the communication topology diagram The expression is: in, Represents a set of intelligent agents. The number of agents. Indicates the first An intelligent agent. The edge set representing communication between intelligent agents; S102: Construct the model of the intelligent agent, expressed as: in, Indicates time, superscript Represents the derivative with respect to time. Indicates the first The location of each agent. and They represent the first The position of an agent is on the components of the horizontal and vertical axes of the coordinate system. Indicates the first Control input for an intelligent agent and Indicates the first The components of the control input of an agent on the horizontal and vertical axes of the coordinate system. It is the set of real numbers; S103: Construct the local cost function of the agent, and under the condition that the communication topology of the swarm intelligence system is an undirected graph, based on the local cost function of the agent, abstract the collaborative protection problem of the swarm intelligence system into a distributed aggregation optimization problem, expressed as: 。 3. The cluster intelligent system protection method based on time-distributed aggregation optimization according to claim 2, characterized in that, The communication topology diagram includes: A Laplace matrix is ​​constructed to represent the connection relationships in the communication topology. The expression for the Laplace matrix is ​​as follows: in, It is an adjacency matrix. For the first The first intelligent agent and the first Communication weights between agents Let be the in-degree matrix. For the first The in-degree of an agent, It is a diagonal matrix. It is a Laplace matrix.

4. The cluster intelligent system protection method based on time-distributed aggregation optimization according to claim 3, characterized in that, The expression for the sampling time series is: in, It is a sampled time series. For the first Each sampling time, For the first Each sampling time, The initial sampling time, Indicates the first The sampling time and the first The time interval between each sampling moment Represents the Euler sequence. The predetermined convergence time.

Images