A distributed situation maintenance method for unmanned clusters based on information extraction

Abstract

The present invention relates to an unmanned cluster distributed situation maintenance method based on information extraction. The unmanned cluster includes multiple intelligent agents, each of which maintains global layer, local layer, and single machine layer situation information; all single machine layer situation information in the local area is shared to the local area temporary information center node; the local temporary center node extracts information to form local layer situation information, and shares the local layer information to all local intelligent agents; after all local areas complete information extraction and sharing, the local layer information extracted by all local area temporary information center nodes is shared to the global temporary information center node; the global temporary center node extracts information to form global layer situation information, and shares the global situation information to all intelligent agents in the unmanned cluster. The present invention realizes distributed situation maintenance of unmanned clusters based on information extraction, which can improve the execution efficiency of unmanned cluster tasks.

Description

Technical Field

[0001] The invention belongs to the field of distributed decision-making, and in particular relates to an unmanned cluster distributed situation maintenance method based on information extraction. Background Art

[0002] Since the 1990s, unmanned systems have been significantly developed and widely used in many fields. However, under complex conditions, the load capacity and range of a single unmanned system are limited and cannot meet the needs. Therefore, unmanned swarms composed of multiple unmanned systems that are interconnected and work together to complete tasks have emerged. For unmanned swarm systems, the common task coordination method is centralized control, which has a simple and clear principle and has been applied in various task scenarios. However, the centralized collaborative control method is a centralized control method. Once the central node fails, the task execution efficiency will be seriously reduced.

[0003] Another unmanned cluster collaboration method is distributed collaboration, which is decentralized control. Each node only needs to make decisions based on the information it has. Therefore, this collaboration method is more robust and has stronger environmental adaptability, and has received more and more attention. However, the existing distributed collaboration method still has certain problems, which limit its further application, as detailed below:

[0004] 1) It requires sharing a large amount of data and cannot adapt to mission scenarios with poor communication;

[0005] 2) Failure to establish a hierarchical situation results in a lack of hierarchical decision-making and inability to cope with complex scenarios;

[0006] 3) No information extraction mechanism has been established, and the information used for decision-making is raw information.

[0007] In summary, how to solve the current problems of distributed collaboration is the key to the further application of distributed collaboration. Summary of the invention

[0008] Aiming at the defects of the prior art and the problem that the distributed situation of unmanned clusters is difficult to maintain in complex real-life scenarios, the present invention proposes an unmanned cluster distributed situation maintenance method based on information extraction, which can well maintain the situation in the distributed collaboration of unmanned clusters.

[0009] In order to achieve the above object, the technical solution adopted by the present invention is:

[0010] A method for distributed situation maintenance of an unmanned cluster based on information extraction, wherein the unmanned cluster includes a plurality of intelligent agents; the method comprises the following steps:

[0011] (1) Each agent establishes its own single-machine situation information.

[0012] (2) Sharing the single-machine-layer situation information of all agents in the local area to the temporary information center node of the local area;

[0013] (3) The local area temporary information center node extracts information to form the local layer situation information of the corresponding local area, and shares the local layer information with all intelligent agents in the corresponding local area;

[0014] (4) After all local layer information is shared in their respective local areas, the local layer information extracted by all local area temporary information center nodes is shared to the global temporary information center node;

[0015] (5) The global temporary information center node extracts information to form global situation information, and then shares the global situation information with all intelligent agents in the unmanned cluster.

[0016] Furthermore, each intelligent agent has communication capabilities and information sharing capabilities; the global layer situation information is the overall situation information of the unmanned cluster during the task execution process, including information on task execution efficiency, unmanned cluster task execution capability, and resource redundancy; the local layer situation information is the situation information of the local area, including information on task execution efficiency, sub-unmanned cluster task execution capability, target movement trend, and resource redundancy in the local area; the single-machine layer situation information is the original information of the intelligent agent, including information on positioning, task execution capability, and covered targets.

[0017] Furthermore, the selection rule of the local area temporary information center node is:

[0018] The intelligent agent located at the center of the local area communication topology will take over. If this intelligent agent fails, the intelligent agent located at the secondary center of the local area communication topology will take over, and so on.

[0019] Furthermore, the information extraction methods described in step (3) and step (5) include statistics, weighting, analysis and prediction; wherein, in step (3), the local area temporary information center node extracts information to form the local layer situation information of the corresponding local area; wherein, the task execution efficiency is calculated as follows:

[0020]

[0021] In the above formula, Tee l represents the task execution efficiency of this local area, U lnum represents the number of agents in this local area, te i represents the task execution capability of the ith agent, Obj lnumIndicates the number of coverage targets in this local area;

[0022] The calculation method of the task execution capability of the sub-unmanned cluster is:

[0023]

[0024] In the above formula, Te l Indicates the task execution capability of this local area, U lnum represents the number of agents in this local area, te i represents the task execution capability of the ith agent;

[0025] The target movement trend is calculated as:

[0026] Obj ltrend =f(Obj0,Obj1,...,Obj9)

[0027] In the above formula, Obj ltrend Indicates the target movement trend of this local area, Obj0 indicates the cluster center of all targets in this local area at the current moment, Obj1 indicates the cluster center of all targets in this local area at the previous moment, and so on, Obj9 indicates the cluster center of all targets in this local area at the previous 10 moments, and f indicates the target trend prediction function;

[0028] Resource redundancy is calculated as follows:

[0029]

[0030] In the above formula, U lre Indicates the resource redundancy of this local area, U lnum represents the number of agents in this local area, U lenum Indicates the number of agents performing tasks in this local area;

[0031] In step (5), the global temporary information center node extracts information to form global situation information; wherein the task execution efficiency is calculated as follows:

[0032]

[0033] In the above formula, Tee represents the global task execution efficiency, L represents the number of local regions, and Tee l represents the task execution efficiency of the lth local area;

[0034] The calculation method of unmanned cluster task execution capability is:

[0035]

[0036] In the above formula, Te represents the task execution capability of the global unmanned cluster, Te l represents the task execution capability of the lth local area;

[0037] Resource redundancy is calculated as follows:

[0038]

[0039] In the above formula, U re represents the global resource redundancy, U lre Represents the resource redundancy of the lth local area.

[0040] Furthermore, all local area temporary communication center nodes form a communication network, and the selection rule of the global temporary information center node is:

[0041] The intelligent agent located at the communication topology center of the communication network takes over. If the intelligent agent fails, the intelligent agent located at the secondary center of the communication topology of the communication network takes over, and so on.

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] 1. The present invention adopts an information extraction mechanism to extract a large amount of original information into a small amount of high-level information, which can effectively limit the amount of communication data and solve the problem of excessive bandwidth occupation in the current method.

[0044] 2. The situation maintenance method of the present invention can be applied to distributed decision-making, and each intelligent agent can make autonomous decisions based on the existing situation information.

[0045] 3. The situation information of the present invention is divided into global layer, local layer and single machine layer, which can cope with the global decision-making, local decision-making and single machine decision-making of unmanned clusters and improve the execution efficiency of unmanned cluster tasks.

[0046] 4. The present invention adopts a temporary information center node mechanism to improve the anti-damage capability of the unmanned cluster. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 It is a flow chart of a method according to an embodiment of the present invention.

[0048] Figure 2 It is a schematic diagram of situation maintenance information of each layer in an embodiment of the present invention.

[0049] Figure 3 It is a schematic diagram of selecting a local area temporary information center in an embodiment of the present invention.

[0050] Figure 4 It is a schematic diagram of selecting a global temporary information center in an embodiment of the present invention. DETAILED DESCRIPTION

[0051] The following will be combined with the embodiments to clearly and completely describe the concept, technical solution advantages and technical effects of the present invention, so as to fully understand the purpose, features and effects of the present invention. It should be noted that the specific embodiments described herein are only used to explain the present invention and do not limit the present invention.

[0052] A distributed situation maintenance method for unmanned clusters based on information extraction, referring to Figure 1 , the implementation steps of this method are as follows:

[0053] Step 1: The unmanned cluster contains multiple agents, such as Figure 2 As shown in the figure, each agent needs to maintain the situation information of the global layer, local layer, and single machine layer, which are described as follows:

[0054] 1a) Each intelligent agent has the ability to communicate and share information;

[0055] 1b) Global situation information is the overall situation information of the unmanned cluster during the task execution process, including information on task execution efficiency, unmanned cluster task execution capability, and resource redundancy;

[0056] 1c) Local layer situation information is the situation information of the local area, including the task execution efficiency, sub-unmanned cluster task execution capability, target movement trend, and resource redundancy information in the local area;

[0057] 1d) Single-machine layer situation information is the original information of the agent, including positioning, task execution capability, covered targets, etc.

[0058] Step 2: All single-machine layer situation information in the local area is shared to the local area temporary information center node. The specific selection rules are as follows:

[0059] like Figure 3 As shown, the local area temporary information center node is served by the agent currently located at the center of the regional communication topology. If the agent fails, it can be served by the agent currently located at the secondary center of the regional communication topology, and so on.

[0060] Step 3: The local temporary central node extracts information to form the local situation information of the local area, and shares the local information with all local agents. The information extraction methods include: statistics, weighting, analysis and prediction, etc.

[0061] Step 4: After all local regions have completed steps (2)-(3), the local layer information extracted by the temporary information center nodes of all local regions is shared to the global temporary information center node;

[0062] Step 5: The global temporary central node extracts information to form global situation information, and shares the global situation information with all agents in the unmanned cluster. The global temporary central node selection rules and information extraction methods are described as follows:

[0063] 5a) The global temporary central node selection rule is: Figure 4 As shown, the global temporary central node is assumed by the intelligent agent at the center of the communication topology formed by all local temporary communication nodes. If this intelligent agent fails, it can be assumed by the intelligent agent at the secondary center of the communication topology formed by all local temporary communication nodes, and so on.

[0064] 5b) Information extraction methods include: statistics, weighting, analysis and prediction, etc. Among them, the calculation method of local area task execution efficiency is:

[0065]

[0066] In the above formula, Tee l represents the task execution efficiency of this local area, U lnum represents the number of agents in this local area, te i represents the task execution capability of the ith agent, Obj lnum Indicates the number of coverage targets in this local area;

[0067] The calculation method of the task execution capability of the local area sub-unmanned cluster is:

[0068]

[0069] In the above formula, Te l Indicates the task execution capability of this local area, U lnum represents the number of agents in this local area, te i represents the task execution capability of the ith agent;

[0070] The calculation method of the local area target movement trend is:

[0071] Obj ltrend =f(Obj0,Obj1,...,Obj9)

[0072] In the above formula, Obj ltrend Indicates the target movement trend of this local area, Obj0 indicates the cluster center of all targets in this local area at the current moment, Obj1 indicates the cluster center of all targets in this local area at the previous moment, and so on, Obj9 indicates the cluster center of all targets in this local area at the previous 10 moments, and f indicates the target trend prediction function. In this example, the long short-term memory network (LSTM) is used;

[0073] The calculation method of local area resource redundancy is:

[0074]

[0075] In the above formula, U lre Indicates the resource redundancy of this local area, U lnum represents the number of agents in this local area, U lenum Indicates the number of agents performing tasks in this local area;

[0076] The calculation method of global layer task execution efficiency is:

[0077]

[0078] In the above formula, Tee represents the global task execution efficiency, L represents the number of local regions, and Tee l represents the task execution efficiency of the lth local area;

[0079] The calculation method of the global layer unmanned cluster task execution capability is:

[0080]

[0081] In the above formula, Te represents the task execution capability of the global unmanned cluster, Te l represents the task execution capability of the lth local area;

[0082] The global layer resource redundancy is calculated as follows:

[0083]

[0084] In the above formula, U re represents the global resource redundancy, U lre Represents the resource redundancy of the lth local area.

Claims

1. A distributed situation maintenance method for an unmanned cluster based on information extraction, wherein the unmanned cluster includes a plurality of intelligent agents; characterized in that: The following steps are involved: (1) Each agent establishes its own single-machine situation information; (2) Share the single-machine-layer situation information of all agents in the local area to the temporary information center node of the local area; (3) The local area temporary information center node extracts information to form the local layer situation information of the corresponding local area, and shares the local layer situation information with all intelligent agents in the corresponding local area; (4) After all local layer situation information is shared in their respective local areas, the local layer situation information extracted by all local area temporary information center nodes is shared to the global temporary information center node; (5) The global temporary information center node extracts information to form global situation information, and then shares the global situation information with all intelligent agents in the unmanned cluster; Each intelligent agent has communication and information sharing capabilities; the global layer situation information is the overall situation information of the unmanned cluster during the task execution process, including information on task execution efficiency, unmanned cluster task execution capability, and resource redundancy; the local layer situation information is the situation information of the local area, including information on task execution efficiency, sub-unmanned cluster task execution capability, target movement trend, and resource redundancy in the local area; the single machine layer situation information is the original information of the intelligent agent, including information on positioning, task execution capability, and covered targets; The selection rules for the local area temporary information center node are as follows: The agent located at the center of the local area communication topology takes over. If this agent fails, the agent located at the secondary center of the local area communication topology takes over, and so on. The information extraction methods described in step (3) and step (5) include statistics, weighting, analysis and prediction; wherein, in step (3), the local area temporary information center node extracts information to form the local layer situation information of the corresponding local area; wherein, the task execution efficiency is calculated as follows: In the above formula, Indicates the task execution efficiency of this local area, represents the number of agents in this local area, represents the task execution capability of the ith agent, Indicates the number of coverage targets in this local area; The calculation method of the task execution capability of the sub-unmanned cluster is: In the above formula, Indicates the task execution capability of this local area, represents the number of agents in this local area, represents the task execution capability of the i-th agent; The target movement trend is calculated as: In the above formula, Indicates the target movement trend in this local area. Represents the cluster center of all targets in the local area at the current moment, Represents the cluster center of all targets in the local area at the previous moment, and so on. represents the cluster center of all targets in the local area in the previous 9 moments, and f represents the target trend prediction function; Resource redundancy is calculated as follows: In the above formula, Indicates the resource redundancy of this local area. represents the number of agents in this local area, Indicates the number of agents performing tasks in this local area; In step (5), the global temporary information center node extracts information to form global situation information; the calculation method of task execution efficiency is: In the above formula, Represents the global task execution efficiency, represents the number of local regions, Indicates The task execution efficiency of a local area; The calculation method of unmanned cluster task execution capability is: In the above formula, Represents the task execution capability of the global unmanned cluster, Indicates The mission execution capability of a local area; Resource redundancy is calculated as follows: In the above formula, Indicates the global resource redundancy, Indicates The resource redundancy of a local area; All local temporary information center nodes form a communication network, and the selection rules of the global temporary information center node are: The intelligent agent located at the communication topology center of the communication network takes over. If the intelligent agent fails, the intelligent agent located at the secondary center of the communication topology of the communication network takes over, and so on.

Images