Intention target prediction method and system based on path recursive network model

Abstract

The present application relates to a method and system for predicting intention targets based on a path recursive network model, which is used to reduce the computational complexity of intention target prediction and meet the needs of on-the-spot decision-making in a real-time confrontation environment. The method includes: constructing a direct edge relationship between the first entity node and / or the second entity node according to a predefined interaction relationship between a first entity node included in the first camp and a second entity node included in the second camp; determining a direct intention probability matrix between the first entity node and the second entity node according to entity attribute information of the first entity node and the second entity node; determining an intention target prediction probability matrix between the first entity node and the second entity node according to the direct intention probability matrix and several levels of potential intention probability matrices between the first entity node and the second entity node.

Description

Technical Field

[0001] The present application relates to the technical field of intention target prediction, and in particular to an intention target prediction method and system based on a path recursive network model. Background Art

[0002] In the process of confrontational games, it is often necessary to predict the intentions and goals of each subject, so as to better understand the development of the situation and assist commanders in making command decisions. Target intention prediction methods can be divided into several categories: Bayesian network, gray correlation method, evidence theory, artificial intelligence-based methods, cognitive model-based methods, information theory-based methods, etc. Bayesian network, artificial intelligence method and cognitive model method are the most commonly used. Among them, Bayesian network models the problem of uncertain reasoning and is interpretable, but the analysis and calculation are more complicated; artificial intelligence methods require a large amount of training data, and generalization needs to be improved; cognitive model methods also have the same problems as artificial intelligence methods.

[0003] Existing intention target prediction methods tend to comprehensively analyze the impact of various factors on the subject's intention, which has high computational complexity and cannot meet the needs of on-the-spot decision-making in real-time confrontation environments. Summary of the invention

[0004] In view of this, in response to the above-mentioned technical problems, the present application provides an intention target prediction method and system based on a path recursive network model to reduce the computational complexity of intention target prediction and meet the needs of on-the-spot decision-making in a real-time confrontation environment.

[0005] In a first aspect, a method for predicting an intention target based on a path recursive network model is provided, the method comprising:

[0006] For a predefined interactive relationship between a first entity node included in the first camp and a second entity node included in the second camp, construct a direct edge relationship between the first entity node and / or the second entity node, wherein the direct edge relationship includes a strengthening relationship and a weakening relationship;

[0007] Determine a direct intention probability matrix between the first entity node and the second entity node according to entity attribute information of the first entity node and the second entity node, wherein the entity attribute information includes a gaming capability parameter of the first entity node or the second entity node, a power comparison parameter between a first camp and a second camp at a spatial point where the first entity node or the second entity node is located, a distance between the first entity node and the second entity node, a theoretical speed of the first entity node or the second entity node, a similarity parameter of a movement direction of the first entity node or the second entity node to a direction of a target entity node, and a preset adjustment coefficient;

[0008] According to the direct intention probability matrix and several levels of potential intention probability matrices between the first entity node and the second entity node, the intention target prediction probability matrix between the first entity node and the second entity node is determined, wherein any level of potential intention probability matrix is ​​determined based on the entity attribute information of the first entity node, the second entity node and the adjacent entity nodes, wherein the first entity node is connected to the second entity node through several levels of adjacent entity nodes.

[0009] In one embodiment, for an entity node The power comparison parameter is based on the entity node The potential field of the spatial point is determined, where the entity node The first camp potential field of the space point Determined according to the following formula:

[0010]

[0011] Entity Node The second camp potential field of the space point Determined according to the following formula:

[0012]

[0013] Indicates the distance from the entity node Set the collection of all first-camp entity nodes within the range. Indicates the distance from the entity node Set the collection of all second-camp entity nodes within the range; For entity nodes With entity node The distance between For predefined entity nodes Gaming ability parameters.

[0014] In one embodiment, if the entity node If it is the first entity node, the power comparison parameter between the first camp and the second camp at the spatial point where it is located is determined according to the following formula:

[0015]

[0016] If the entity node If it is the second entity node, the power comparison parameter between the first camp and the second camp at the spatial point where it is located is determined according to the following formula:

[0017] .

[0018] In one embodiment, according to the entity attribute information of the first entity node and the second entity node, the direct intention probability matrix of the first entity node to the second entity node is determined according to the following formula:

[0019]

[0020] in: for function, and is the preset adjustment coefficient, For entity nodes The direction of its own motion and the direction to the entity node Similarity of direction, For entity nodes With entity node The distance For entity nodes Theoretical speed, if the entity node is a fixed position node, then To launch weapons or detect the speed of electromagnetic waves, Represents a pre-set entity node Importance parameter;

[0021]

[0022] Represents an entity node From historical moments To the current time The average value of the velocity direction, Represents an entity node From historical moments To the current time The direction of velocity, Represents an entity node From historical moments To the current time The average velocity scalar, Represents an entity node To entity node The direction vector, Represents the calculation operator of vector similarity, Represents variance.

[0023] In one embodiment, based on the direct intention probability matrix and the potential intention probability matrices of the first entity node and the second entity node, the intention target prediction probability matrix between the first entity node and the second entity node is determined according to the following formula: :

[0024]

[0025] in, is the adjacency matrix of the direct intent probability matrix, and Elements Indicates that from the entity node To entity node of The sum of the level potential intention probability matrices.

[0026] In a second aspect, a system for predicting an intention target based on a path recursive network model is provided, the system comprising:

[0027] A construction module, configured to construct a direct edge relationship between the first entity node included in the first camp and the second entity node included in the second camp, with respect to a predefined interactive relationship between the first entity node included in the first camp and the second entity node included in the second camp, wherein the direct edge relationship includes an enhancement relationship and a weakening relationship;

[0028] A direct intention determination unit, configured to determine a direct intention probability matrix between the first entity node and the second entity node according to entity attribute information of the first entity node and the second entity node, wherein the entity attribute information includes a gaming capability parameter of the first entity node or the second entity node, a power comparison parameter between a first camp and a second camp at a spatial point where the first entity node or the second entity node is located, a distance between the first entity node and the second entity node, a theoretical speed of the first entity node or the second entity node, a similarity parameter of a motion direction of the first entity node or the second entity node to a direction of a target entity node, and a preset adjustment coefficient;

[0029] A prediction unit is used to determine the intention target prediction probability matrix between the first entity node and the second entity node based on the direct intention probability matrix and several levels of potential intention probability matrices between the first entity node and the second entity node, wherein any level of potential intention probability matrix is ​​determined based on entity attribute information of the first entity node, the second entity node and adjacent entity nodes, and wherein the first entity node is connected to the second entity node through several levels of adjacent entity nodes.

[0030] In one embodiment, for an entity node The power comparison parameter is based on the entity node The potential field of the spatial point is determined;

[0031] The direct intention determination unit is used to determine the entity node according to the following formula The first camp potential field of the space point :

[0032]

[0033] Determine the entity node according to the following formula The second camp potential field of the space point :

[0034]

[0035] Indicates the distance from the entity node Set the collection of all first-camp entity nodes within the range. Indicates the distance from the entity node Set the collection of all second-camp entity nodes within the range; For entity nodes With entity node The distance between For predefined entity nodes Gaming ability parameters.

[0036] In one embodiment, the direct intention determination unit is used if the entity node If it is the first entity node, the power comparison parameter between the first camp and the second camp at the spatial point where it is located is determined according to the following formula:

[0037] If the entity node If it is the second entity node, the power comparison parameter between the first camp and the second camp at the spatial point where it is located is determined according to the following formula:

[0038] .

[0039] In one embodiment, the direct intention determination unit is used to determine the direct intention probability matrix of the first entity node to the second entity node according to the following formula:

[0040]

[0041] in: for function, and is the preset adjustment coefficient, For entity nodes The direction of its own motion and the direction to the entity node Similarity of direction, For entity nodes With entity node The distance For entity nodes Theoretical speed, if the entity node is a fixed position node, then To launch weapons or detect the speed of electromagnetic waves, Represents a pre-set entity node Importance parameter;

[0042] Represents an entity node From historical moments To the current time The average value of the velocity direction, Represents an entity node From historical moments To the current time The direction of velocity, Represents an entity node From historical moments To the current time The average velocity scalar, Represents an entity node To entity node The direction vector, Represents the calculation operator of vector similarity, Represents variance.

[0043] In one embodiment, the prediction unit is used to determine the intended target prediction probability matrix between the first entity node and the second entity node according to the following formula: :

[0044]

[0045] in, is the adjacency matrix of the direct intent probability matrix, and Elements Indicates that from the entity node To entity node of The sum of the level potential intention probability matrices.

[0046] The above-mentioned intention target prediction method based on the path recursive network model adopts a network representation method to abstract all subjects into entity nodes, and abstract the relationship between them into edges. The connection weight of the network, that is, the size of direct intention, is obtained through modeling analysis; the potential intention of each subject is further determined through the path recursion of the matrix; finally, the direct intention and potential intention are comprehensively considered to achieve a fast and accurate prediction of the subject's intention target. In the above process, since the potential influence of the intentions between the entity nodes is taken into account and the calculation amount is small, it meets the needs of on-the-spot decision-making in a real-time confrontation environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 A schematic diagram of an implementation flow of an intention target prediction method based on a path recursive network model in one embodiment;

[0048] Figure 2 In one embodiment, the entity node To entity node Schematic diagram of the connection edge;

[0049] Figure 3 In one embodiment, a node-to-node For Node Schematic diagram of potential intentions;

[0050] Figure 4 For entity nodes , , , Schematic diagram of the connection edge;

[0051] Figure 5 FIG. 4 is a diagram showing the internal structure of a computing device in one embodiment. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0053] like Figure 1 As shown, it is a schematic diagram of the implementation process of the intention target prediction method based on the path recursive network model provided by the present application, which includes the following steps:

[0054] S11. Construct a direct edge relationship between the first entity node and / or the second entity node according to a predefined interaction relationship between the first entity node included in the first camp and the second entity node included in the second camp.

[0055] In this step, we first construct the direct intention network structure of the game based on the prior information. Specifically, we regard each entity as an entity node, with a common Take entity nodes as an example. Different camps are distinguished by labels. The entity node label of the first camp is , the entity node label of the second camp is For the convenience of description, this application refers to the entity nodes of the first camp as first entity nodes, and the entity nodes of the second camp as second entity nodes.

[0056] Obtain the common interaction relationship between entity nodes. When implementing it, it can be converted into prior knowledge based on common sense or military knowledge to build a direct edge relationship between entity nodes. Specifically, if there is a fixed action relationship within the same camp, there is a directed edge between the fixed entity nodes, from the action initiating entity node to the action receiving entity node, and the symbol "+" is given to represent enhancement. For example, the first group (entity node) of the same camp ) and the second cluster (physical nodes ) has a relationship of covering and being covered, then the entity node is given To entity node A directed “+” edge.

[0057] If there is a common attack / interference / detection relationship between the first entity node and the second entity node, a directed “-” edge will be created between the first entity node and the second entity node, representing weakening, with the direction from the action initiating entity node to the action receiving entity node. ) and the first cluster of the second camp (physical nodes ) have a mutual attack relationship, then assign entity nodes To entity node , entity node To entity node A directed “-” edge. Figure 2 As shown, the directed “+” edges are gray and the directed “-” edges are black.

[0058] S12: Determine a direct intention probability matrix between the first entity node and the second entity node according to entity attribute information of the first entity node and the second entity node.

[0059] First, the entity attribute information corresponding to the entity node is determined according to the entity characteristics. In specific implementation, the entity vertical information may include the gaming ability parameter of the first entity node or the second entity node, the power comparison parameter between the first camp and the second camp at the spatial point where the first entity node or the second entity node is located, the distance between the first entity node and the second entity node, the theoretical speed of the first entity node or the second entity node, the similarity parameter of the first entity node or the second entity node's own movement direction to the target entity node direction and a pre-set adjustment coefficient, etc.

[0060] In specific implementation, the set of all entity nodes is For example, Indicates The attribute vector of an entity node includes the location, speed, importance and entity inherent attributes of the entity node. Given by humans a priori, the value range is , the closer The higher the importance, the higher the level. The inherent attributes of entities include strike capability, defense capability, detection capability, etc. These inherent attributes of entities can be further divided into smaller capabilities. For example, strike capability can be divided into short-range strike capability and long-range strike capability, or point strike capability and surface strike capability, etc. In this application, these attributes are uniformly abstracted into two categories: game capability parameters With entity node The power comparison parameter between the first camp and the second figure at the spatial point .

[0061] (1) Gaming ability parameters That is, the capabilities of the entity node itself, including the ability to attack, detect, support, etc., and the defense capabilities of the entity itself. In this application, the external forces and the defense capabilities are unified as game capability parameters. The stronger the external forces of the entity, the stronger the game capability; the smaller the degree of damage to the entity itself and the stronger the ability to resist attacks, the stronger the defense.

[0062] (2) Entity Node The power comparison parameters of the first camp and the second camp at the spatial point Affects the probability of physical nodes being attacked or detected. In this application, an artificial potential field is used Indicates the strength of the first camp / second camp. When implementing it, the weighted average can be used to get the entity node The potential field size of the space point where the entity node is located. The first camp potential field of the space point Determined according to the following formula:

[0063]

[0064] Entity Node The second camp potential field of the space point Determined according to the following formula:

[0065]

[0066] in, Indicates the distance from the entity node Set the collection of all first-camp entity nodes within the range. Indicates the distance from the entity node Set the collection of all second-camp entity nodes within the range; For entity nodes With entity node The distance between For predefined entity nodes Gaming ability.

[0067] If the entity node Belong to the first camp, then ,here For entity nodes The first camp potential field of the space point, For entity nodes The second camp potential field of the spatial point; if the entity node Belong to the second camp, then ,here For entity nodes The second camp potential field of the space point is, For entity nodes The first camp potential field of the spatial point.

[0068] After the attributes of the entity nodes themselves are given, the direct intention probability matrix between the entity nodes is further calculated.

[0069] During specific implementation, the connection weight between two entity nodes can be determined based on the bargaining power of both entity nodes, the power comparison between the first camp and the second camp at the spatial point where the node is located, the direction of movement, the distance and the importance, so as to obtain the strength of the direct intention between the two entity nodes. Specifically, the stronger the bargaining power, the greater the bargaining intention; the stronger the power of the same camp at the spatial point where the entity node is located, the greater the possibility of being detected or attacked; the more similar the direction from the entity node to the target node is to the direction of the entity node itself, the greater the bargaining intention; the higher the importance of the target node, the greater the bargaining intention; the greater the distance, the smaller the bargaining intention. In this application, the entity nodes can be determined according to the following formula: For node entities The direct intention probability matrix :

[0070]

[0071] in for function, and is the preset adjustment coefficient, For entity nodes The direction of its own motion and the direction to the entity node Similarity of direction, For entity nodes With entity node The distance For entity nodes Theoretical speed (i.e. physical node The standard speed, not the actual speed ), if the entity node For a fixed position node, let the theoretical speed To launch weapons or detect the speed of electromagnetic waves, is a priori given entity node Gaming ability parameters, Represents a pre-set entity node Importance parameter. Among them, The expression is:

[0072]

[0073] in, Represents an entity node From historical moments To the current time The average value of the velocity direction, Represents an entity node From historical moments To the current time The direction of velocity, Represents an entity node From historical moments To the current time The average velocity scalar, Represents an entity node To entity node The direction vector, The operator representing the vector similarity calculation can be implemented using Euclidean distance or cosine similarity. Represents variance. When the entity node When in mobile state, It is represented by vector similarity and actual speed variance; if the entity node From historical moments To the current time The speed is always 0 ( For fixed position nodes or has not moved in the past period of time), then let .

[0074] S13. Determine an intent target prediction probability matrix between the first entity node and the second entity node based on the direct intent probability matrix and several levels of potential intent probability matrices between the first entity node and the second entity node.

[0075] Among them, any level of potential intention probability matrix is ​​determined based on the entity attribute information of the first entity node, the second entity node and the adjacent entity node, wherein the first entity node is connected to the second entity node through several levels of adjacent entity nodes.

[0076] In specific implementation, the direct intention probability matrix between the first entity node and the second entity node can be obtained through step S12. In fact, there are multi-level linkage effects between entities. The intention of the entity node is not only related to its intention target, but also to the intention of the intention target. Therefore, to obtain accurate intention prediction results, it is necessary to further analyze the multi-level potential intentions of each entity, that is, it is necessary to perform path recursion on the intention. Assume that the path recursion result of the first-level potential intention is , the path recursion result of the secondary potential intention is , the path recursion result of the three-level potential intention is .

[0077] In this step, based on the direct intention probability matrix and the potential intention probability matrix between the first entity node and the second entity node, the intention target prediction probability matrix between the first entity node and the second entity node can be determined according to the following formula: :

[0078]

[0079] in, is the adjacency matrix of the direct intent probability matrix, and Elements Indicates that from the entity node To entity node of The sum of the level potential intention probability matrices.

[0080] Taking the first-level potential intention as an example, that is, considering the influence of the intention of the intention target on the node intention, we adopt

[0081]

[0082] The total intention obtained by combining the direct intention and the first-level potential intention ,in is the adjacency matrix of the direct intent probability matrix, and Elements Represents a slave node To Node The sum of first-level latent intentions (i.e., the intentions of the intention targets).

[0083] According to the embodiment of the present invention, the role of the edge is divided into strengthening (+) and weakening (-). Through the entity node Point to entity node There are four combinations of two-hop links: ++, --, +-, -+: (1) ,but right There is a positive reinforcement effect, so ++ gives +; (2) ,but inhibition , inhibition , right It has a positive reinforcing effect, so - gets +; (3) , inhibition , Promote , inhibition , so +- gives -; (4) , Promote , inhibition , inhibition , so -+ gives +. From this we can see that The calculation principle conforms to common sense.

[0084] Take the example to better understand the present invention, Figure 3 As shown, the entity With base When the entity Approaching the base When Serious threat to base , thus protecting the base Other entities For Entity The attack intention is strong. In matrix calculation, due to the entity With base The distance is small, so the base For Entity The weakening effect Large; entity protecting the base With base The distance is small, so the entity To base The enhancing effect Big; then protect the entity of the base Through two hops (base-entity ) path recursively connects the entity The weakening effect is great:

[0085]

[0086] Both the direct impact between entities and the indirect impact of two hops are considered. In specific implementation, the level of potential intention impact can be determined according to actual needs, for example, considering the calculation complexity and impact attenuation reasons.

[0087] Get the intention relationship matrix between all subjects After that, you can Determine the subject's intended goal. Matrix No. Line represents the body The size of the game intention towards other subjects. The greater the intention, the stronger the subject The more inclined to implement the game, the more likely the subject is to Main body If the intended goal , then the subject No intended goal.

[0088] The intention target prediction method based on the path recursive network model provided in this application adopts an interpretable method based on the path recursive network model. Compared with the conventional intention prediction method, it comprehensively considers the influence of potential intentions and realizes more concise modeling and calculation. In this technical scheme, a direct intention calculation model is constructed, which can be used as the knowledge input of other studies, for example, as the calculation basis of potential intentions in the present invention. In this technical scheme, by setting the number of layers of path recursion, that is, multi-level potential intentions, the depth of influence of potential factors on intentions can be vertically extended. In this application, the power comparison of the first camp and the second camp at the spatial point where the entity node is located is used as the basis for calculating the probability of being attacked / detected. Compared with the conventional calculation method based on defense capabilities, it can more realistically approach the real scene and improve the accuracy of modeling. This application comprehensively summarizes the factors affecting intention by comprehensively considering the attributes, position, speed, importance, and power comparison of the first camp and the second camp of the node, and abstracts these factors into several key parameters through classification and induction, forming a generalized process of intention analysis.

[0089] The following takes a specific red-blue drone confrontation scenario as an example to illustrate the specific implementation method of this application.

[0090] Figure 4 Midpoint , , , They represent the base of the first camp, the reconnaissance drone of the second camp, the mobile aircraft group of the first camp, and the reconnaissance and attack drone of the second camp. The reconnaissance drone group of the second camp performs reconnaissance missions on the base of the first camp. right The one-way arrow is used as a representation, indicating right The second camp is to provide cover and assist in reconnaissance with the drone swarm. right The one-way arrow is used as a representation, indicating right The first level intention; the first camp mobile aircraft is ready to take off to protect the first camp base. right The one-way arrow is used as a representation, indicating right At this time, because the first camp's mobile aircraft group intends to protect the first camp's base, the second camp's reconnaissance and strike drones need to attack the first camp's mobile aircraft group first to ensure the smooth completion of the second camp's reconnaissance drone's reconnaissance mission. Therefore, there is a second-level intention of the second camp's reconnaissance and strike drones against the first camp's mobile aircraft group, that is, right One-way arrow. The main intention and goal of this application are taken as an example to illustrate the specific implementation steps of this application.

[0091] On the basis of the fixed collocation network constructed a priori, the direct intention probability matrix is ​​calculated according to the position, speed, fixed attributes and other parameters of each node, and the direct intention probability matrix is ​​obtained.

[0092]

[0093] Calculating first-level latent intent

[0094]

[0095] Comprehensive calculation of the overall intention

[0096]

[0097] According to the principle of intention maximization, the intention target of each subject is obtained, such as node The intended goal is , then the node The maximum intention goal is .

[0098] It should be understood that although Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 1At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0099] Based on the same technical concept, the present application also provides an intention target prediction system based on a path recursive network model, the system comprising:

[0100] A construction module, configured to construct a direct edge relationship between the first entity node included in the first camp and the second entity node included in the second camp, with respect to a predefined interactive relationship between the first entity node included in the first camp and the second entity node included in the second camp, wherein the direct edge relationship includes an enhancement relationship and a weakening relationship;

[0101] A direct intention determination unit, configured to determine a direct intention probability matrix between the first entity node and the second entity node according to entity attribute information of the first entity node and the second entity node, wherein the entity attribute information includes a gaming capability parameter of the first entity node or the second entity node, a power comparison parameter between a first camp and a second camp at a spatial point where the first entity node or the second entity node is located, a distance between the first entity node and the second entity node, a theoretical speed of the first entity node or the second entity node, a similarity parameter of a motion direction of the first entity node or the second entity node to a direction of a target entity node, and a preset adjustment coefficient;

[0102] A prediction unit is used to determine the intention target prediction probability matrix between the first entity node and the second entity node based on the direct intention probability matrix and several levels of potential intention probability matrices between the first entity node and the second entity node, wherein any level of potential intention probability matrix is ​​determined based on entity attribute information of the first entity node, the second entity node and adjacent entity nodes, and wherein the first entity node is connected to the second entity node through several levels of adjacent entity nodes.

[0103] In one embodiment, for an entity node The power comparison parameter is based on the entity node The potential field of the spatial point is determined;

[0104] The direct intention determination unit is used to determine the entity node according to the following formula The first camp potential field of the space point :

[0105]

[0106] Determine the entity node according to the following formula The second camp potential field of the space point :

[0107]

[0108] Indicates the distance from the entity node Set the collection of all first-camp entity nodes within the range. Indicates the distance from the entity node Set the collection of all second-camp entity nodes within the range; For entity nodes With entity node The distance between For predefined entity nodes Gaming ability parameters.

[0109] In one embodiment, the direct intention determination unit is used if the entity node If it is the first entity node, the power comparison parameter between the first camp and the second camp at the spatial point where it is located is determined according to the following formula:

[0110]

[0111] If the entity node If it is the second entity node, the power comparison parameter between the first camp and the second camp at the spatial point where it is located is determined according to the following formula:

[0112] .

[0113] In one embodiment, the direct intention determination unit is used to determine the direct intention probability matrix of the first entity node to the second entity node according to the following formula:

[0114]

[0115] in: for function, and is the preset adjustment coefficient, For entity nodes The direction of its own motion and the direction to the entity node Similarity of direction, For entity nodes With entity node The distance For entity nodes Theoretical speed, if the entity node is a fixed position node, then To launch weapons or detect the speed of electromagnetic waves, Represents a pre-set entity node Importance parameter;

[0116]

[0117] Represents an entity node From historical moments To the current time The average value of the velocity direction, Represents an entity node From historical moments To the current time The direction of velocity, Represents an entity node From historical moments To the current time The average velocity scalar, Represents an entity node To entity node The direction vector, Represents the calculation operator of vector similarity, Represents variance.

[0118] In one embodiment, the prediction unit is used to determine the intended target prediction probability matrix between the first entity node and the second entity node according to the following formula: :

[0119]

[0120] in, is the adjacency matrix of the direct intent probability matrix, and Elements Indicates that from the entity node To entity node of The sum of the level potential intention probability matrices.

[0121] In one embodiment, the present application provides a computing device, which may be a server, and its internal structure diagram may be as follows: Figure 5As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for predicting an intention target based on a path recursive network model is implemented.

[0122] In one embodiment, the present application provides a computer device, which may be a terminal, and its internal structure diagram may be as follows: Figure 5 As shown. The computer device includes a processor, a memory, a network interface, a display screen and an input device connected through a system bus, a computer program is stored in the memory, and the processor implements the steps of the method described in any of the above embodiments when executing the computer program. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for predicting an intention target based on a path recursive network model is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a button, trackball or touchpad set on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0123] Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0124] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described in any of the above embodiments are implemented.

[0125] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0126] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0127] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A method for predicting intention targets based on a path recursive network model, characterized in that: The method is applied to a drone confrontation scenario, and the method comprises: According to a predefined interactive relationship between a first entity node included in the first camp and a second entity node included in the second camp, a direct edge relationship between the first entity node and / or the second entity node is constructed, wherein the direct edge relationship includes an enhancement relationship and a weakening relationship, wherein the first entity node includes a base and a mobile aircraft, and the second entity node includes a reconnaissance drone and a reconnaissance and strike drone; Determine a direct intention probability matrix between the first entity node and the second entity node according to entity attribute information of the first entity node and the second entity node, wherein the entity attribute information includes a gaming capability parameter of the first entity node or the second entity node, a power comparison parameter between a first camp and a second camp at a spatial point where the first entity node or the second entity node is located, a distance between the first entity node and the second entity node, a theoretical speed of the first entity node or the second entity node, a similarity parameter of a movement direction of the first entity node or the second entity node to a direction of a target entity node, and a preset adjustment coefficient; According to the direct intention probability matrix and several levels of potential intention probability matrices between the first entity node and the second entity node, the intention target prediction probability matrix between the first entity node and the second entity node is determined, wherein any level of potential intention probability matrix is ​​determined based on the entity attribute information of the first entity node, the second entity node and the adjacent entity nodes, wherein the first entity node is connected to the second entity node through several levels of adjacent entity nodes.

2. The method according to claim 1, characterized in that For any entity node The power comparison parameter is based on the entity node The potential field of the spatial point is determined, where the entity node The first camp potential field of the space point Determined according to the following formula: Entity Node The second camp potential field of the space point Determined according to the following formula: Indicates the distance from the entity node Set the collection of all first-camp entity nodes within the range. Indicates the distance from the entity node Set the collection of all second-camp entity nodes within the range; For entity nodes With entity node The distance between For predefined entity nodes Gaming ability parameters.

3. The method according to claim 2, characterized in that If the entity node If it is the first entity node, the power comparison parameter between the first camp and the second camp at the spatial point where it is located is determined according to the following formula: ； If the entity node If it is the second entity node, the power comparison parameter between the first camp and the second camp at the spatial point where it is located is determined according to the following formula: 。 4. The method according to claim 3, characterized in that According to the entity attribute information of the first entity node and the second entity node, the direct intention probability matrix of the first entity node to the second entity node is determined according to the following formula: in: is the sigmoid function, and is the preset adjustment coefficient, For entity nodes The direction of its own motion and the direction to the entity node Similarity of direction, For entity nodes With entity node The distance For entity nodes Theoretical speed, if the entity node is a fixed position node, then To launch weapons or detect the speed of electromagnetic waves, Represents a pre-set entity node Importance parameter; Represents an entity node From historical moments To the current time The average value of the velocity direction, Represents an entity node From historical moments To the current time The direction of velocity, Represents an entity node From historical moments To the current time The average velocity scalar, Represents an entity node To entity node The direction vector, Represents the operator for calculating vector similarity, Represents variance.

5. The method according to claim 1, characterized in that According to the direct intention probability matrix and the potential intention probability matrices of the first entity node and the second entity node, the intention target prediction probability matrix between the first entity node and the second entity node is determined according to the following formula: : in, is the adjacency matrix of the direct intent probability matrix, and Elements Indicates that from the entity node To entity node of The sum of the level potential intention probability matrices.

6. An intention target prediction system based on a path recursive network model, characterized in that: The system comprises: A construction module, configured to construct a direct edge relationship between the first entity node and / or the second entity node according to a predefined interactive relationship between the first entity node included in the first camp and the second entity node included in the second camp, wherein the direct edge relationship includes an enhancement relationship and a weakening relationship; A direct intention determination unit, configured to determine a direct intention probability matrix between the first entity node and the second entity node according to entity attribute information of the first entity node and the second entity node, wherein the entity attribute information includes a gaming capability parameter of the first entity node or the second entity node, a power comparison parameter between a first camp and a second camp at a spatial point where the first entity node or the second entity node is located, a distance between the first entity node and the second entity node, a theoretical speed of the first entity node or the second entity node, a similarity parameter of a motion direction of the first entity node or the second entity node to a direction of a target entity node, and a preset adjustment coefficient; A prediction unit is used to determine the intention target prediction probability matrix between the first entity node and the second entity node based on the direct intention probability matrix and several levels of potential intention probability matrices between the first entity node and the second entity node, wherein any level of potential intention probability matrix is ​​determined based on entity attribute information of the first entity node, the second entity node and adjacent entity nodes, and wherein the first entity node is connected to the second entity node through several levels of adjacent entity nodes.

7. The system according to claim 6, characterized in that For any entity node The power comparison parameter is based on the entity node The potential field of the spatial point is determined; The direct intention determination unit is used to determine the entity node according to the following formula The first camp potential field of the space point : Determine the entity node according to the following formula The second camp potential field of the space point : Indicates the distance from the entity node Set the collection of all first-camp entity nodes within the range. Indicates the distance from the entity node Set the collection of all second-camp entity nodes within the range; For entity nodes With entity node The distance between For predefined entity nodes Gaming ability parameters.

8. The system according to claim 7, characterized in that The direct intention determination unit is used if the entity node If it is the first entity node, the power comparison parameter between the first camp and the second camp at the spatial point where it is located is determined according to the following formula: If the entity node If it is the second entity node, the power comparison parameter between the first camp and the second camp at the spatial point where it is located is determined according to the following formula: 。 9. The system according to claim 8, characterized in that The direct intention determination unit is used to determine the direct intention probability matrix of the first entity node to the second entity node according to the following formula: in: is the sigmoid function, and is the preset adjustment coefficient, For entity nodes The direction of its own motion and the direction to the entity node Similarity of direction, For entity nodes With entity node The distance For entity nodes Theoretical speed, if the entity node is a fixed position node, then To launch weapons or detect the speed of electromagnetic waves, Represents a pre-set entity node Importance parameter; Represents an entity node From historical moments To the current time The average value of the velocity direction, Represents an entity node From historical moments To the current time The direction of velocity, Represents an entity node From historical moments To the current time The average velocity scalar of Represents an entity node To entity node The direction vector, Represents the operator for calculating vector similarity, Represents variance.

10. The system according to claim 9, characterized in that The prediction unit is used to determine the intended target prediction probability matrix between the first entity node and the second entity node according to the following formula: : in, is the adjacency matrix of the direct intent probability matrix, and Elements Indicates that from the entity node To entity node of The sum of the level potential intention probability matrices.

Images