A method for representing battlefield environment knowledge

Abstract

The present invention belongs to the technical field of knowledge graphs, and specifically relates to a method for representing battlefield environment knowledge, including: extracting battlefield environment knowledge based on battlefield environment data, where the battlefield environment knowledge includes battlefield environment entity knowledge, battlefield environment event knowledge, battlefield environment impact process knowledge, and battlefield environment service decision-making knowledge; representing each type of the above knowledge respectively based on a knowledge hypergraph to obtain knowledge hypergraphs of the corresponding battlefield environment entity layer, battlefield environment event layer, battlefield environment impact process layer, and battlefield environment service decision-making layer; in each layer, representing the relationship between the nodes in the knowledge hypergraph of a certain layer and the nodes in the knowledge hypergraphs of other layers using hyperedges / edges to obtain a cross-layer knowledge hypergraph; associating and visually displaying the obtained knowledge hypergraphs corresponding to each layer and the cross-layer knowledge hypergraph; thereby, the present invention solves the problem that it is difficult for the knowledge representation model in the prior art to comprehensively reflect the association relationships among all elements.

Description

Technical Field

[0001] The present invention belongs to the technical field of knowledge graphs, and particularly relates to a method for representing battlefield environment knowledge. Background Art

[0002] The battlefield environment is the objective environment within a fixed combat area, excluding personnel and weapons and equipment. It covers the "entire domain" of land, sea, sky, and space in terms of space, and includes "multi-dimensional" environments such as geography, meteorology, electromagnetics, network, and nuclear, biological, and chemical in terms of elements. The simulation modeling and expression based on multi-source heterogeneous battlefield environment data is an important means for commanders to understand the objective environment, formulate combat plans, and dispel the battlefield fog. The types of battlefield environment sensors integrating air, space, land, and sea are gradually becoming rich, the data recording means are becoming more comprehensive, and the data types are also increasing. Processing and processing massive battlefield environment data to achieve the improvement from data to knowledge and make machines more intelligent is an important way for current battlefield environment intelligent service support.

[0003] A knowledge graph is a knowledge representation method in the field of artificial intelligence. Its goal is to achieve the cognitive intelligence of machines. The core is to form a knowledge base by extracting, modeling, and representing multi-source heterogeneous data, and it shines in directions such as data association search, intelligent question answering, and decision analysis. In the military field, it mainly uses knowledge graphs to conduct in-depth research on the organization of military equipment knowledge, military knowledge, the construction of target knowledge graphs, etc., and is used in fields such as intelligent matching of combat plans, intelligent comprehensive identification of sea and air targets, and natural language question answering of military knowledge. The research and application of knowledge graphs is an inevitable trend in the development of military intelligence, but the construction and application of battlefield environment knowledge graphs have not been systematically studied. What is of reference significance for the research of battlefield environment knowledge graphs is the research of geographical knowledge graphs, which aims to solve the problem of "geographical information explosion and knowledge scarcity". The research in aspects such as the construction of geographical knowledge ontology models, event knowledge representation, geographical knowledge extraction, storage indexing, and intelligent services has important reference significance for battlefield environment knowledge modeling and application. However, most knowledge representation models are difficult to comprehensively reflect the relationships between all relevant elements or the relationships between element attributes. The relationships between the elements represented by the model and the elements are incomplete. For example, in the model represented by triples, the representation method of triples makes the fragmentation of battlefield environment knowledge relatively serious, making it difficult to reflect the temporal characteristics and scene characteristics of battlefield environment knowledge and lacking some correlation relationships between elements. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for representing battlefield environment knowledge to solve the problem that the knowledge representation model in the prior art is difficult to comprehensively reflect the correlation relationships between all elements.

[0005] To solve the above technical problems, the technical solutions provided by the present invention and the corresponding beneficial effects of the technical solutions are as follows:

[0006] A method for representing battlefield environment knowledge of the present invention includes the following steps:

[0007] 1) Based on battlefield environment data, extract battlefield environment knowledge therefrom. The battlefield environment knowledge includes battlefield environment entity knowledge, battlefield environment event knowledge, battlefield environment impact process knowledge, and battlefield environment service decision-making knowledge;

[0008] The battlefield environment entity knowledge includes an objectified logical description of each element of the battlefield environment, either independently or jointly, within a certain battlefield area; the battlefield environment event knowledge includes an event in which the behavior, spatial structure, attributes, or combination of battlefield environment elements within a certain battlefield area has changed, and the change has had an important impact on the combat unit or the environment itself; the battlefield environment impact process knowledge includes knowledge used in the fields of battlefield environment impact assessment and combat simulation and simulation, relying on relevant materials and expert experience, and forming knowledge including a marine environment model, an atmospheric environment model, a terrain impact model, and a comprehensive environment impact model; the battlefield environment service decision-making knowledge includes knowledge of the comprehensive effectiveness impact of comprehensive environmental elements including geography, meteorology, electromagnetics, and network on each stage of combat operations;

[0009] 2) Based on the knowledge hypergraph, represent the battlefield environment entity knowledge, battlefield environment event knowledge, battlefield environment impact process knowledge, and battlefield environment service decision-making knowledge respectively to obtain knowledge hypergraphs of the corresponding battlefield environment entity layer, battlefield environment event layer, battlefield environment impact process layer, and battlefield environment service decision-making layer;

[0010] 3) In the battlefield environment entity layer, battlefield environment event layer, battlefield environment impact process layer, and battlefield environment service decision-making layer, represent the relationship between the nodes in the knowledge hypergraph of a certain layer and the nodes in the knowledge hypergraphs of other layers using hyperedges / edges to obtain a cross-layer knowledge hypergraph;

[0011] 4) Parallelly connect and visually display the obtained knowledge hypergraphs of the battlefield environment entity layer, battlefield environment event layer, battlefield environment impact process layer, and battlefield environment service decision-making layer, as well as the cross-layer knowledge hypergraph.

[0012] The beneficial effects of the above technical solution are as follows: The battlefield environment knowledge modeling is the core basic work of the intelligent support for the battlefield environment. Modeling and analyzing various types of knowledge such as entities, events, influence processes, and decision-making services is a huge systematic project. The present invention mainly conducts a systematic analysis and modeling of the battlefield environment knowledge from the perspectives of the classification of the knowledge system, the construction of the knowledge hypergraph model, the construction and association of the knowledge hypergraph, etc., and realizes the comprehensive representation of the knowledge hypergraph of various types of knowledge from the single-element data, the relationships between single-layer elements to the association relationships between layers. Thus, it solves the problem that the fragmentation of the battlefield environment knowledge is relatively serious when using the triple representation method in the prior art, and it is difficult to reflect the temporal characteristics and scenario characteristics of the battlefield environment knowledge.

[0013] Further, the relationships in step 3) include the mapping relationship between the battlefield environment entity layer and the battlefield environment event layer, the mapping relationship between the battlefield environment event layer and the battlefield environment influence process layer, and the mapping relationship between the battlefield environment influence process layer and the battlefield environment service decision layer; the mapping relationship between the battlefield environment entity layer and the battlefield environment event layer is used to represent the relationships between entities included in the battlefield environment event knowledge and the battlefield environment knowledge; the mapping relationship between the battlefield environment event layer and the battlefield environment influence process layer is used to represent the influence on weapons and equipment or personnel caused by the events occurring in the battlefield environment; the mapping relationship between the battlefield environment influence process layer and the battlefield environment service decision layer is used to represent the decision-making influence of the influence parameters on weapons and equipment or personnel on combat operations.

[0014] The beneficial effects of the above technical solution are as follows: The present invention maps the separate battlefield environment entity layer and the battlefield environment event layer through the entities included in the battlefield environment event knowledge, and then, according to the influence on weapons and equipment or personnel caused by the events occurring in the battlefield environment, and finally, according to the decision-making influence of the influence parameters on weapons and equipment or personnel on combat operations, realizes the linkage relationship between layers and comprehensively reflects the association relationships of the mutual influences of various elements.

[0015] Further, the battlefield environment entity layer includes several battlefield environment entities, and the battlefield environment entities are represented using entity representation attributes; the entity representation attributes include entity type, location, area, time, status, and semantic relationship; the semantic relationship includes the relationships of all entity representation attributes of an entity; the semantic relationship is represented using a hyperedge / edge; each entity representation attribute except the semantic relationship is represented using a node; after the battlefield environment entity layer is represented based on the knowledge hypergraph, an undirected attribute hypergraph network is obtained.

[0016] The beneficial effects of the above technical solution are as follows: In the present invention, the relationships between entity-represented attributes and entities are represented by edges, which can intuitively represent the relationships between entities and attributes. Moreover, the relationships between the entity-represented attributes are represented by hyperedges, which can intuitively represent the scattered attributes with the same entity, making it more conducive to the utilization of knowledge and statistical analysis.

[0017] Furthermore, the battlefield environment event layer includes several battlefield environment events, and the battlefield environment events are represented by using event-represented attributes. The event-represented attributes include event type, time element, location element, event subject, event logical relationship, action element, state set, and event description. The event logical relationship includes the event-represented attribute values forming the event and the inheritance, development, and causal relationships between events, and the event logical relationship is represented by hyperedges / edges. Each event-represented attribute other than the event logical relationship is represented by a node. After the battlefield environment event layer is represented based on the knowledge hypergraph, a directed hypergraph network with a directed logical relationship is obtained.

[0018] Furthermore, the battlefield environment impact process layer includes several battlefield environment impact processes, and the battlefield environment impact processes are represented by using process-represented attributes. The process-represented attributes include environmental impact factors, personnel, weapons and equipment, affected objects, subject attributes, time, and impact weight relationships. The impact weight relationship includes the comprehensive impact of multiple environmental impact factors on weapons and equipment or personnel, and the impact weight relationship is represented by hyperedges / edges. Each process-represented attribute other than the impact weight relationship is represented by a node. After the battlefield environment impact process layer is represented based on the knowledge hypergraph, a weighted directed hypergraph network is obtained.

[0019] Furthermore, the battlefield environment service decision-making layer includes several battlefield environment service decisions, and the battlefield environment service decisions are represented by using decision-represented attributes. The decision-represented attributes include type, terrain, meteorology, ocean impact factors, affected entities, combat operations, regions, and process relationships. The process relationship includes the decision-making impact of the environment and equipment performance on combat operations, and the process relationship is represented by a hyperedge. Each decision-represented attribute other than the process relationship is represented by a node. After the battlefield environment service decision-making layer is represented based on the knowledge hypergraph, a fuzzy hypergraph network is obtained.

[0020] Furthermore, the hyperedge logic used in step 3) is expressed as:

[0021]

[0022] where, o i ∈G BE_Object ,e j ∈G BE_Event ,a k ∈GBE_Affect , d l ∈ G BE_Decision respectively represent any node in four layers of G BE_Object , G BE_Event , G BE_Affect , G BE_Decision , any node in four layers represents o i and e j whether there is a mapping relationship between them represents e i and a k whether there is a mapping relationship between them represents a k and d l whether there is a mapping relationship between them

[0023] Furthermore, the hyperedges include unordered hyperedges and ordered hyperedges, and the corresponding unordered hyperedges and ordered hyperedges are constructed according to preset rules

[0024] Furthermore, the formal representation of the knowledge hypergraph of the battlefield environment entity layer, battlefield environment event layer, battlefield environment impact process layer, and battlefield environment service decision layer in step 4) and the overall representation model used for representing the cross-layer knowledge hypergraph is as follows

[0025] G = {G BE_Object , G BE_Event , G BE_Affect , G BE_Decision , R}

[0026] where R represents the set of mapping relationships between layers; G BE_Object represents the battlefield environment entity layer, including the semantic hyperedge relationships between various entities in the battlefield environment; G BE_Event represents the battlefield environment event layer, including the causal, sequential, and inheritance event logical relationships between various events in the battlefield environment; G BE_Affect represents the battlefield environment impact process layer, including the impact factor and impact weight relationships of various entities in the battlefield environment; G BE_Decision represents the battlefield environment service decision layer, including the comprehensive environmental analysis models of combat operations, combat training, and weapon strike effect assessment and the process relationships of combat operations BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 is the schematic diagram of the classification of battlefield environment knowledge of the present invention

[0028] Figure 2 is the schematic diagram of the meta-modeling and relationships of various types of knowledge in the battlefield environment of the present invention

[0029] Figure 3 is the schematic diagram of the battlefield environment temporal hypergraph knowledge representation model of the present invention

[0030] Figure 4 is the hypergraph of battlefield environment knowledge of the present invention;

[0031] Figure 5 is the conceptual model of the multi - level and time - series knowledge hypergraph of the battlefield environment of the present invention;

[0032] Figure 6 is the schematic diagram of map element knowledge representation taking a certain city as an example of the present invention;

[0033] Figure 7 is the schematic diagram of event knowledge representation of the present invention;

[0034] Figure 8 is the knowledge hypergraph representation of the fuzzy influence diagram of the present invention;

[0035] Figure 9 is the schematic diagram of interaction with a three - dimensional scene based on atlas semantic nodes of the present invention;

[0036] Figure 10-1 is the schematic diagram of the hypergraph representation of the flood disaster event in the example of the hypergraph representation of events of the present invention;

[0037] Figure 10-2 is the schematic diagram of the hypergraph representation of the political event in the example of the hypergraph representation of events of the present invention;

[0038] Figure 11-1 is the schematic diagram of the hypergraph representation with the influencing factor being the hydrological factor in the example of the knowledge graph representation of the battlefield environment influence process of the present invention;

[0039] Figure 11-2 is the schematic diagram of the hypergraph representation with the influencing factor being the meteorological factor in the example of the knowledge graph representation of the battlefield environment influence process of the present invention;

[0040] Figure 11-3 is the schematic diagram of the hypergraph representation with the influencing factor being the soil quality factor in the example of the knowledge graph representation of the battlefield environment influence process of the present invention;

[0041] Figure 11-4 is the schematic diagram of the hypergraph representation with the influencing factor being the vegetation factor in the example of the knowledge graph representation of the battlefield environment influence process of the present invention;

[0042] Figure 12-1 is the schematic diagram of an example of an ordered hyper - edge of the present invention;

[0043] Figure 12-2 is the schematic diagram of another example of an ordered hyper - edge of the present invention;

[0044] Figure 13-1 is the schematic diagram of an example of an unordered hyper - edge of the present invention;

[0045] Figure 13-2 It is a schematic diagram of another example of unordered hyperedges of the present invention;

[0046] Figure 13-3 It is a schematic diagram of yet another example of unordered hyperedges of the present invention;

[0047] Figure 14 It is a schematic diagram of an example of hyperedge modeling of battlefield environment knowledge of the present invention;

[0048] Figure 15-1 It is the first schematic diagram of the hypergraph representation of nodes and non-independent nodes of the present invention;

[0049] Figure 15-2 It is the second schematic diagram of the hypergraph representation of nodes and non-independent nodes of the present invention;

[0050] Figure 15-3 It is the third schematic diagram of the hypergraph representation of nodes and non-independent nodes of the present invention;

[0051] Figure 15-4 It is the fourth schematic diagram of the hypergraph representation of nodes and non-independent nodes of the present invention. Detailed implementation manners

[0052] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0053] Method embodiment:

[0054] The following is an explanation in combination with the steps of the present invention.

[0055] First step: Based on battlefield environment data, extract battlefield environment knowledge, where the battlefield environment knowledge includes battlefield environment entity knowledge, battlefield environment event knowledge, battlefield environment impact process knowledge, and battlefield environment service decision-making knowledge. The following is an introduction to the logical representation of battlefield environment knowledge.

[0056] First, an explanation is given for the classification of battlefield environment knowledge. Battlefield environment knowledge mainly depends on battlefield environment information, and the modality of information storage determines the form of knowledge storage. The main knowledge storage modalities include: text, images, audio, and video, etc. The battlefield environment knowledge storage modality mainly relies on the structured storage form of battlefield environment information storage, such as maps, images, military geographic manuals, surveillance audio / video, combat operation rules, etc. On the basis of modalities such as text, images, audio / video, etc., it records the domain attributes and characteristics unique to the battlefield environment itself. For example, maps and images are essentially images that record information such as geographical spatial positions and attributes. Battlefield environment knowledge can be classified into environmental basic knowledge, environmental impact process knowledge, and environmental service decision-making knowledge according to the application direction, as Figure 1 shown.

[0057] (1) Basic environmental knowledge.

[0058] Basic environmental knowledge mainly describes and represents the basic objective information of the environment, which is the result of information recording of the battlefield objective environment through various sensors, mainly including element knowledge and event knowledge. For example: the natural undulation state of the earth's surface terrain, the location, attributes, and spatial relationships of geographical information entities, the temperature, humidity, air pressure, rainfall, visibility, etc. of meteorology, the soil quality, soil type, rock formation attributes, etc. of geology, the temperature, salinity, density, ocean air pressure, underwater terrain undulation, etc. of the ocean, the magnetic field, signals, etc. of electromagnetism, and the human knowledge such as population, religion, culture, etc. Event knowledge includes geographical events, human events, and information events. For example, geographical events include events such as landslides, mudslides, floods, rainfall, soil erosion, and water and soil loss, human events include political events, economic events, military events, and cultural events, and information events include events such as signal interference, online public opinion, network reconnaissance, network attacks, and network defenses.

[0059] (2) Knowledge of environmental impact processes.

[0060] Knowledge of environmental impact processes mainly refers to the knowledge formed by experts in different environmental fields, such as cartographers, geologists, geodesists, oceanographers, etc., based on their professional knowledge, through the knowledge reprocessing of information on the basis of the basic information of environmental exploration. It includes explicit knowledge and implicit knowledge. Explicit knowledge is mainly the environmental impact index system summarized by domain experts, such as environmental effectiveness impact model indicators, geopolitical environment influence indicators, etc. Implicit knowledge is mainly the professional knowledge used by domain experts in the summarization process, as well as expert knowledge that is difficult to formalize, such as map symbol design knowledge, map cartography knowledge, etc.

[0061] (3) Knowledge of environmental service decision-making.

[0062] Knowledge of environmental service decision-making mainly refers to the knowledge that ultimately provides environmental guarantee services for combat operation decisions, mostly referring to various environmental comprehensive impact assessment models, combat operation environmental impact effectiveness assessment methods, etc., which are closely related to combat operations and weapons and equipment. For example, the impact assessment of adverse meteorological environments on combat operations, and environmental auxiliary decision-making knowledge such as land mobility, air transportation, and target damage.

[0063] Step 2: Construct a multi-level temporal knowledge hypergraph model for the battlefield environment. The construction method of this multi-level temporal knowledge hypergraph model for the battlefield environment is as follows:

[0064] First, an explanation is given for the design of the battlefield environment knowledge meta-model.

[0065] Knowledge representation is to study how to organize various required knowledge in the most appropriate form. Such as Figure 2As shown in the figure, according to the classification of battlefield environment knowledge, various types of knowledge are abstracted and modeled, and its logical structure can be represented as a knowledge set:

[0066] BEKM ∷={BE_Concepts, BE_Object, BE_Event, BE_Affect, BE_Decision, Relations}

[0067] Among them, BE_Concepts represents the set of battlefield environment concept classifications; BE_Object represents the set of battlefield environment entity knowledge, BE_Event represents the set of battlefield environment event knowledge, BE_Affect represents the set of battlefield environment impact knowledge, BE_Decision represents the set of battlefield environment decision - analysis knowledge, and Relations represents the set of relationships in the battlefield environment, mainly referring to the semantic relationships, spatial relationships, and interaction relationships among various types of knowledge.

[0068] (1) Representation of battlefield environment entity knowledge.

[0069] A battlefield environment entity refers to the object - oriented logical description of each element of the battlefield environment, either independently or jointly, within a certain battlefield area. It includes not only environmental objects existing in the real world (such as ground features, vegetation, clouds, rain, etc.), but also environmental objects that are difficult to visualize and exist virtually. Battlefield environment entities have spatial attributes, time attributes, regional attributes, process attributes, etc. The relationship types include spatio - temporal relationships such as azimuth relationships, distance relationships, and time relationships between entities, as well as semantic relationships such as subordination, instance, and inclusion. The logical structure of the representation of battlefield environment entities is a six - tuple:

[0070] BE_Object ∷=<TYPE, TIME, POS, RGN, STA, REL>

[0071] Among them, TYPE, TIME, POS, RGN, STA, and REL represent entity type, time, position, region, state, and relationship respectively.

[0072] (2) Representation of battlefield environment event knowledge.

[0073] A battlefield environment event is an event in which the behavior, spatial structure, attributes, or combination of the theme object (phenomenon) of the battlefield environment changes within a certain battlefield area, and this change has an important impact on the combat unit or the environment itself. For example, changes in battlefield environment entities within a certain spatio - temporal range, such as rainfall, debris flow, soil erosion, etc.

[0074] Battlefield environment events highlight the relationships between objects, including both the evolution relationships within objects and the association relationships between environmental objects. An event includes seven elements: the event subject (who), time (when), location (where), action (action), state (state), situation (what), and type (type). Formally, an event is represented as:

[0075] BE_Event ::= <TYPE, TIME, POS, OBJ, REL, ACT, STA, REL>

[0076] Among them, TYPE, TIME, POS, OBJ, REL, ACT, and STA respectively represent the event type, time element, location element, event subject, event relationship, action element (such as the action trigger word), state set, and situation (event description) of the event.

[0077] (3) Knowledge representation of the battlefield environment impact process.

[0078] Knowledge of the battlefield environment impact process is mainly used in fields such as battlefield environment impact assessment and combat simulation. It mainly relies on relevant materials and expert experience to form knowledge such as marine environment models, atmospheric environment models, terrain impact models, and comprehensive environment impact models. Environmental impact knowledge mainly exists in the form of impact factors and is represented by mathematical models in the process of simulation modeling. The logical structure of the battlefield environment impact process is represented as:

[0079] BE_Affect ::= <ENV, PER, WEP, OBJ, PRO, TIME, REL>

[0080] Among them, ENV, PER, WEP, OBJ, PRO, TIM, and R respectively represent environmental impact factors, personnel, weapons and equipment, objects of action, subject attributes, time, and relationships.

[0081] (4) Knowledge representation of battlefield environment service decision-making.

[0082] The battlefield environment service decision-making knowledge emphasizes considering the comprehensive impact results of comprehensive environmental factors such as geography, meteorology, electromagnetics, and networks on the comprehensive effectiveness of each stage of combat operations. It is necessary to consider the comprehensive modeling of military operation elements, weaponry, and the battlefield environment, which is a complex system knowledge modeling problem. The battlefield environment service decision-making knowledge comes from two aspects: on the one hand, it is the knowledge modeling of existing effectiveness evaluation methods. Representative evaluation methods include the analytic hierarchy process, ADC model method, combat simulation method, correlation analysis method, support vector machine evaluation method, etc. The representation of expert knowledge can be achieved through graph modeling of the evaluation methods; on the other hand, it is data- or knowledge-driven combat effectiveness evaluation. A qualitative and quantitative combined evaluation method is adopted. For example, the fuzzy influence diagram analysis method is used to model the knowledge hypergraph, and computational methods such as graph neural networks or large graph estimation can be used to achieve battlefield environment effectiveness evaluation. The logical structure is expressed as:

[0083] BE_Decision:: = <TYPE, TER, MET, SEA, OBJ, ACT, REG, REL>

[0084] Among them, TYPE, TER, MET, SEA, OBJ, ACT, REG, and REL respectively represent type, terrain, meteorology, ocean impact factor, impact entity, combat operation, region, and relationship.

[0085] Secondly, use the above-mentioned battlefield environment knowledge meta-model to construct the battlefield environment temporal knowledge hypergraph representation model of the present invention.

[0086] The following explains the battlefield environment knowledge hypergraph. Currently, most knowledge representations use triples. Although it can better represent the logical reasoning at the conceptual level, with the large-scale expansion of entity-level data, the temporal characteristics and scenario characteristics of battlefield environment knowledge become more and more obvious. The triple representation method makes the fragmentation of battlefield environment knowledge relatively serious. And it simplifies the complexity of the data stored in the knowledge graph. Especially for hyper-relation data connecting two or more entities, it cannot be represented by binary relations, and the loss of high-order structure information therein will lead to limitations in the representation and reasoning capabilities of the knowledge hypergraph.

[0087] The present invention adopts a knowledge hypergraph model to realize the modeling of complex and diverse relationships in the battlefield environment.

[0088] Construct a battlefield environment temporal hypergraph knowledge representation model, such as Figure 3As shown below. First, construct basic classes such as BE_Object, BE_Event, BE_Affect, BE_Decision, BE_Person, and BE_Weapon and establish relationships between the classes; then, extract the basic core elements TIME, POS, RGN, STA, and ACT and represent them using inheritance in the owl language; the instance layer mainly inherits from the concept layer, establishing relationships between instance layers and between instances and the concept layer from four levels: entity, event, influence process, and service decision. The entire hypergraph network is driven by time elements.

[0089] The knowledge hypergraph can be represented as HG = (V, E), where V = {v_1,..., v_n} is the set of entities (nodes), and E = {E_1,..., E_n} represents the set of non-empty ordered tuples of V, called the set of hyperedges. The hyperedge e ∈ E corresponds to a relation type mapping function , R is the set of relations, indicating that each hyperedge corresponds to a specific type of relation r ∈ R, and the arity |r| of the relation r is fixed, that is, the number of entities involved in the relation r is fixed.

[0090] In the knowledge hypergraph, a fact can be represented as a multi-tuple (r, v_1,..., v_n), where r ∈ R, v_i ∈ V, and (v_1,..., v_n) ∈ E. As Figure 4 shown: (1) The hyperedge "flood disaster" event connects multiple entities such as (time, location, type, disaster type), and can clearly represent the correlation between entities; (2) The hyperedge "city" can connect multiple entity attributes such as (time, location, area, population, type), and can comprehensively represent various attribute relationships of an entity; (3) The hyperedge "vehicle maneuver environment impact" connects multiple influencing factors such as (atmospheric pressure, altitude, fog, terrain slope, visibility), and can comprehensively describe the impact of the environment on the maneuver speed; (4) The hyperedge "anti-terrorism operation analysis and decision" connects multiple entity elements such as (snow cover, wind direction, wind speed, temperature, equipment), and can comprehensively analyze the comprehensive decision-making of the environment, weapon equipment, and combat operations. Secondly, introduce the multi-level temporal knowledge hypergraph model of the battlefield environment.

[0091] The knowledge of the battlefield environment exhibits obvious hierarchical, spatio-temporal, and scale characteristics. Referring to the hierarchical hypergraph model, as Figure 5 shown, the multi-level temporal knowledge hypergraph model of the battlefield environment can be formally represented as: G = {G BE_object , G BE_Event , G BE_Affect , G BE_Decision, R}, where R represents the set of relationships between layers. G BE_Object represents the element layer, which is composed of elements in various fields of the battlefield environment and the semantic hyper-edge relationships between them; G BE_Event represents the event layer, which is composed of various events in the battlefield environment and the causal, sequential, inheritance, and other event logical relationships between the events; G BE_Affect represents the environmental impact layer, which is composed of the impact factors of various elements in the battlefield environment and the impact weight relationships; G BE_Decision represents the environmental analysis and decision-making layer, which is composed of comprehensive environmental analysis models such as combat operations, combat training, and weapon strike effect evaluation, and the relationships in the combat operation process. The dotted line represents the cross-layer linkage of associated entities, time, and location at each level. G BE_object is an undirected attribute hypergraph network, representing the relationships of entity attributes; G BE_Event is a directed hypergraph network, representing the directed logical relationships of various events in the battlefield environment; G BE_Affect is a weighted directed hypergraph network, representing the relationships of various environmental element impacts, and the edges represent the weights of various impact factors; G BE_Decision is a fuzzy hypergraph network, which does not require discretization of continuous attributes. Let G BE_Decision = <d, e, λ>, where d = {d1, d2,..., d n} represents the vertex set, e = {e1, e2,..., e n} is the hyper-edge set, and λ is the optimal fuzzy similarity threshold of the fuzzy hypernetwork model. The conditional attribute set of the hyper-edge is c = {c1, c2,..., c n}, D is the decision attribute of the hyper-edge, and e1 is the hyper-edge in the hyper-edge set E that connects k vertices d i1 , d i2 ,..., d in . Among them, the vertex d i is a sample, and the samples in one hyper-edge have the same attribute set.

[0092] As Figure 4 shown, the dynamic mapping relationship of the time-series hypernetwork model, the change of the self-attributes of environmental entity elements or the mutual influence between entities, causes the occurrence of relevant geographical events, human events, or information events, and the single-element influence of weapon equipment and combat operations is caused by energy fusion or flow, and finally comprehensively affects combat command and strike effect.

[0093] The modeling of the battlefield environment knowledge hyper-edge is described below.

[0094] The battlefield environment knowledge hyper-edge mainly includes two categories: one is the knowledge hyper-edge of each subgraph, and the other is the cross-layer hyper-edge between subgraphs. The first type of hyper-edge is modeled according to the subgraph hyper-edge in 3.1, and the second type of cross-layer hyper-edge mainly includes G BE_object , G BE_Event , GBE_Affect , G BE_Decision The inter-layer mapping relationship between the four subgraphs.

[0095] The first type of hyperedges mainly include: (1) the hyperedges inside the element knowledge sub-network, mainly representing the attribute values of various entity elements; (2) the hyperedges inside the event sub-network, mainly representing the "5W1H" attribute values forming the events, as well as the inheritance, development, causal and other relationships between events; (3) the hyperedges inside the environmental impact process sub-network, mainly representing the comprehensive impacts of multiple environmental impact elements on weaponry or personnel; (4) the hyperedges inside the environmental service decision sub-network, mainly representing the decision-making impacts of the environment and equipment performance on combat operations.

[0096] The second type of hyperedges mainly include: (1) the mapping relationship between the element knowledge sub-network and the event sub-network (G BE_object →G BE_Event ), representing the element entities included in the battlefield environment events, such as the "occurrence location" relationship between the "flood disaster event" and the "city"; (2) the mapping relationship between the event sub-network and the environmental impact process sub-network (G BE_Event →G BE_Affect ), representing the impacts on weaponry or personnel caused by events occurring in the battlefield environment, such as the "impact factor" relationship between the "heavy rainfall event (event layer)" and the "visibility (environmental impact layer)"; (3) the mapping relationship between the environmental impact process sub-network and the environmental service decision sub-network (G BE_Affect →G BE_Decision ), representing the decision-making impacts of the impact parameters on weaponry or personnel on combat operations, such as the "fuzzy impact degree" relationship between the "impact of terrain slope on vehicles (environmental impact layer)" and the "coordinated pursuit (combat decision)".

[0097] Let o i ∈G BE_Object , e i ∈G BE_Event , a k ∈G BE_Affect , d l ∈G BE_Decision respectively represent any node of the four sub-networks of G BE_Object , G BE_Event , G BE_Affect , G BE_Decision , and the variable respectively represents whether there is a mapping relationship between different types. Thus, the battlefield environment knowledge hyperedges can be expressed as:

[0098] Among them represents whether there is a mapping relationship between o i and e j ; Indicates e j and a k whether there is a mapping relationship; Indicates a k and d l whether there is a mapping relationship.

[0099] The battlefield environment knowledge hyperedges can be divided into unordered hyperedges and ordered hyperedges. The element knowledge sub-networks are mostly unordered hyperedges, and the event knowledge sub-networks are mostly ordered hyperedges. The ordered hyperedges can be divided into chain hyperedges and cyclic hyperedges. The hyperedges need to be obtained by performing connectivity analysis on binary ordinary edges and constructing subgraphs, such as Figure 12-1 、 Figure 12-2 、 Figure 13-1 、 Figure 13-2 and Figure 13-3 shown.

[0100] Define the rules for constructing battlefield environment knowledge hyperedges:

[0101] Rule 1 constructs ordered hyperedges based on binary ordered ordinary edges:

[0102] (?r rdfs:subPropertyOf betho

[0103] :OrderedChainHE),(?x?r?y),(?y?r?i),makeChainHE(?x,?r,?z)->(?z rdf:type betho:OrderedHE),(?z rdf:first?x)

[0104] Described as: r is a type of chain-like ordered relationship in the environmental knowledge hypergraph. x, y, and i all represent nodes. (?x?r?y) means that nodes x and y are linked through relationship r. (?y?r?i) means that nodes y and i are linked through relationship r. (?x,?r,?z) means that a chain hyperedge z can be generated with node x as the starting node and satisfying relationship r, where z is an ordered hyperedge and x is the starting node.

[0105] Rule 2 constructs unordered hyperedges based on binary unordered ordinary edges:

[0106] (?r rdfs:subPropertyOf betho

[0107] :unorderedChainHE),(?x?r?y),(?y?r?i),makeChainHE(?x,?r,?z)->(?z rdf:type betho:UnorderedHE),(?z rdf:first?x),(?z rdf:member?y)

[0108] It is described as: r is a kind of chain-like disordered relationship of the environmental knowledge hypergraph. x, y, and i all represent nodes. (?x?r?y) means that node x and node y are linked through relationship r. (?y?r?i) means that node y and node i are linked through relationship r. (?x,?r,?z) means that a chain-like hyperedge z starting from node x and satisfying relationship r can be generated, where z is an unordered hyperedge, x is the starting node, and y is a member of z.

[0109] Rule 3: Constructing a circular ordered hyperedge from circular ordered ordinary edges:

[0110] (?r rdfs:subPropertyOf betho:orderedCircleHE), (?x?r?y), makeCircleHE(?x,?r,?z) -> (?z rdf:type betho:OrderedHE)

[0111] It is described as: r is a kind of circular ordered relationship of the environmental knowledge hypergraph. (?x,?r,?y) means checking the next node y on the circular hyperedge starting from node x and satisfying relationship r. If node y satisfies relationship r, then node y is connected to this circular hyperedge. (x, r, z) means that a circular hyperedge starting from node x and satisfying relationship r can be generated, then z is an ordered hyperedge.

[0112] Rule 4: Constructing a circular unordered hyperedge from circular unordered ordinary edges:

[0113] (?r rdfs:subPropertyOf betho:unorderedCircleHE), (?x?r?y), makeCircleHE(?x,?r,?z) -> (?z rdf:type betho:UnOrderedHE)

[0114] It is described as: r is a kind of circular unordered relationship of the environmental knowledge hypergraph. (?x,?r,?y) means checking the next node y on the circular hyperedge starting from node x and satisfying relationship r. If node y satisfies relationship r, then node y is connected to this circular hyperedge. (x, r, z) means that a circular hyperedge starting from node x and satisfying relationship r can be generated, then z is an unordered hyperedge.

[0115] Using the above rules to achieve knowledge hyperedge modeling. Thus, the modeling of edges, hyperedges, and nodes is completed, and a multi-level temporal knowledge hypergraph model of the battlefield environment is obtained.

[0116] Step 3: Use the multi-level time-series knowledge hypergraph model of the battlefield environment. Based on the knowledge hypergraph, represent the battlefield environment entity knowledge, battlefield environment event knowledge, battlefield environment impact process knowledge, and battlefield environment service decision-making knowledge respectively to obtain the knowledge hypergraphs of the corresponding battlefield environment entity layer, battlefield environment event layer, battlefield environment impact process layer, and battlefield environment service decision-making layer.

[0117] Step 4: In the battlefield environment entity layer, battlefield environment event layer, battlefield environment impact process layer, and battlefield environment service decision-making layer, represent the relationship between the nodes in the knowledge hypergraph of a certain layer and the nodes in the knowledge hypergraphs of other layers using hyperedges or edges to obtain a cross-layer knowledge hypergraph.

[0118] Step 5: Visualize the obtained knowledge hypergraphs of the battlefield environment entity layer, battlefield environment event layer, battlefield environment impact process layer, and battlefield environment service decision-making layer, as well as the cross-layer knowledge hypergraph, as Figure 14 shown.

[0119] The effects of the present invention will be described below in combination with experiments and analysis.

[0120] Taking the analysis of the impact of counter-terrorism operation environment as an example, model the battlefield environment knowledge from four levels: entity, event, impact process, and analysis decision-making. The data and models are shown in Table 1. The experimental hardware processor is Intel Core i7-8750H 2.20GHz, the memory is 32G, and the graphics card is GeForce GTX 1050Ti. The software development IDE uses Pycharm, the front-end JavaScript is used for graph visualization, and the knowledge graph is stored in the Neo4j graph database.

[0121] Table 1 Experimental data

[0122]

[0123] The following is a comparative description of the element knowledge representation.

[0124] (1) The map element knowledge adopts structured shp data, as Figure 6 shown. The names, types, locations, regions, times, and states of each map element are extracted, and hyperedges are used for storage in neo4j.

[0125] (2) Event knowledge representation.

[0126] Taking geopolitical events as an example for event knowledge extraction, relevant news events are crawled from news websites, and the events are sorted according to the event trigger words, obtaining 34,729 nodes and 69,457 edges. Example representations of hyperedges are shown in Table 2, and example visualization results are as Figure 7 shown.

[0127] Example of the hypergraph representation of the events in Table 2

[0128]

[0129]

[0130] The knowledge representation of the battlefield environment impact process will be described below.

[0131] Taking the analysis of the terrain passing performance of tracked vehicles as an example, its impact analysis model is expressed as follows:

[0132] In the formula, P is the power of the vehicle engine, f1 is the influence coefficient of atmospheric pressure on the power output of the vehicle engine, η is the mechanical efficiency of the vehicle, V is the moving speed of the vehicle, f is the friction coefficient of the vehicle during driving, G is the total weight of the vehicle, α is the slope of the ground, C is the air resistance coefficient, A is the frontal projected area of the vehicle, and V f is the wind speed component in the opposite direction of the vehicle movement. The result of its impact hypergraph representation is shown in Table 3.

[0133] Table 3 Example of the knowledge graph representation of the battlefield environment impact process

[0134]

[0135] The knowledge representation of the battlefield environment service decision will be described below.

[0136] Taking the environmental impact decision knowledge of anti-terrorism operations in the plateau area as an example, the relationship layer, numerical layer and function layer of the environmental comprehensive fuzzy impact diagram and the fuzzy relationship between nodes are converted into a hypergraph model. The state and frequency of independent nodes are converted into hyperedge relationships, as shown in Table 4. The frequency fuzzy set, the state fuzzy sets of independent nodes and non-independent nodes are respectively converted into frequency and state nodes. The fuzzy relationship between nodes is converted into a hyperedge relationship. The experimental results are as Figure 8 shown, where VH, H, M, L, VL represent frequency fuzzy nodes, and the membership degree is used as the node attribute value; G, S, B are state fuzzy nodes of independent nodes, and the membership degree is used as the node attribute value; HD, MD, LD, NO are state fuzzy nodes of non-independent nodes, and the membership degree is used as the node attribute value. Using the state and frequency of independent nodes to construct hyperedges, and the node fuzzy relationship to construct the relationship between various nodes, the results are as Figure 8 shown.

[0137] Table 4 Hypergraph representation of nodes - non-independent nodes

[0138] Node Frequency Status Non-independent node Non-independent node status Hypergraph representation 1 cloud type M G 11 helicopters LD As Figure 15-1 shown 1 cloud type H S 10 satellites NO As Figure 15-2 shown 3 wind speeds VL G 11 helicopters MD As Figure 15-3 shown 9 air pressures H B 11 helicopters HD As Figure 15-4 shown

[0139] The association of various types of knowledge in the battlefield environment will be described below.

[0140] Based on battlefield environment element entities, cross-domain multi-source data such as entity attributes, images / street views, events, military geography documents, maps, etc. are associated. Its advantage is to provide a multi-source environment data retrieval based on entity semantics, such as Figure 9 shown. It realizes the retrieval of multi-source heterogeneous battlefield environment data semantically and can be associated and mapped with the three-dimensional virtual battlefield environment, and realizes the interaction with the three-dimensional scene based on the semantic nodes of the atlas.

[0141] The modeling of battlefield environment knowledge is the core basic work of intelligent support for the battlefield environment. Modeling and analyzing various types of knowledge such as elements, events, influence processes, decision-making services, etc. is a huge systematic project. The present invention mainly analyzes and models battlefield environment knowledge from the perspectives of knowledge system classification, knowledge hypergraph model construction, knowledge hypergraph construction and association visualization, etc. It mainly realizes the representation of various types of knowledge from data, models to knowledge hypergraphs, and based on multi-source battlefield environment data, an experiment on knowledge hypergraph modeling is carried out. The present invention supports the associated search of hypergraph networks of various types of battlefield element types, and then associates the knowledge of battlefield environment influence processes and decision analysis knowledge to comprehensively analyze the battlefield environment, laying a foundation for providing intelligent services for subsequent situation prediction, command and control, etc. For example, the knowledge hypergraph of the present invention can provide services for the following application scenarios: the integration of battlefield environment knowledge and knowledge networks such as battlefield situation, command and control, etc., knowledge completion based on multi-modal battlefield environment knowledge, knowledge reasoning based on graph neural networks, intelligent analysis and prediction of battlefield scenarios based on knowledge graphs, etc.

Claims

1. A method for representing battlefield environment knowledge, characterized in that: Including the following steps: 1) Based on battlefield environment data, extract battlefield environment knowledge therefrom, where the battlefield environment knowledge includes battlefield environment entity knowledge, battlefield environment event knowledge, battlefield environment impact process knowledge, and battlefield environment service decision-making knowledge; The battlefield environment entity knowledge includes an objectified logical description of each element of the battlefield environment, either independently or jointly, within a certain battlefield area; the battlefield environment event knowledge includes an event in which the behavior, spatial structure, attributes, or combination of elements of the battlefield environment has changed within a certain battlefield area, and the change has had an important impact on the combat unit or the environment itself; the battlefield environment impact process knowledge includes knowledge used in the fields of battlefield environment impact assessment and combat simulation and relies on relevant materials and expert experience to form knowledge including marine environment models, atmospheric environment models, terrain impact models, and comprehensive environment impact models; the battlefield environment service decision-making knowledge includes knowledge of the comprehensive effectiveness impact of comprehensive environmental elements including geography, meteorology, electromagnetism, and network on each stage of combat operations; 2) Based on the knowledge hypergraph, represent the battlefield environment entity knowledge, battlefield environment event knowledge, battlefield environment impact process knowledge, and battlefield environment service decision-making knowledge respectively to obtain the knowledge hypergraphs of the corresponding battlefield environment entity layer, battlefield environment event layer, battlefield environment impact process layer, and battlefield environment service decision-making layer; 3) In the battlefield environment entity layer, battlefield environment event layer, battlefield environment impact process layer, and battlefield environment service decision-making layer, represent the relationship between the nodes in the knowledge hypergraph of a certain layer and the nodes in the knowledge hypergraphs of other layers using hyperedges / edges to obtain a cross-layer knowledge hypergraph; 4) Associate and visually display the obtained knowledge hypergraphs of the battlefield environment entity layer, battlefield environment event layer, battlefield environment impact process layer, and battlefield environment service decision-making layer, as well as the cross-layer knowledge hypergraph.

2. The method for representing battlefield environment knowledge according to claim 1, wherein: The relationships described in step 3) include the mapping relationship between the battlefield environment entity layer and the battlefield environment event layer, the mapping relationship between the battlefield environment event layer and the battlefield environment impact process layer, and the mapping relationship between the battlefield environment impact process layer and the battlefield environment service decision-making layer; the mapping relationship between the battlefield environment entity layer and the battlefield environment event layer is used to represent the relationship between the entities included in the battlefield environment event knowledge and the battlefield environment knowledge; the mapping relationship between the battlefield environment event layer and the battlefield environment impact process layer is used to represent the impact on weapons and equipment or personnel caused by the events occurring in the battlefield environment; the mapping relationship between the battlefield environment impact process layer and the battlefield environment service decision-making layer is used to represent the decision-making impact of the impact parameters on weapons and equipment or personnel on combat operations.

3. The method for representing battlefield environment knowledge according to claim 1 or 2, characterized in that: The battlefield environment entity layer includes several battlefield environment entities, and the battlefield environment entities are represented using entity representation attributes; the entity representation attributes include entity type, location, area, time, status, and semantic relationship; the semantic relationship includes the relationship of all entity representation attributes of an entity; the semantic relationship is represented using a hyperedge / edge; each entity representation attribute other than the semantic relationship is represented using a node; after representing the battlefield environment entity layer based on the knowledge hypergraph, an undirected attribute hypergraph network is obtained.

4. The method for representing battlefield environment knowledge according to claim 1 or 2, characterized in that: The battlefield environment event layer includes several battlefield environment events, and the battlefield environment events are represented using event representation attributes. The event representation attributes include event type, time element, location element, event subject, event logical relationship, action element, state set, and event description. The event logical relationship includes the event representation attribute values that form an event and the inheritance, development, and causal relationships between events, and the event logical relationship is represented using hyperedges / edges. Each event representation attribute other than the event logical relationship is represented using a node. After representing the battlefield environment event layer based on a knowledge hypergraph, a directed hypergraph network with a directed logical relationship is obtained.

5. The method for representing battlefield environment knowledge according to claim 1 or 2, characterized in that: The battlefield environment impact process layer includes several battlefield environment impact processes, and the battlefield environment impact processes are represented using process representation attributes. The process representation attributes include environmental impact factors, personnel, weapons and equipment, objects of action, subject attributes, time, and impact weight relationship. The impact weight relationship includes the comprehensive impact of multiple environmental impact factors on weapons and equipment or personnel, and the impact weight relationship is represented using hyperedges / edges. Each process representation attribute other than the impact weight relationship is represented using a node. After representing the battlefield environment impact process layer based on a knowledge hypergraph, a weighted directed hypergraph network is obtained.

6. The method for representing battlefield environment knowledge according to claim 1 or 2, characterized in that: The battlefield environment service decision layer includes several battlefield environment service decisions, and the battlefield environment service decisions are represented using decision representation attributes. The decision representation attributes include type, terrain, meteorology, ocean impact factors, impact entities, combat operations, regions, and process relationships. The process relationship includes the decision-making impact of the environment and equipment performance on combat operations. The process relationship is represented using hyperedges. Each decision representation attribute other than the process relationship is represented using a node. After representing the battlefield environment service decision layer based on a knowledge hypergraph, a fuzzy hypergraph network is obtained.

7. The method for representing battlefield environment knowledge according to claim 1, characterized in that: The hyperedge logic used in step 3) is represented as: where o i ∈ G BE_Object , e j ∈ G BE_Event , a k ∈ G BE_Affect , d l ∈ G BE_Decision respectively represent any node in the four layers of G BE_Object , G BE_Event , G BE_Affect , G BE_Decision , represents whether there is a mapping relationship between o i and e j , represents whether there is a mapping relationship between e j and a k , represents whether there is a mapping relationship between a k and d l 。 8. The method for representing battlefield environment knowledge according to claim 1, wherein: The hyperedges include unordered hyperedges and ordered hyperedges, and the corresponding unordered hyperedges and ordered hyperedges are constructed according to preset rules.

9. The method for representing battlefield environment knowledge according to claim 1, characterized in that: The formal representation of the overall representation model used when representing the knowledge hypergraphs of the battlefield environment entity layer, battlefield environment event layer, battlefield environment impact process layer, and battlefield environment service decision layer and the cross-layer knowledge hypergraphs in step 4) is: G = {G BE_Object , G BE_Event , G BE_Affect , G BE_Decision , R} Among them, R represents the set of mapping relationships between layers; G BE_Object represents the battlefield environment entity layer, including the entities in each field of the battlefield environment and the semantic hyper-edge relationships between them; G BE_Event represents the battlefield environment event layer, including various events in the battlefield environment and the causal, sequential, and inheritance event logical relationships between the events; G BE_Affect represents the battlefield environment impact process layer, including the impact factors of each entity in the battlefield environment and the impact weight relationship; G BE_Decision represents the battlefield environment service decision-making layer, including the comprehensive environmental analysis models of combat operations, combat training, and weapon strike effect assessment, as well as the process relationships of combat operations.

Images