Multi-domain battlefield task decision-making method and system based on knowledge graph

Abstract

The invention discloses a multi-domain battlefield task decision-making method and system based on a knowledge graph. The multi-domain battlefield task decision-making method based on the knowledge graph comprises the steps that the relation between battlefield elements and tasks in a database is analyzed, and situation entity entries of the battlefield elements are extracted; utilizing a predefined relation template to construct a relation between the situation entity entries; constructing a decision knowledge base based on fuzzy reasoning; based on the decision knowledge base, constructing a knowledge reasoning model by using a dynamic Bayesian network technology; and receiving a combat instruction, generating a multi-dimensional decision suggestion by using the knowledge reasoning model based on a combat rule matched with the knowledge graph in the decision knowledge base, and displaying a decision result through the user interaction module. According to the method, multi-source information can be integrated in a multi-domain combat environment, cross-domain collaborative decision is realized, and the adaptability of a complex battlefield is improved; the knowledge graph and the fuzzy reasoning technology are combined, so that the uncertainty and fuzziness of the battlefield situation are effectively processed, and the decision-making ability is remarkably enhanced.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of artificial intelligence and military decision support, and in particular to a multi-domain battlefield task decision method and system based on a knowledge graph. BACKGROUND

[0002] At present, the modern battlefield environment is complex and changeable, and the commander needs to process a large amount of intelligence, analyze the environment situation and make decisions in a short time. However, the existing battlefield decision support system has many shortcomings:

[0003] First, the decision-making in complex battlefield environment faces challenges: the existing battlefield decision support system has problems such as data island, ineffective integration of information and slow decision response when dealing with complex battlefield environment; the existing battlefield decision support system cannot effectively integrate multi-source battlefield situation, and it is difficult to quickly generate efficient decision path. Second, the uncertainty and fuzziness in the decision-making process: the uncertainty and dynamic change in the battlefield environment make the decision-making process challenging. The existing decision support system often has difficulty in dealing with the uncertainty and fuzziness in the battlefield situation. Third, the demand for multi-domain operation has not been met: the existing battlefield decision support system is mostly for single combat domain and cannot effectively deal with the demand for multi-domain complex combat tasks. SUMMARY

[0004] The embodiment of the present application provides a multi-domain battlefield task decision method and system based on a knowledge graph to solve the technical problems of how to effectively integrate multi-source battlefield situation, quickly integrate heterogeneous information and improve the decision-making efficiency and accuracy in the multi-domain complex battlefield environment.

[0005] A multi-domain battlefield task decision method based on a knowledge graph, comprising:

[0006] S10, analyzing the relationship between battlefield elements and tasks in the database to extract situation entity items of combat elements;

[0007] S20, constructing the relationship between the situation entity items using a predefined relationship template;

[0008] S30, constructing a decision knowledge base based on fuzzy reasoning;

[0009] S40, constructing a knowledge reasoning model based on the decision knowledge base using dynamic Bayesian network technology;

[0010] S50, receiving combat instructions, and based on the combat rules matched with the knowledge graph in the decision knowledge base, generating multi-dimensional decision suggestions using the knowledge reasoning model, evaluating and screening the decision suggestions through decision analysis, and displaying the decision results through a user interaction module.

[0011] A multi-domain battlefield mission decision-making system based on knowledge graph, including:

[0012] An analysis and extraction module, used to analyze the relationship between battlefield elements and tasks in the database to extract situation entity entries of combat elements;

[0013] A relationship building module, used to build relationships between situation entity items using predefined relationship templates;

[0014] Decision knowledge base construction module, used to construct decision knowledge base based on fuzzy reasoning;

[0015] The knowledge reasoning model construction module is used to construct a knowledge reasoning model based on the decision knowledge base using dynamic Bayesian network technology;

[0016] The decision generation and display module is used to receive combat instructions and generate multi-dimensional decision suggestions based on combat rules that match the knowledge graph in the decision knowledge base using the knowledge reasoning model. The decision suggestions are evaluated and screened through decision analysis, and the decision results are displayed through the user interaction module.

[0017] The present invention has the following technical effects:

[0018] First, in response to the shortcomings of existing technologies in adaptability to combat environments, the present invention proposes a system capable of making multi-task decisions in a multi-domain combat environment. Existing technologies mainly focus on a single combat domain (such as airspace), which makes it difficult to meet the needs of complex multi-domain combat tasks. The knowledge graph-based multi-task decision-making method and system of the present invention can integrate and process multi-source information from different combat domains, realizing cross-domain information fusion and collaborative decision-making. This capability significantly improves the system's adaptability and decision-making efficiency in complex battlefield environments, laying the foundation for subsequent technical improvements.

[0019] Secondly, to address the limitations of existing technologies in knowledge graph applications, this invention deeply integrates knowledge graphs with fuzzy reasoning technology. After constructing a battlefield situation knowledge graph, existing technologies lack effective multi-task decision-making mechanisms and are unable to fully utilize the complex relationships within the knowledge graph for efficient reasoning. By constructing a decision-making knowledge base based on fuzzy reasoning, this invention can effectively handle the uncertainty and ambiguity in battlefield situations, thus addressing the shortcomings of existing technologies and further enhancing the system's decision-making capabilities.

[0020] Thirdly, to solve the static problem of the prior art in battlefield situation awareness and prediction, the application introduces dynamic Bayesian network technology. The prior art relies on static models and is difficult to reflect the dynamic changes of the battlefield situation in real time. However, the application can perceive, reason and predict the changes of the battlefield situation by constructing a dynamic knowledge base model, and update the probability relationship of the battlefield events in real time, dynamically adjust the decision path. This improvement ensures that the decision result is always based on the latest battlefield information, significantly improving the timeliness and dynamic adaptability of the decision.

[0021] Finally, in order to improve the actual application effect of the system, the application adopts modular design and optimizes the user interaction module. The modular design of the system includes knowledge reasoning module, decision analysis module, user interaction module, etc. Each functional module works independently and cooperatively, has high scalability and flexibility, and can be customized and optimized according to different combat requirements. In the design of the user interaction module, the system provides a friendly graphical interface and multiple interaction modes, ensuring the use experience and operation efficiency of the commander, and ensuring the efficiency and ease of use of the system in practical application.

[0022] In summary, the application solves the deficiencies of the prior art in multi-domain combat environment adaptability, knowledge graph application, battlefield situation awareness and prediction, and user interaction, etc., and realizes the efficient, dynamic and easy-to-use decision-making ability of the system in complex battlefield environment. These advantages complement each other and improve the overall performance and practicality of the system. BRIEF DESCRIPTION OF DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the application, the following will briefly introduce the drawings needed to be used in the description of the embodiments of the application. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor.

[0024] Figure 1 is a flowchart of the multi-domain battlefield task decision method based on knowledge graph in an embodiment of the application;

[0025] Figure 2 is a flowchart of constructing the relationship between situation entity items in an embodiment of the application;

[0026] Figure 3 is a flowchart of constructing the decision knowledge base in an embodiment of the application;

[0027] Figure 4 is a structural diagram of the multi-domain battlefield task decision system based on knowledge graph in an embodiment of the application. DETAILED DESCRIPTION

[0028] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0029] The multi-domain battlefield task decision-making method based on knowledge graph provided by the embodiment of the present invention can be applied as follows: Figure 1 Specifically, the multi-domain battlefield task decision method based on knowledge graph is applied in a multi-domain battlefield task decision system based on knowledge graph, and the multi-domain battlefield task decision system based on knowledge graph includes the following: Figure 1 The client and server shown communicate over a network. The client, also known as the user end, is the program that corresponds to the server and provides local services to clients. The client can be installed on, but is not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices. The server can be implemented as a standalone server or a server cluster consisting of multiple servers.

[0030] In one embodiment, if Figure 1 As shown in the figure, a multi-domain battlefield task decision method based on knowledge graph is provided. Figure 1 The server in the example is used as an example, and the steps are as follows:

[0031] S10, analyzing the relationship between battlefield elements and tasks in the database to extract situation entity entries of combat elements;

[0032] S20, constructing relationships between situation entity entries using a predefined relationship template;

[0033] S30. Constructing a decision knowledge base based on fuzzy reasoning;

[0034] S40. Based on the decision knowledge base, a knowledge reasoning model is constructed using dynamic Bayesian network technology;

[0035] S50, receiving combat instructions, and based on combat rules that match the knowledge graph in the decision knowledge base, using the knowledge reasoning model to generate multi-dimensional decision suggestions, evaluating and screening the decision suggestions through decision analysis, and displaying the decision results through the user interaction module.

[0036] In the above-mentioned embodiments of the present invention, the following core improvements are achieved to address the shortcomings of existing technologies in adaptability to multi-domain combat environments, knowledge graph application, battlefield situation perception and prediction, and user interaction: First, it is possible to integrate multi-source information in a multi-domain combat environment, realize cross-domain collaborative decision-making, and improve adaptability to complex battlefields; second, by combining knowledge graphs with fuzzy reasoning technology, it is possible to effectively handle the uncertainty and ambiguity of the battlefield situation and enhance decision-making capabilities; third, dynamic Bayesian network technology is introduced to construct a dynamic knowledge base model to perceive, reason and predict battlefield situation changes in real time, ensuring the timeliness and dynamic adaptability of decisions; finally, a modular design is adopted and the user interaction module is optimized to provide a friendly graphical interface and multiple interaction methods, thereby improving scalability, flexibility and ease of use, and being more practical and efficient in complex and changing battlefield environments.

[0037] In one embodiment, if Figure 1 As shown, the S10 analyzes the relationship between battlefield elements and tasks in the database to extract situation entity entries of combat elements; it includes the following sub-steps:

[0038] S101. Analyze battlefield elements and tasks in the database, and then sort out the inherent connections between battlefield elements and tasks;

[0039] S102. Combine combat instructions and real-time perception information to obtain specific data from different information sources, including extracting combat node data from battlefield resource information, extracting lighting conditions from environmental information, and extracting tactical objectives from combat instructions;

[0040] S103. Utilize natural language processing (NLP) technology to automatically extract and summarize specific situational knowledge units from specific data as situational entity entries for combat elements.

[0041] Understandably, in the process of extracting combat element situation entity entries, the system follows a series of clear and logically rigorous steps to automatically extract and summarize key information in the battlefield situation.

[0042] First, the system analyzes the relationship between battlefield elements and tasks in the database. This fundamental analysis is a prerequisite for subsequent work. By sorting out the inherent connections between battlefield elements and tasks, it provides a basis for the subsequent extraction of specific situation entity entries. Based on this foundation, it further extracts and summarizes specific situation entity entries by combining combat orders and real-time perception information. Combining combat orders with real-time perception of the battlefield situation makes the extracted situation entity entries more targeted and real-time, accurately reflecting the actual situation on the battlefield.

[0043] In the process of extracting situation entity entries, specific data is obtained from different information sources, and combat node data is extracted from battlefield resource information, such as "<f><Available> indicates that the unmanned aerial vehicle node of model IF_1 can be used, which provides data support for understanding the combat forces on the battlefield and their available states; the light condition is extracted from the environmental information, and the tactical purpose is extracted from the combat order, such as "<light condition> <includes> <sufficient light> and <light condition> <includes> <insufficient light> ", and further corresponding tactical decisions are made according to the light condition, such as "<sufficient light> <take> <high-speed driving> and <insufficient light> <take> <low-speed driving> ", so that the ability to flexibly adjust tactics according to different environmental factors can be achieved.

[0044] In order to realize the extraction of the above-mentioned situation entity items, it is necessary to analyze structured and unstructured data sources. These data sources include battlefield sensor information, text information, etc., covering various types of data forms. By using natural language processing (NLP) technology and other methods, relevant situation knowledge units are automatically extracted from these complex data sources. Based on the analysis of battlefield elements and tasks, the analysis representation method of combat orders, and the extraction of battlefield situation knowledge units from historical and real-time perception information of battlefield engagement states, specific situation entity items are induced, such as the number of combat elements, terrain, strategic and tactical purposes. In specific implementation, entity items such as "good light", "stationary small obstacle", "XX station reconnaissance task" are identified and extracted, and stored in an entity set. These extracted entity items are the basis for subsequent relationship construction and reasoning, and provide important data support for subsequent battlefield situation analysis and decision-making. In summary, the entire process is an orderly process from basic analysis to data extraction, to knowledge unit identification and storage. Through the comprehensive processing of multi-source data and the application of natural language processing technology, the automatic analysis and knowledge extraction of the battlefield situation are realized, providing strong support for subsequent combat decision-making.

[0045] In an embodiment, the step S20 constructs the relationship between the situation entity items using a predefined relationship template; further comprising the following sub-steps:

[0046] S201, using a predefined relationship template to construct the relationship between the task and the target, the relationship between the task and the allocation requirement, and the relationship between the task and the rule;

[0047] S202, further refining the relationship and conditions between the task and the allocation requirement during the execution of the task, and considering the requirements for the capabilities of the unmanned node during the execution of the task, to ensure that the node has specific capabilities;

[0048] S204, the relationship between the constructed entity items is displayed in a visual manner, each entity item is represented in the form of a node, and the relationship between the nodes is connected by directed edges, and the relationship type is marked on the edge.

[0049] Understandably, when constructing the relationship between the situation entity entries, the following logic and steps are followed:

[0050] First, based on the extracted situation entity entries, various relationships between the entity entries can be constructed using predefined relationship templates, which are key links for situation analysis and decision support, including the relationships between tasks and targets, tasks and allocation requirements, and tasks and rules, etc. Among them, the relationship between tasks and targets is the most basic, and different tasks determine different targets. For example, when it is detected that there is a relationship between tasks and targets, such as <采取> ", and store this correspondence in the form of a dictionary so that the decision-making system can quickly retrieve and use it.

[0051] Different tasks will have different node requirements, for example <分配> <fmyn>" indicates that m UAV nodes and n UCAV nodes are required to execute Task B. This clarifies the number of resources required for task execution and provides a specific basis for resource allocation.

[0052] Furthermore, we must also consider the requirements for the capabilities of unmanned nodes during task execution. In some cases, it is necessary not only to meet the number requirements of unmanned nodes, but also to ensure that these nodes have specific capabilities. In this case, the system will formulate <allocations> <fmyn> ， <需要能力> <c>The rule of " indicates that when executing task B, in addition to m drone nodes and n unmanned vehicle nodes, these nodes must also have capability C. This further refines the conditions for task allocation and ensures that the allocated resources can meet the actual needs of the task.

[0053] Finally, these constructed relationships are displayed visually, such as< / c> < / fmyn> < / fmyn> Figure 2 As shown in the figure, each entity entry is represented as a node, and the relationships between nodes are connected by directed edges, with specific relationship types labeled on the edges, such as "include," "take," "allocate," "need capability," and "possess capability." This visual representation clearly illustrates the complex relationships between entity entries, facilitating understanding and analysis and providing an intuitive reference for subsequent decision-making. Through this series of logically clear steps, the system effectively constructs relationships between situational entity entries, providing strong support for subsequent situation analysis and task decision-making.

[0054] Understandably, when constructing the relationships between situational entity entries, by clarifying the logical relationships between tasks, objectives, organizations, and rules, we ensure that the interactions between these elements are clearly reflected in the situational analysis and decision-making process: First, the relationship between tasks and objectives is fundamental. Objectives are the fundamental reason for the existence of tasks and determine their content and direction. In other words, tasks are designed to achieve specific objectives, providing a clear context, operational environment, and requirements for the task. A single objective may require the coordinated completion of multiple tasks, and the execution of tasks is intended to gradually achieve the objective. This relationship reflects the decisive and guiding role of objectives in tasks.

[0055] Secondly, the relationship between tasks and organizations reflects the execution aspect of tasks. Organizations are the principal actors in task execution; each task requires at least one organization. An organization's capabilities determine the types of tasks it can undertake, so the organization's actual capabilities must be considered when assigning tasks. Furthermore, the match between tasks and organizations is not completely fixed; the task determination process requires constant comparison and adjustment with organizational capabilities to ensure feasibility and effectiveness. This relationship emphasizes the central role of organizations in task execution and the mutual adaptability between tasks and organizational capabilities.

[0056] Furthermore, the relationship between tasks and rules involves the specific strategies and methods for executing tasks. Essentially, rules describe the methods and strategies for using an organization to achieve its goals. Rules determine how tasks are matched to organizations and also dictate when, where, and how the organization executes tasks. Therefore, rules not only influence the allocation of tasks but also directly impact the execution of tasks. This relationship highlights the guiding and constraining role of rules in task execution.

[0057] Finally, the relationship between goals, organizations, and rules is interactive. Goals guide the direction of an organization, which achieves them through the execution of tasks. Rules, on the other hand, provide the specific strategies and methods used to achieve them. Goals determine the formulation of rules, as rules must be aligned with them. At the same time, rules also react to goals, and different rules can lead to varying degrees of goal achievement. Rules have enforceable power, constraining organizational behavior and ensuring that they adhere to established strategies during operations. Therefore, organizations must fully understand and apply rules to achieve their goals.

[0058] In summary, the relationship between tasks, objectives, organizations, and rules is interdependent and interactive. Objectives are the starting point of tasks, tasks are the basis for organizational action, and rules provide guidance and constraints for task execution. This clear logical relationship ensures that the system accurately reflects the complexity and dynamic nature of battlefield situations when constructing situation entity entries, providing a solid foundation for subsequent situation analysis and decision-making.

[0059] In one embodiment, if Figure 3 As shown, step S30 constructs a decision knowledge base based on fuzzy reasoning; it includes the following sub-steps:

[0060] S301. Based on the fuzzy reasoning and rough set theory and methods, valuable knowledge is extracted from the original observation data to build a fuzzy decision model.

[0061] S3011. Based on the extracted valuable knowledge, construct a continuous value decision table for combat command mode;

[0062] S3012. Use fuzzy processing to convert the continuous value decision table into a fuzzy decision table;

[0063] S3013. Calculate the inclusion degree of the fuzzy decision table and select decision rules with high inclusion degree;

[0064] S3014. Extract fuzzy inference rules based on high-inclusion decision rules, build a decision knowledge base and store it.

[0065] As can be understood, in the process of constructing a decision knowledge base based on fuzzy reasoning, fuzzy rough set theory and methods are utilized to extract valuable knowledge from raw observation data, which forms the foundation for building a fuzzy decision model. Specifically, the first step is to construct a continuous-valued decision table for the combat command method. This stage involves quantizing the relevant parameters of the combat command method to form a decision table with specific numerical values, providing a clear data foundation for subsequent processing. Next, the system converts the continuous-valued decision table into a fuzzy one. This step is achieved through fuzzification, which converts quantized continuous values ​​into fuzzy values ​​to accommodate the uncertainty and ambiguity of the battlefield situation. Fuzzification can better reflect the complexity and diversity of the battlefield environment, providing data forms that are more realistic for subsequent reasoning. Subsequently, the inclusion degree of the fuzzy decision table is calculated. The inclusion degree calculation is used to select decision rules with high inclusion degrees, thereby ensuring that the extracted rules are highly reliable and applicable. By calculating the inclusion degree, the system can identify which rules are more valuable in fuzzy environments, thereby providing stronger support for decision making. Finally, the fuzzy reasoning rules are extracted and stored in the decision knowledge base. These rules are highly inclusive rules selected from the fuzzy decision table, which can provide a basis for reasonable decision-making in complex battlefield situations. The construction process of the decision knowledge base ensures the effective processing of complex battlefield situation information and scientific decision-making based on fuzzy reasoning rules. The whole process is as follows Figure 3 As shown in the figure, it includes key steps such as data input, fuzzy processing, and rule extraction. These steps are closely linked and gradually advanced, ultimately building a decision-making knowledge base that can adapt to battlefield ambiguity and uncertainty, providing strong support for combat command.

[0066] In one embodiment, if Figure 1 As shown, step S40 constructs a knowledge reasoning model based on the decision knowledge base using dynamic Bayesian network technology; it includes the following sub-steps:

[0067] S401. Using a Bayesian network to establish a probability model for battlefield events, a static model is constructed by defining conditional independence and probability relationships between variables; the expression of the static model is:

[0068] P(x1,x2,…,x n )=∏ i P(x i |pa i ) (1)

[0069] Among them, (x1,x2,…,x n ) represents the implicit state variable, pa i Represents the variable x i The parent node set of x in the Bayesian network structure i The set of upstream variables that are directly related and have an impact on it, P(x i |pa i ) indicates that in the parent node pa i Under the condition of the value of i Take the probability distribution of a specific state;

[0070] S402: Expand the static model on the time axis through a dynamic Bayesian network to adapt to time changes. The expression of the dynamic Bayesian network is:

[0071]

[0072] Among them, P(x i |x i-1 ) represents the probability function between states at different time slices; P(y i |x i ) represents the probability function of each node on the same time slice; P(x0) represents the initial state at the beginning of the process;

[0073] S403. Extract the characteristics of battlefield situation elements and acquire domain knowledge, adjust and optimize the dynamic Bayesian network structure to construct a knowledge reasoning model of the Bayesian network to ensure accurate reflection of the causal and dependency relationships of the battlefield situation.

[0074] As can be understood, the above implementation utilizes dynamic Bayesian network technology to construct a Bayesian network knowledge base model capable of perceiving, reasoning, and predicting battlefield situations. This serves as the foundation for the entire knowledge reasoning model. The choice of dynamic Bayesian network technology enables the model to adapt to dynamic changes in the battlefield situation and provide real-time analysis and prediction of battlefield events. By modeling the probabilistic relationships between battlefield events, the Bayesian network can dynamically update its estimate of the battlefield situation. This is based on the core function of the Bayesian network, which utilizes probabilistic reasoning to handle uncertainty and dynamics. By establishing probabilistic relationships between events, the estimate of the battlefield situation can be adjusted in real time based on new observations, thereby providing more accurate information for decision-making. Furthermore, to ensure that the Bayesian network model accurately reflects the causal and dependency relationships within the battlefield situation, the Bayesian network undergoes necessary structural adjustments. This adjustment process involves adding and removing nodes and rewiring edges. Node addition and removal introduce or remove new or no longer relevant information related to the battlefield situation, while edge rewiring adjusts the causal and dependency relationships between events to adapt to real-time changes in the battlefield situation. These adjustments enable the Bayesian network to always maintain an accurate description of the battlefield situation, thereby improving the reliability and effectiveness of the knowledge reasoning model.

[0075] Furthermore, it can be understood that the process of establishing a decision-making knowledge base inference model based on a dynamic Bayesian network follows the following: First, in response to the uncertainty and incompleteness of battlefield situation information, Bayesian network technology is used to establish a probabilistic model of battlefield events. Bayesian networks are an uncertainty knowledge representation and inference model based on probability analysis and graph theory. By defining conditional independence between variables, they significantly reduce the number of probabilities required to define the full joint probability distribution, thus gaining widespread application in uncertainty reasoning and decision-making problems. This can connect multiple combat entities at different battlefield situation levels, constructing and evolving a Bayesian network knowledge base model capable of perceiving, reasoning, and predicting the battlefield situation, making it an effective modeling method for battlefield situation estimation.

[0076] Each node in a Bayesian network represents a random variable, and arrows connect nodes, indicating causal relationships. The relationships between nodes are described using conditional probability tables. Given a node's prior probability and conditional probability, the state probabilities of each node can be calculated. Probabilities are then propagated along directed edges, completing the reasoning of the entire Bayesian network. Bayesian networks exploit the conditional independence of variables to decompose the joint distribution into the product of several local distributions, thereby simplifying the probabilistic analysis of complex systems.

[0077] A dynamic Bayesian network (DBN) is a time-dependent extension of a static Bayesian network, used to describe models that change and evolve over time. Composed of a probability distribution function for a sequence of implicit state variables and a sequence of observed variables, a dynamic Bayesian network can reflect state changes across time slices and the probabilistic relationships between nodes within the same time slice. By defining probability functions and initial states over time series, a dynamic Bayesian network can more accurately describe the dynamics of battlefield situations.

[0078] When using a Bayesian network knowledge base for operational decision-making reasoning and prediction, two key issues must be addressed: extracting battlefield situational features and acquiring domain knowledge. These two factors together determine the structure of the Bayesian network. Because the increasing number of situational elements exponentially increases the complexity of the network structure, dynamic operational decision-making estimation first relies on the domain knowledge of military experts to make necessary network modifications. By rationally selecting features that reflect the battlefield situation and constructing correlations between these features based on domain knowledge, a Bayesian network inference model that reflects the operational situation can be constructed. This process ensures that the model accurately reflects the dynamic changes in the battlefield situation, providing a scientific basis for operational decision-making.

[0079] In one embodiment, if Figure 1 As shown, step S50 receives combat instructions, generates multi-dimensional decision suggestions using a knowledge reasoning model based on combat rules matched with the knowledge graph in the decision knowledge base, evaluates and screens the decision suggestions through decision analysis, and displays the decision results through a user interaction module. The steps include the following:

[0080] S501. Receive the commander's combat order, match the combat order with entities and relationships in the current battlefield situation knowledge graph, and identify military rules and situation elements related to the current combat mission;

[0081] S502. Utilize the fuzzy inference rules in the decision knowledge base to generate multi-dimensional decision suggestions based on the matched operational rules and situation information, and rank these solutions according to probability;

[0082] S503. Using the decision analysis module, combined with actual operational requirements, the generated decision suggestions are evaluated and screened from multiple dimensions to generate one or more optimal decision solutions.

[0083] S504: Display the decision result to the commander through the user interaction module and receive feedback from the commander.

[0084] Understandably, the generation of decision paths is an advanced stage of multi-source information fusion. Its core lies in multi-objective optimization based on expert experience and established rules, with a focus on building a suitable knowledge base and designing efficient and reasonable fusion mechanisms and generalized reasoning capabilities. The decision-making process is typically divided into three stages: situational awareness and assessment, threat assessment, and decision-making. Specifically, through real-time perception of the battlefield situation, situational knowledge units are rapidly extracted and a database is constructed, which in turn generates a knowledge graph. This process ensures the timely acquisition and organization of critical battlefield information, providing data support for subsequent decision-making analysis. Real-time perception and knowledge extraction capabilities enable rapid response to battlefield changes and provide commanders with the latest situational information.

[0085] Next, fuzzy reasoning is performed using the constructed knowledge graph. The fuzzy reasoning module can handle the uncertainty and ambiguity in battlefield situations and generate multiple possible decision recommendations. This process combines fuzzy logic with the structured information in the knowledge graph, providing flexible and reasonable decision options in complex and changing battlefield environments.

[0086] At the same time, dynamic decision analysis is also performed in conjunction with Bayesian networks. By modeling the probabilistic relationships between battlefield events, the Bayesian network module can dynamically update battlefield situation estimates and ensure that decision plans accurately reflect the causal and dependency relationships within the battlefield situation, further enhancing the system's decision-making capabilities and adaptability.

[0087] After the commander inputs the operational order through the user interaction module, the system matches the operational order with the information in the knowledge graph, and combines fuzzy reasoning and Bayesian network analysis to finally generate a specific decision-making scheme. This process reflects the intelligent and automated capabilities of the system, which can quickly generate the optimal decision-making scheme according to the commander's intentions and battlefield situation. In addition, the military rules stored in the knowledge base are matched with the actual "scenario space" knowledge, and the corresponding decision result is selected, and the corresponding decision scheme is inferred, and it can also assist the commander to carry out sand table deduction on the future situation.

[0088] Finally, the decision-making scheme is displayed to the commander through a graphical interface, so that the commander can intuitively understand the decision-making result and make adjustments or provide feedback as needed. This user interaction link not only improves the usability of the system, but also further optimizes the decision-making ability of the system through the feedback mechanism.

[0089] It should be understood that the size of the serial number of each step in the above embodiments does not mean the order of execution, and the execution order of each process should be determined according to its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0090] In a specific embodiment, the operation process of the multi-domain battlefield task decision-making method based on the knowledge graph is as follows:

[0091] In step S10, the operational element situation entity item is extracted, and the included relationship data is extracted <包括> ， <采取> <c> ， <c><分配> <fmyn> ， <c><demand capacity> <d>, <F_x> <have the ability> <d>,<Y_y><分配能力> <d>That is, goal A includes sub-goal B, sub-goal B takes task C, task C requires m drone nodes and n unmanned vehicle nodes, task C requires unmanned nodes with capability D, drone nodes of model F_x have capability D, and unmanned vehicle nodes of model Y_y have capability D.

[0092] In step S20, the relationship between the situation entity entries is constructed, the uploaded text file is parsed, and the entities and the triple relationships between them (subject, predicate, object) are extracted. According to different predicate types, the code classifies these entities into entity nodes or action nodes and writes them into corresponding collections. Then, based on the extracted relationships, the code constructs a hierarchical structure between the entities and determines the ancestors and child nodes of each entity through a recursive method. In addition, the code also manages rules and policies to record the specific associations between entities and actions. The resource allocation part analyzes the needs of each entity and matches them to ensure the rational allocation of capabilities and resources. Finally, all these relationships and hierarchical structures are organized into a directed graph structure to form a systematic relationship network for subsequent query and analysis. The entire process realizes the systematic management and intuitive display of complex relationships between situation entities.

[0093] In step S30, a decision-making knowledge base is constructed based on fuzzy reasoning. The relevant parameters of the combat command method are quantified to form a continuous-valued decision table. This is then converted into a fuzzy decision table through fuzzification to accommodate the uncertainty and ambiguity of the battlefield situation. The inclusion degree of each entry is then calculated, and decision rules with high inclusion degrees are selected and stored in the decision-making knowledge base.

[0094] In step S40, a knowledge inference model is constructed using dynamic Bayesian network technology. By establishing a probabilistic relationship model for battlefield events, the system can dynamically update the battlefield situation estimate and rank possible actions. This model provides probabilistic support for subsequent fuzzy decision-making.

[0095] In step S50, fuzzy decision-making is performed based on the knowledge graph and operational instructions. The system first receives the commander's operational instructions and matches them with the entities and relationships in the knowledge graph. Fuzzy inference rules are used to generate multiple possible decision suggestions. These suggestions are then evaluated and filtered through the decision analysis module to produce the optimal solution. The decision results are presented to the commander through the user interaction module, allowing the commander to make adjustments and provide feedback as needed. The system then dynamically optimizes the decision-making process based on this feedback.

[0096] Taking the input example in step S10 as an example, if A is input, the system will output A's subtask B, the number of unmanned nodes required for B, the unmanned node allocation results based on the required capacity, and the required task C. If E is input (E is not in the database, but the character or semantic similarity with A is greater than a set value), the system will provide A's decision result as a reference based on the fuzzy inference decision model. This process embodies the system's complete logic, from data extraction to relationship construction to fuzzy decision-making, ensuring the scientific nature and dynamic adaptability of decision-making.

[0097] In one embodiment, a multi-domain battlefield task decision system based on a knowledge graph is provided, and the multi-domain battlefield task decision system of the knowledge graph corresponds one-to-one to the multi-domain battlefield task decision method of the knowledge graph in the above embodiment.< / d> < / d> < / d> < / c> < / fmyn> < / c> < / c> Figure 4 As shown in the figure, the multi-domain battlefield task decision system of this knowledge graph:

[0098] An analysis and extraction module 100 is used to analyze the relationship between battlefield elements and tasks in the database to extract situation entity entries of combat elements;

[0099] A relationship building module 200 is used to build relationships between situation entity entries using predefined relationship templates;

[0100] A decision knowledge base construction module 300 is used to construct a decision knowledge base based on fuzzy reasoning;

[0101] The knowledge reasoning model construction module 400 is used to construct a knowledge reasoning model based on the decision knowledge base using dynamic Bayesian network technology;

[0102] The decision generation and display module 500 is used to receive combat instructions and generate multi-dimensional decision suggestions based on combat rules that match the knowledge graph in the decision knowledge base using a knowledge reasoning model, evaluate and screen the decision suggestions through decision analysis, and display the decision results through a user interaction module.

[0103] The decision-making system of the above-described embodiment of the present invention provides commanders with accurate and efficient decision-making support through a series of closely connected steps and functional modules. First, the system perceives the battlefield situation in real time, rapidly extracts situational knowledge units, constructs a database, and then generates a knowledge graph. This process is the foundational function of the system, ensuring that the system can promptly acquire and organize key battlefield information, providing data support for subsequent decision-making analysis. The real-time perception and knowledge extraction capabilities enable the system to quickly respond to battlefield changes and provide commanders with the latest situational information.

[0104] The system then uses the constructed knowledge graph to perform fuzzy reasoning. This fuzzy reasoning module can handle the uncertainty and ambiguity of battlefield situations and generate multiple possible decision recommendations. This process combines fuzzy logic with the structured information in the knowledge graph, enabling the system to provide flexible and reasonable decision options in complex and changing battlefield environments.

[0105] The system also incorporates a Bayesian network for dynamic decision analysis. By modeling the probabilistic relationships between battlefield events, the Bayesian network module dynamically updates battlefield situation estimates and ensures that decision plans accurately reflect the causal and interdependent relationships within the battlefield situation. The introduction of this module further enhances the system's decision-making capabilities and adaptability.

[0106] After the commander inputs the combat order through the user interaction module, the system matches the combat order with the information in the knowledge graph, and combines fuzzy reasoning and Bayesian network analysis to finally generate a specific decision scheme. This process reflects the intelligent and automated capabilities of the system, which can quickly generate the optimal decision scheme according to the commander's intentions and battlefield situation.

[0107] Finally, the decision scheme is displayed to the commander through a graphical interface, allowing the commander to intuitively understand the decision results and make adjustments or provide feedback as needed. This user interaction not only improves the usability of the system, but also further optimizes the decision-making capabilities of the system through the feedback mechanism.

[0108] The multi-task decision system based on the knowledge graph has complete battlefield situation awareness, knowledge graph construction, fuzzy reasoning, and decision analysis functions. Through real-time sensing and knowledge extraction, combined with fuzzy reasoning and Bayesian network technology, the system can effectively handle the uncertainty and complexity of the battlefield situation, providing accurate and efficient decision support for commanders.

[0109] The specific limitations of the multi-domain battlefield task decision system based on the knowledge graph can be referred to the limitations of the multi-domain battlefield task decision method based on the knowledge graph in the above, which will not be repeated here. Each module in the above multi-domain battlefield task decision system based on the knowledge graph can be realized by software, hardware, and their combinations. The above modules can be embedded in or independent of the processor in the computer device in hardware form, or stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0110] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above functional units and modules is exemplified, and in actual application, the above functions can be completed by different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above.

[0111] The above-described embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for some technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application. < / f>

Claims

1. A multi-domain battlefield task decision-making method based on knowledge graph, characterized by: include: S10, analyzing the relationship between battlefield elements and tasks in the database to extract situation entity entries of combat elements; S20, constructing relationships between situation entity entries using a predefined relationship template; S30. Constructing a decision knowledge base based on fuzzy reasoning; S40. Based on the decision knowledge base, a knowledge reasoning model is constructed using dynamic Bayesian network technology; S50, receiving combat instructions, and based on combat rules that match the knowledge graph in the decision knowledge base, using the knowledge reasoning model to generate multi-dimensional decision suggestions, evaluating and screening the decision suggestions through decision analysis, and displaying the decision results through the user interaction module.

2. The multi-domain battlefield task decision-making method based on knowledge graph according to claim 1 is characterized in that: The step S10 includes: S101. Analyze battlefield elements and tasks in the database, and then sort out the inherent connections between battlefield elements and tasks; S102. Combine combat instructions and real-time perception information to obtain specific data from different information sources, including extracting combat node data from battlefield resource information, extracting lighting conditions from environmental information, and extracting tactical objectives from combat instructions; S103. Utilize natural language processing (NLP) technology to automatically extract and summarize specific situational knowledge units from specific data as situational entity entries for combat elements.

3. The multi-domain battlefield task decision-making method based on knowledge graph according to claim 2 is characterized in that: The step S20 includes: S201, using predefined relationship templates to construct relationships between tasks and goals, between tasks and allocation requirements, and between tasks and rules; S202. During the task execution process, further refine the relationship and conditions between the task and the allocation requirements, and consider the requirements for the unmanned node capabilities during task execution to ensure that the nodes have specific capabilities; S204 : Display the relationship between the constructed entity entries in a visual manner. Each entity entry is represented in the form of a node. The relationship between the nodes is connected by directed edges, and the relationship type is marked on the edge.

4. The multi-domain battlefield task decision-making method based on knowledge graph according to claim 3 is characterized in that: The step S30 includes: S301. Based on the fuzzy reasoning and rough set theory and methods, valuable knowledge is extracted from the original observation data to build a fuzzy decision model.

5. The multi-domain battlefield task decision-making method based on knowledge graph according to claim 4 is characterized in that: The step S301 includes: S3011. Based on the extracted valuable knowledge, construct a continuous value decision table for combat command mode; S3012. Use fuzzy processing to convert the continuous value decision table into a fuzzy decision table; S3013. Calculate the inclusion degree of the fuzzy decision table and select decision rules with high inclusion degree; S3014. Extract fuzzy inference rules based on high-inclusion decision rules, build a decision knowledge base and store it.

6. The multi-domain battlefield task decision-making method based on knowledge graph according to claim 4 is characterized in that: The step S40 includes: S401. Using a Bayesian network to establish a probability model for battlefield events, a static model is constructed by defining conditional independence and probability relationships between variables; the expression of the static model is: P(x1,x2,…,x n )=∏ i P(x i |when i ) (1) Among them, (x1,x2,…,x n ) represents the implicit state variable, P(x i |pa i ) indicates that in the parent node pa i Under the condition of the value of i Take the probability distribution of a specific state; S402: Expand the static model on the time axis through a dynamic Bayesian network to adapt to time changes. The expression of the dynamic Bayesian network is: Among them, P(x i |x i-1 ) represents the probability function between states at different time slices; P(y i |x i ) represents the probability function of each node on the same time slice; P(x0) represents the initial state at the beginning of the process; S403. Extract the characteristics of battlefield situation elements and acquire domain knowledge, adjust and optimize the dynamic Bayesian network structure to construct a knowledge reasoning model of the Bayesian network to ensure accurate reflection of the causal and dependency relationships of the battlefield situation.

7. The multi-domain battlefield task decision-making method based on knowledge graph according to claim 6 is characterized in that: The step S50 includes: S501. Receive the commander's combat order, match the combat order with entities and relationships in the current battlefield situation knowledge graph, and identify military rules and situation elements related to the current combat mission; S502. Utilize the fuzzy inference rules in the decision knowledge base to generate multi-dimensional decision suggestions based on the matched operational rules and situation information, and rank these solutions according to probability; S503. Using the decision analysis module, combined with actual operational requirements, the generated decision suggestions are evaluated and screened from multiple dimensions to generate one or more optimal decision solutions. S504: Display the decision result to the commander through the user interaction module and receive feedback from the commander.

8. A multi-domain battlefield mission decision system based on knowledge graph, characterized by: include: An analysis and extraction module, used to analyze the relationship between battlefield elements and tasks in the database to extract situation entity entries of combat elements; The relationship building module is used to build relationships between situation entity entries using predefined relationship templates; Decision knowledge base construction module, used to construct decision knowledge base based on fuzzy reasoning; The knowledge reasoning model construction module is used to construct a knowledge reasoning model based on the decision knowledge base using dynamic Bayesian network technology; The decision generation and display module is used to receive combat instructions and generate multi-dimensional decision suggestions based on combat rules that match the knowledge graph in the decision knowledge base using the knowledge reasoning model. The decision suggestions are evaluated and screened through decision analysis, and the decision results are displayed through the user interaction module.

Images