CN-121980921-A - Multi-dimensional model intelligent modeling method based on intention analysis and knowledge driving

CN121980921ACN 121980921 ACN121980921 ACN 121980921ACN-121980921-A

Abstract

The application provides an intelligent modeling method of a multi-dimensional model based on intention analysis and knowledge driving, which relates to the technical field of simulation calculation, and comprises the steps of receiving multi-source heterogeneous input data of a target; the method comprises the steps of analyzing modeling intention and characteristic information set based on a first AI model for multi-source heterogeneous input data, generating an entity layered semantic ontology model instance based on a predefined entity layered semantic ontology model, a modeling characteristic knowledge base and a second AI model according to the modeling intention and the characteristic information set, and generating an executable simulation construction script of a target through a third AI model based on the entity layered semantic ontology model instance. The application realizes agile, extensible and standardized entity intelligent construction, and improves the simulation deduction efficiency.

Inventors

WANG YINGHE
WEI XI
ZHENG YANG
ZHANG JING
WANG JING
ZHANG JINGYA
HAN JING
LIN HUI
GUO XIAOLEI
HE CHANGYU
ZHANG QIANYUE
WANG YINGXUE
ZHANG XINHAI
LI HUIBO

Assignees

中国电子科技集团有限公司电子科学研究院

Dates

Publication Date: 20260505
Application Date: 20251226

Claims (10)

1. The intelligent modeling method of the multi-dimensional model based on intention analysis and knowledge driving is characterized by comprising the following steps of: Receiving multi-source heterogeneous input data of a target, wherein the multi-source heterogeneous input data comprises natural language instructions, structured data and image information; Analyzing modeling intention and characteristic information set based on a first AI model for the multi-source heterogeneous input data; Generating an entity layered semantic ontology model instance based on a predefined entity layered semantic ontology model, a preset modeling feature knowledge base and a second AI model according to a modeling intention and a feature information set, wherein the entity layered semantic ontology model comprises attribute representations of a physical domain, a perception domain, a network communication domain and a behavior domain of an entity; And generating an executable simulation construction script of the target through a third AI model based on the entity layered semantic ontology model instance.
2. The method for intelligent modeling of a multi-dimensional model based on intent resolution and knowledge actuation as claimed in claim 1, wherein, The physical domain describes static and dynamic physical attributes of the entity; describing the type, performance parameters, information processing logic and performance degradation of the sensor carried by the entity in the actual environment; the network communication domain describes the communication capability and/or information interaction capability of the entity; The behavior domain describes task targets, decision rules and behavior logic under uncertain conditions of the entity.
3. The method for intelligent modeling of a multi-dimensional model based on intent resolution and knowledge actuation as claimed in claim 1, wherein, The generating an entity layered semantic ontology model instance based on a predefined entity layered semantic ontology model, a preset modeling feature knowledge base and a second AI model according to the modeling intention and the feature information set specifically comprises the following steps: Carrying out semantic alignment on the analyzed characteristic information set and the entity layered semantic ontology model; retrieving modeling knowledge related to the modeling intent from the modeling feature knowledge base; and generating the entity layered semantic ontology model instance based on the second AI model according to the semantic alignment result and the retrieved modeling knowledge.
4. The method for intelligent modeling of a multi-dimensional model based on intent resolution and knowledge actuation as claimed in claim 3, Constructing the modeling feature knowledge base, including: Extracting semantic vectors of component definitions and parameter assignment and combination relations thereof from source codes of a historical simulation model through code analysis and natural language processing technologies, and constructing knowledge triples; Forming a knowledge graph of standardized modeling knowledge based on the knowledge triples; Mapping the symbolic representations in the knowledge graph into vectors for semantic retrieval.
5. The method for intelligent modeling of a multi-dimensional model based on intent resolution and knowledge actuation as claimed in claim 1, wherein, The generating, based on the entity layered semantic ontology model instance, an executable simulation construction script of the target through a third AI model includes: Decomposing the complete modeling task into a plurality of subtasks, wherein the subtasks are determined according to components in the entity layered semantic ontology model instance and are used for generating executable simulation modeling codes at a component level; aiming at each subtask, constructing prompt information comprising a target simulation deduction platform code template, parameter constraint and a generation rule, wherein the prompt information is used for guiding the third AI model to generate a gap-filling code; The component relation in the entity layering semantic ontology model instance is spliced with the generated executable simulation modeling codes of the component level to form the executable simulation construction script; In the process of generating the executable simulation modeling codes at the component level, limiting an output vocabulary of the third AI model according to grammar rules of the target script language so as to ensure grammar correctness; And carrying out real-time verification on the generated parameter assignment according to the parameter range knowledge in the modeling feature knowledge base so as to ensure semantic correctness.
6. The method for intelligent modeling of a multi-dimensional model based on intent resolution and knowledge driving as claimed in any one of claims 1-5, wherein, The method further comprises the steps of model verification and closed loop optimization: performing static grammar and logic check on the generated script; Presenting the model generated according to the script to a user in the form of a structured view and a natural language abstract for confirmation and feedback; and carrying out iterative optimization on the first AI model, the second AI model and the third AI model based on feedback data.
7. The utility model provides a multidimensional model intelligent modeling device based on intention analysis and knowledge drive which characterized in that includes: The data input module is used for receiving multi-source heterogeneous input data of a target, wherein the multi-source heterogeneous input data comprises natural language instructions, structural parameters and image information; The data analysis module is used for analyzing modeling intention and characteristic information set of the multi-source heterogeneous input data based on a first AI model; The system comprises an entity layering semantic ontology model instance construction module, a modeling feature knowledge base and a modeling feature knowledge base, wherein the entity layering semantic ontology model instance construction module is used for generating an entity layering semantic ontology model instance based on a predefined entity layering semantic ontology model, a preset modeling feature knowledge base and a second AI model according to modeling intention and feature information sets, wherein the entity layering semantic ontology model comprises attribute representations of a physical domain, a perception domain, a network communication domain and a behavior domain of an entity; And the script intelligent generation module is used for generating an executable simulation construction script of the target through a third AI model based on the entity layered semantic ontology model instance.
8. The multi-dimensional model intelligent modeling apparatus based on intention analysis and knowledge driving according to claim 7, wherein, The physical domain describes static and dynamic physical attributes of the entity; describing the type, performance parameters, information processing logic and performance degradation of the sensor carried by the entity in the actual environment; the network communication domain describes the communication capability and/or information interaction capability of the entity; The behavior domain describes task targets, decision rules and behavior logic under uncertain conditions of the entity.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, which when executed by the processor performs the steps of the intent resolution and knowledge driven multi-dimensional model intelligent modeling method according to any of claims 1 to 6.
10. A readable storage medium, characterized in that it has stored thereon a program or instructions, which when executed by a processor, implement the steps of the intent resolution and knowledge driven based intelligent modeling method of a multidimensional model according to any of claims 1 to 6.

Description

Multi-dimensional model intelligent modeling method based on intention analysis and knowledge driving Technical Field The invention relates to the technical field of simulation calculation, in particular to an intention analysis and knowledge driving-based multi-dimensional model intelligent modeling method, device, equipment and medium. Background Driven by new technological revolution and industrial revolution, the demand of simulation deduction is growing, task entities (such as space, air, land and sea multi-domain equipment) are increasingly presenting highly complex, systematic and intelligent characteristics, and the types of equipment, interaction relations and the like related to the corresponding models in the simulation system are also becoming more diverse. However, the modeling simulation technology in the related art still depends on expert experience and manual coding, and has long modeling period, high cost and insufficient flexibility and adaptability. In the face of rapid iteration of equipment load, continuous evolution of task theory and instantaneous change of actual environment, the method can not meet actual combat requirements of agile simulation and rapid response, and is difficult to support deduction and presumption verification of high frequency. Furthermore, each mechanism and even different simulation deduction systems often adopt an independent design and non-standardized special model system, the types of data resources relied and generated by the simulation deduction are also rich and various (including structural parameters, unstructured texts, images, signal data and the like), the quantity is huge, the granularity difference is obvious, serious repeated development and resource waste are caused, and more importantly, the deep fusion and efficiency evaluation of the cross-domain and cross-system simulation are fundamentally blocked, and the comprehensive integrated and efficient collaborative simulation deduction is difficult to realize. Disclosure of Invention The invention provides a multi-dimensional model intelligent modeling method, device, equipment and medium based on intention analysis and knowledge driving, which solve the problem of how to realize agile, extensible and standardized solid modeling to improve the simulation deduction efficiency. In order to achieve the above purpose, the application adopts the following technical scheme: In a first aspect, an intent analysis and knowledge driven based multi-dimensional model intelligent modeling method is provided, including: Receiving multi-source heterogeneous input data of a target, wherein the multi-source heterogeneous input data comprises natural language instructions, structured data and image information; Analyzing modeling intention and characteristic information set based on a first AI model for the multi-source heterogeneous input data; Generating an entity layered semantic ontology model instance based on a predefined entity layered semantic ontology model, a preset modeling feature knowledge base and a second AI model according to a modeling intention and a feature information set, wherein the entity layered semantic ontology model comprises attribute representations of a physical domain, a perception domain, a network communication domain and a behavior domain of an entity; And generating an executable simulation construction script of the target through a third AI model based on the entity layered semantic ontology model instance. In a second aspect, there is provided a multi-dimensional model intelligent modeling apparatus based on intent resolution and knowledge driving, comprising: The data input module is used for receiving multi-source heterogeneous input data of a target, wherein the multi-source heterogeneous input data comprises natural language instructions, structural parameters and image information; The data analysis module is used for analyzing modeling intention and characteristic information set of the multi-source heterogeneous input data based on a first AI model; The system comprises an entity layering semantic ontology model instance construction module, a modeling feature knowledge base and a modeling feature knowledge base, wherein the entity layering semantic ontology model instance construction module is used for generating an entity layering semantic ontology model instance based on a predefined entity layering semantic ontology model, a preset modeling feature knowledge base and a second AI model according to modeling intention and feature information sets, wherein the entity layering semantic ontology model comprises attribute representations of a physical domain, a perception domain, a network communication domain and a behavior domain of an entity; And the script intelligent generation module is used for generating an executable simulation construction script of the target through a third AI model based on the entity layered semantic ontology model instance. In a third asp