CN-122021851-A - Knowledge graph-based complex system component level design automation method

CN122021851ACN 122021851 ACN122021851 ACN 122021851ACN-122021851-A

Abstract

The invention provides a complex system component level design automation method based on a knowledge graph, which comprises the steps of defining a semantic mapping rule of functions and components based on formalized logic, wherein the semantic mapping rule comprises logic association of function types, interface attributes, component types and interface attributes, automatically mapping function requirements into a candidate component set conforming to semantic constraints through knowledge graph semantic reasoning, automatically carrying out combined search on the candidate component set based on compatibility constraints of input and output interfaces among components, screening to generate a feasible component level design scheme meeting a complete function chain, and carrying out quantitative evaluation on the design scheme through heterogeneous index standardization and weighted aggregation by adopting a multi-attribute decision model integrating performance, cost, risk and robustness indexes to output an optimal design scheme.

Inventors

NIU BIAO
Sang Shuhan
LI WENLONG
LUO WEIFENG
CHEN JIAWEN
FAN XIAOYANG
WANG KAIDI

Assignees

杭州电子科技大学
湖南云箭集团有限公司

Dates

Publication Date: 20260512
Application Date: 20260119

Claims (10)

1. The complex system component level design automation method based on the knowledge graph is characterized by comprising the following steps of: the method comprises the steps of defining a semantic mapping rule of functions and components based on formalized logic, wherein the semantic mapping rule comprises logic association of function types, interface attributes, component types and interface attributes, and automatically mapping functional requirements into a candidate component set conforming to semantic constraint through knowledge graph semantic reasoning; Based on compatibility constraint of input and output interfaces among components, carrying out automatic combined search on the candidate component set, and screening to generate a feasible component level design scheme meeting a complete functional chain; and (3) adopting a multi-attribute decision model integrating performance, cost, risk and robustness indexes, carrying out quantitative evaluation on the design scheme through heterogeneous index standardization and weighted aggregation, and outputting an optimal design scheme.
2. The method for automating the design of complex system component level based on knowledge graph as set forth in claim 1, wherein the semantic mapping rule is configured to infer one or more components having a specific type capable of implementing a function based on the type of the function and constraints of an input stream thereof, and each component has a predefined association relationship with the function.
3. The method for automating the design of the component level of the complex system based on the knowledge graph of claim 1, wherein the candidate components are derived from a standardized component library, and the construction process of the component library comprises the steps of collecting the type, the performance parameters and the interface information of the components, carrying out standardized processing on the component data, storing the standardized data and updating and maintaining the component library periodically.
4. The method for automating the design of the component level of the complex system based on the knowledge graph of claim 1, wherein the automatized combined search adopts a depth-first search algorithm, traverses all component paths meeting the requirements in the search process, screens out a subset of components with data sources in the input and purposeful in the output, and forms a feasible component level design scheme.
5. The method for automating the design of complex system component level based on knowledge graph of claim 1, wherein the compatibility constraint between components further comprises a component type matching constraint, ensuring that the combined component chain is continuous in function type.
6. The knowledge graph-based complex system component level design automation method is characterized in that heterogeneous indexes of a multi-attribute decision model are standardized and weighted and aggregated, and specifically comprises the steps of respectively distributing weight coefficients for performance, cost, risk and robustness indexes, calculating comprehensive scores of a design scheme by combining component quantity adjustment factors after summarizing index data of all components, and normalizing the comprehensive scores through forward weighting performance, the robustness indexes, reverse weighting cost and the risk indexes by the component quantity adjustment factors.
7. The method for automating the design of the complex system component level based on the knowledge graph of claim 1, wherein the quantized evaluation further comprises a scheme optimizing step of selecting an optimal design scheme according to the comprehensive scoring result, optimizing and adjusting the suboptimal scheme to generate a plurality of alternative optimized schemes, and outputting a detailed evaluation report containing evaluation data.
8. The method for automating the design of complex system components based on knowledge graph as set forth in claim 1, wherein knowledge graph reasoning dynamically updates the semantic mapping rules and integrates interdisciplinary component properties to optimize component selection for dynamic adaptation of components within a system design cycle.
9. A complex system component level design automation system based on knowledge graph, for implementing the method of claim 1, comprising: the semantic mapping module is configured to automatically map the functional requirements into a candidate component set conforming to semantic constraint through knowledge graph semantic reasoning based on the semantic mapping rules of formalized logic definition functions and components; The component combination module is configured to automatically and combinedly search the candidate component set based on the compatibility constraint of the input and output interfaces among the components, and screen and generate a feasible component level design scheme meeting the complete function chain; the multidimensional evaluation optimization module is configured to quantitatively evaluate the design scheme by adopting a multi-attribute decision model integrating performance, cost, risk and robustness indexes and output an optimal design scheme through heterogeneous index standardization and weighted aggregation.
10. A non-transitory computer readable storage medium, characterized in that the storage medium has stored thereon a computer program which, when executed by a processor, implements the method of any of claims 1-8.

Description

Knowledge graph-based complex system component level design automation method Technical Field The invention belongs to the technical field of complex system design and knowledge graph, and particularly relates to a complex system component level design automation method based on a knowledge graph. Background As the scale and complexity of modern complex systems continue to increase, traditional system design approaches face a number of challenges. Design of complex systems often involves the fusion of interdisciplinary knowledge in conjunction with multi-domain components, requiring accurate and efficient translation of natural language user requirements into specific, machine-understandable component-level design solutions. The traditional method mainly relies on a designer with abundant experience to manually finish the requirement disassembly and the component screening and proper assembly, which is time-consuming and labor-consuming, is easily affected by artificial cognition limitation and experience difference, causes uneven design quality, and is difficult to cope with the dynamic update requirement caused by rapid iteration of component technology. Meanwhile, the attribute isomerism and the interface diversity of the interdisciplinary components enable the compatibility difficulty among the manual coordination components to be greatly increased, and further restrict the design efficiency and the system reliability. In recent years, model-based system engineering (MBSE) has become an important methodology for complex system design. MBSE describes the system requirements, structures and behaviors through a formal model, so that the consistency and the integrity of the design are effectively improved, and a unified communication carrier is provided for cross-team cooperation. However, the conventional MBSE method generally relies on manually constructing and maintaining formalized models, so that not only is the modeling threshold high and the period long, but also unstructured natural language requirements are difficult to process efficiently, and an effective semantic conversion and ambiguity resolution mechanism is lacked, so that distortion risks exist in the transmission between the requirements and the models. In addition, the model reusability of the existing MBSE scheme is poor, design results among different projects are difficult to quickly migrate, and a systematic automatic support is lacked in a component level design level, so that accurate mapping and dynamic adaptation from a function body to a component body cannot be realized. KG (knowledge graph) technology provides a new idea for complex system design by virtue of strong semantic representation and reasoning capability. By structurally storing and associating the domain knowledge, the knowledge graph can break the information island, provide knowledge support for the automatic mapping from the functional requirements to the component realization, and further lay a foundation for the component combination and the design scheme generation. However, the existing research still has obvious defects in practical application, namely on one hand, the existing KG reasoning process lacks strong structural constraint, is easy to generate a phenomenon of 'illusion', causes insufficient mapping accuracy of functions and components, and is difficult to ensure the reliability of design, on the other hand, the component combination is dependent on single semantic matching, the comprehensive consideration of multidimensional indexes such as cost, risk, robustness and the like is ignored, and when facing a large-scale component library, the searching algorithm is low in efficiency, and the candidate combination conforming to multiple constraint is difficult to be screened out quickly. Meanwhile, the compatibility judgment of the existing scheme on heterogeneous components is often limited to interface form matching, the deep verification on function type continuity and performance suitability is lacked, combination conflict is easy to occur, and the effectiveness and economy of the design scheme are difficult to balance. Therefore, how to combine the structural reasoning advantages of the knowledge graph with the efficient component combination algorithm solves the key problems of accurate conversion of natural language requirements, rapid screening of a large-scale component library, combination optimization under multi-dimensional constraint and the like, and develops an efficient and intelligent system design method to be a problem to be solved urgently by those skilled in the art. Disclosure of Invention Aiming at the defects and shortcomings in the prior art, the invention provides a knowledge-graph-based complex system component level design automation method, a knowledge-graph-based complex system component level design automation system and a storage medium. The method is characterized in that a semantic mapping rule between functions and com