Knowledge graph-driven methodology for complex product architecture solution generation and simulation verification

Complex product architecture design is a key process that determines system performance, development efficiency, and reliability. The intelligent transformation of this process is of significant importance in responding to dynamic demands and reducing R&D costs. Traditional architecture design methods face challenges such as reliance on experience, low knowledge reuse, and insufficient efficiency in solution generation and validation. To address these issues, this paper proposes a knowledge graph-driven approach for the generation and simulation validation of complex product architectures. Through structured knowledge management and application, this method enhances the intelligence of product architecture design and decision-making support capabilities. Firstly, a multi-source knowledge graph that integrates product, feature, and decision data is constructed to support semantic associations and dynamic inference of design knowledge. Secondly, an adjacency matrix is created based on graph matching, generating an initial architecture solution. This solution is then refined using multi-objective optimization algorithms and multi-attribute decision-making methods for parameter configuration and trade-off analysis. Finally, a mechanism model is employed to dynamically evaluate the solution's performance, creating a closed-loop feedback system to iteratively optimize and shorten the design verification cycle. A case study of the separation system of a launch vehicle's primary and second stages demonstrates the entire process of function-component matching, solution generation, and simulation validation under the knowledge graph-driven approach.