A MBSE and LLM driven method for intelligent generation of aerial bomb design solution

Abstract Model-based systems engineering (MBSE) faces significant challenges in knowledge reuse and design agility. To address this gap, this study proposes and constructs an intelligent conceptual design framework for weapon systems driven by Large Language Models (LLMs). The framework operates in two stages. First, in an offline process, it constructs a hierarchical knowledge index of tactical capabilities using semantic clustering and an LLM. This structured index then enables a novel multi-agent hierarchical inference mechanism during the online phase, which performs a top-down parallel search and pruning on user requirements to efficiently generate design solutions. Experimental results demonstrate that the framework significantly outperforms generic baseline methods in the accuracy, correctness, and validity of the generated solutions. This study confirms that integrating LLMs with structured reasoning provides an effective path for enhancing the design agility of complex systems.