TY - GEN
T1 - Knowledge Graph Enhanced Large Language Model for Missile Design
AU - Fan, Yi
AU - Sun, Yu
AU - Mi, Baigang
N1 - Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2027.
PY - 2027
Y1 - 2027
N2 - Missile technology, central to precision-guided weapons, demands systematic improvements in overall design methods for enhancing intelligence, precision, and cost-efficiency. Traditional empirical design heavily relies on subjective expert knowledge, leading to inconsistencies and high iterative costs. To address these issues, this study proposes an innovative framework integrating knowledge graph (KG) and large language model (LLM) technologies into missile overall design. We construct a multi-level knowledge association fusion model based on missile modular design philosophy and ontological modeling, achieving structured knowledge representation. Leveraging LLM-based open information extraction with chain-of-thought prompting, we systematically organize and fuse missile design experiences and instances. Furthermore, employing retrieval-augmented generation (RAG), we implement an interpretable missile design knowledge Q&A recommendation system. The developed KG encompasses 18,150 entities and 40,017 relations, demonstrating strong connectivity and comprehensive coverage. The resulting recommendation system effectively supports missile design layouts and instance-specific inquiries, significantly advancing missile overall design technology toward agile, cost-effective, and precise development.
AB - Missile technology, central to precision-guided weapons, demands systematic improvements in overall design methods for enhancing intelligence, precision, and cost-efficiency. Traditional empirical design heavily relies on subjective expert knowledge, leading to inconsistencies and high iterative costs. To address these issues, this study proposes an innovative framework integrating knowledge graph (KG) and large language model (LLM) technologies into missile overall design. We construct a multi-level knowledge association fusion model based on missile modular design philosophy and ontological modeling, achieving structured knowledge representation. Leveraging LLM-based open information extraction with chain-of-thought prompting, we systematically organize and fuse missile design experiences and instances. Furthermore, employing retrieval-augmented generation (RAG), we implement an interpretable missile design knowledge Q&A recommendation system. The developed KG encompasses 18,150 entities and 40,017 relations, demonstrating strong connectivity and comprehensive coverage. The resulting recommendation system effectively supports missile design layouts and instance-specific inquiries, significantly advancing missile overall design technology toward agile, cost-effective, and precise development.
KW - Knowledge graph (KG)
KW - Large language model (LLM)
KW - Missile design
KW - Retrieval augmented generation (RAG)
UR - https://www.scopus.com/pages/publications/105046916580
U2 - 10.1007/978-981-92-1319-1_6
DO - 10.1007/978-981-92-1319-1_6
M3 - 会议稿件
AN - SCOPUS:105046916580
SN - 9789819213184
T3 - Lecture Notes in Mechanical Engineering
SP - 69
EP - 81
BT - Proceedings of The 2025 Asia-Pacific International Symposium on Aerospace Technology - Proceedings of APISAT 2025
A2 - Suk, Jinyoung
A2 - Jeong, Shinkyu
A2 - Park, Donghun
PB - Springer Science and Business Media Deutschland GmbH
T2 - Asia-Pacific International Symposium on Aerospace Technology, APISAT 2025
Y2 - 27 October 2025 through 29 October 2025
ER -