생성형 AI와 RAG 기술을 활용한 국방군수통합정보체계(DELIIS)의 지능형 에이전트 구축 방안 연구

초록

Defense Innovation 4.0 necessitates data-driven logistics, yet current DELIIS systems suffer from complex interfaces and high cognitive loads. This study proposes an intelligent agent architecture integrating LLM and RAG for task automation. A rule-based simulation PoC, using 35,000 mock records, demonstrated an 80% reduction in information access steps and over 90% reduction in processing time for mid-term planning. Integrated security measures, including Prompt Injection prevention and Role-Based Access Control (RBAC), were defined to ensure operational safety. These results confirm the proposed agent's potential to transform military logistics into a proactive, intelligence-led system.

키워드