针对飞机编队对抗仿真场景中行为模型构建过程复杂、人工依赖程度高等问题，通过融合大语言模型技术，实现飞机编队协同对抗行为的智能化建模。构建一种融合大语言模型技术的复合分层行为建模框架，实现语义驱动的自动化建模流程；设计一种基于“系统-用户”分层架构的提示工程方法，实现概念语义到形式化状态机的转化；提出一种基于向量检索机制的“状态-行为”关联方法，实现决策状态与行为树节点的智能高效匹配。基于上述方法，构建典型飞机编队协同战术行为模型，并在典型对抗场景中开展仿真实验。仿真实验表明，该框架与方法可支撑编队协同对抗行为的智能化快速建模，验证了所提出框架的适用性以及建模方法的有效性。
To address the complexity and high manual dependency in behavior modeling process for cooperative combat simulation scenarios, this paper integrates large language model (LLM) technology to enable intelligent modeling of aircraft formation cooperative combat behaviors. A hybrid hierarchical modeling framework integrating LLM technology is proposed to ena-ble a semantics-driven and automated modeling workflow. A two-layer “system-user” prompt engineering method is de-signed to transform conceptual semantics into formalized state machines. A vector-retrieval-based association method is developed to achieve intelligent and efficient matching between decision states and behavior tree nodes. Based on these methods, representative cooperative combat behavior models for aircraft formations are constructed and validated through simulation experiments in typical combat scenarios. The experimental results demonstrate that the proposed framework and methods effectively support rapid and intelligent modeling of cooperative combat behaviors, confirming their applicability and practical effectiveness.
[1]杜梓冰, 张立丰, 陈敬志, 等. 有人/无人机协同作战演示验证试飞关键技术[J]. 航空兵器, 2019, 26(04): 75-81.
[2]张琪. 学习驱动的CGF决策行为建模方法研究[D]. 长沙: 国防科技大学, 2021.
[3]Bourassa N R. Modeling and simulation of fleet air defense systems using EADSIM[D]. Monterey, Cali-fornia. Naval Postgraduate School, 1993.
[4]Clive P D, Johnson J A, Moss M J, et al. Advanced framework for simulation, integration and modeling (AFSIM) [C]//Proceedings of the international confer-ence on scientific computing (CSC). The Steering Committee of The World Congress in Computer Sci-ence, Computer Engineering and Applied Computing (World-Comp), 2015: 73.
[5]王鹏, 刘昊雨, 李妮, 等. 基于改进ABC算法的有-无人协同空战行为建模[J]. 系统仿真学报, 2024, 36(12): 2871-2883.
[6]刘波, 魏潇龙, 屈虹, 等. 有人——无人机协同空战机动决策研究[J]. 航空工程进展, 2023, 14(06): 63-72.
[7]Mnih V, Kavukcuoglu K, Silver D, et al. Playing atari with deep reinforcement learning[J]. arXiv preprint arXiv:1312.5602, 2013.
[8]陈希亮, 张永亮. 基于深度强化学习的陆军分队战术决策问题研究[J]. 军事运筹与系统工程, 2017, 31(03): 20-27+57.
[9]王宇琨, 王泽, 董力维, 等. 基于分层的智能建模方法的多机空战行为建模[J]. 系统仿真学报, 2023, 35(10): 2249-2261.
[10]刘玮, 张永亮, 程旭. 基于深度强化学习的人机智能对抗综述[J]. 指挥信息系统与技术, 2023, 14(2): 28-37.
[11]王彤, 赵美静, 徐沛, 等. 基于大语言模型的兵棋推演智能决策技术[J]. 自动化学报, 2025, 51(06): 1205-1217.
[12]刘大勇, 董志明, 郭齐胜, 等. LLM赋能的战术兵棋决策Agent构建方法[J/OL]. 系统仿真学报, 1-16[2025-10-30]. https://link.cnki.net/urlid/11.3092.v.20250724.1354.004.
[13]Sahoo P, Singh A K, Saha S, et al. A systematic survey of prompt engineering in large language models: Techniques and applications[J]. arXiv preprint arXiv:2402. 07927, 2024.
[14]Lewis P, Perez E, Piktus A, et al. Retrieval-augmented generation for knowledge-intensive NLP tasks[J]. Ad-vances in Neural Information Processing Systems, 2020, 33: 9459-9474.
[15]DeepSeek. DeepSeek API文档[EB/OL]. (2025-09-29)[2025-10-30]. https://api-docs.deepseek.com/zh-cn/.
[16]Alibaba DAMO Academy. Thenlper/gte-base-zh[EB/OL]. [2025-10-30]. https://huggingface.co/thenlper/gte-base-zh.
[17]月之暗面科技有限公司. Moonshot AI开放平台-Kimi大模型API服务[EB/OL]. (2025-10-20)[2025-10-30]. https://platform.moonshot.cn/docs/overview.
[18]阿里云. 阿里云百炼文档[EB/OL]. [2025-10-30]. https://bailian.console.aliyun.com/.
[19]火山引擎. 火山方舟大模型服务平台[EB/OL]. (2025-10-15)[2025-10-30]. https://www.volcengine.com/docs/82379/.
[20]Beijing Academy of Artificial Intelligence. BAAI/bge-base-zh[EB/OL]. (2023-12-10)[2025-10-30]. https:// hugging face.co/BAAI/bge-large-zh.
[21]WANG Y X, SUN Q X, HE S C. M3E: Moka massive mixed embedding model[EB/OL]. [2025-10-30]. https://huggingface.co/moka-ai/m3e-base.
[22]XU M. text2vec: A tool for text to vector[EB/OL]. [2025-10-30]. https://github.com/shibing624/text2vec.
CJA