对抗环境下的智能兵棋系统设计及其关键技术OA北大核心CSTPCD

The intelligent game confrontation has become one of the hot research fields.The system construction and module design of the wargame system,the modeling elements of the wargaming system,including the wargame elements,wargame rules and intelligent interface design,and the overall architecture of the intelligent wargame system.The feasibility of the system design is verified by A3C reinforcement learning intelligent algorithm.The reward setting of the reinforcement learning and training process is improved,and the state input,algorithm driving process and action output process of the intelligent wargame environment are clarified.The system theory and work are verified by the intelligent wargame system realized independently.This work provides a feasible path for the design and implementation of intelligent game system based on reinforcement learning,and provides a basic platform for the future research of intelligent game confrontation based on reinforcement learning.