Research On Virtual Campaign System With Its Agent Behavior

Under the background of strain from international polictical environment and limitation of military budget, the simulation training has been regarded as an important method. Higher discretion level, more simulation liveness and larger simulation scale were needed in modern military simulation training. Therefore the traditional operation simulation method cannot meet the modern needs. In this dissertation, the idea of virtual technology was adopted. The complex adaptive system,and artificial intelligence theory in virtual compaign system were built. It would be helpful to study the regulations of war and military conflict to improve the fight force of PLA.In this dissertation, the evolving model based on agent of virtual campaign system was given,the evolving model system structure of virtual campaign system was set up. The concept of combat agent, and the operating grocess architecture of agent-based virtual campaign system was conceived, the working process of combat agent was constructed. Then, operating process of planning submodel and the evolving of combat agent were studied. Finally, the multi-agents compaign system and its structure of reinforce learning were studied. The operation style of combat agent reinforcement learning and the learning framework of multi-agents compaign system were discussed.The main innovations were summarized as follows: 1. Theoretic innovation: the virtual operation system was studied in system view, including its concept, feature, system structure and the goal of studying. Then, the physical structure, logsitical structure and operating process were studied. Finally, the gaming behavior of operation agent was described. It will be helpful to develop the theoretic research of virtual operation system; 2. Methodological innovation: the evolving modeling method of agent-based virtual compaign system was given. the evolving problem of the combat agent's rules was proposed, the mathematic model of combat agent's rules evolving was given, Which enriched the research method of vrtual operation system. 3. Applied innovation: the evolving model of virtual operation system with the help of multi-agents learning theory was developped, and we study the reinforcement learning structure of combat agent was studied, which would enrich the research view of virtual operation system.