Research On The ATC Agent Modeling Based On BDI Model

Since the birth of artificial intelligence in the 1950 s, it attracted widespred attention. So far, many well-known domestic and foreign universities have set up research institutions which are used for special research of artificial intelligence. Multi-discipline technology has applications in artificial intelligence. Therefore, researching artificial intelligence has some challenges. With the development of computer technology and the promotion of global information tide, research and application of artificial intelligence are expending.Agent is a relatively new software paradigm that brings concepts from the theories of artificial intelligence into the mainstream realm of distributed systems. In China, Agent has been used to build the air traffic simulation system. But it focused on building communication and collaborative relationship between Agents and not established the Agent from logic level.For ATC operational behavior modeling problems in air traffic control simulation, this paper analyzed the behavior characteristic of ATC operation, and adopted the most popular structure of Agent-BDI structure and established the conflict detection and resolution rule library of ATC Agent based on decision tree. Finally, this paper designed the deliberation ATC Agent. Based on Jadex platform, the model of ATC Agent was constructed. Then this paper have assembled the ATC Agent model to communicate and coordinate with other two kinds of Agent models, namely, aircraft Agent and ATC Automation Systems Agent established on JADE platform. The simulation scenario was constructed through simulation system and the BDI reasoning process of ATC Agent was verified and the daily command behavior of ATC was realized. Simultaneously, considering some circumstances, the prior knowledge in an Agent is no perfect, the Agent needs to supplement the knowledge base by learning. Therefore, this paper proposed the learning behavior of ATC Agent. And realized the learning behavior of ATC Agent by using the Q-learning algorithm. The results show that: the ATC Agent model can execute BDI reasoning process and learning behavior smoothly, and can detect and resolute the conflict between aircraft Agents.The results also proved the effectiveness and rationality of the BDI model and learning behavior which are constructed by this paper.