| 英文摘要 | Intelligent vehicle is important part of ITS. The research on intelligent vehicle can help reduce the traffic pressure, and increase driving safety and comfortability. Intelligent vehicle integrates so much technology such as computation, information fusing, communication, AI and automatic control, etc. Advance control algorithm and intelligent decision algorithm can improve the driving performance. Currently, the network has played significant role in vehicle development. Many kinds of networks have been used in vehicles. Therefore, the research on advance control algorithms in the network environment can improve the vehicle performance. This thesis researches the agent-based control on intelligent vehicles. The thesis includes: Firstly, various simple task-specific control agents are designed based on fuzzy rules. After dividing the task that will be operated by the vehicular integral control algorithm into some specific subtasks,control agents which can conduct those subtasks are designed by fuzzy rules. Using this approach, CC fuzzy control agent, ACC fuzzy control agent, and lateral control agent are designed. Based on those agents, hosting mechanism is designed to manage those agents. Secondly, the learning algorithm is studied for remote supervisor. By implementing fuzzy agents based on neural networks, the learning algorithm is designed for adjusting the parameters of fuzzy agents. According to this approach, the control idea of Local Simple and Remote Complex (LSRC) can be implemented. Finally, based on Petri net, we develop the transfer model between local and remote of control agent in intelligent vehicle. After analyzing MAF standard proposed by OMG which is accepted in industrial field, we realize that the analysis of the transfer process must be based on the analysis of executing process beyond agents and agent systems. Because the agents and agent systems are developed on general object, we develop the Petri model of general object at first. Then, Based on the Petri model of general object, we develop the Petri models of agents and agent systems. At last, by integrated relative Petri sub-net, we develop the unified Petri net, and discuss the transfer process based on it. |
修改评论