The skills of software programming and modeling based on Intelligent Agent are more important in software engineering field recent years. Founded on the research of architecture of traditional Agent that is applied in power enterprise, we provide a new Agent model by introducing some non-normal modal operators, and prove its feasibility as well as evolvement in BDI semantic, and it is applied to Intelligent Agent Architecture. The research of the collaboration of multi-agent is a highly active aspect in the domain of multi-agent. We bright forward the modified task decomposition algorithm by summarizing the existing task decomposition and allocation algorithms, and testify its advantage in time complexity by means of comparing with existing algorithms. And we propose the corresponding task allocation algorithm as well as the task's marginal allocation, which is directed to the priority of sub-tasks. It describes the task allocation algorithms of verifiability tasks and concurrency tasks, tests their advantages by experiments, and carries out their mathematical proof. At last all of this that we suggest is applied to power enterprise, and we provide the multi-agent system framework. |