The related concept of intelligent unmanned swarm operation was introduced. In order to reflect the autonomous capability and adaptive capability of intelligent unmanned simulation entity, the OODA-L (observe, orient, decide, act and learning) pattern was proposed to make the learning process explicit, and was further extended to Co-OODA-L pattern. In the general description of the intelligent unmanned simulation entity, Markov decision process was adopted to carry out the mathematical abstraction, and a three-domain hierarchical structure of intelligent unmanned Agent was proposed. In order to embody the characteristics of autonomous cooperation and distribution of intelligent unmanned swarm operation, an architecture for cooperative operation modeling of swarm was proposed, which uses an artificial neural network to integrate a variable number of intelligent unmanned Agents into a homogeneous or heterogeneous swarm.