Knowledge network is a typical social network, which is equipped with the feature of small-world. The paper adopts adaptive modeling method of Multi-Agent in complex adaptive system, applying Multi-Agent simulation platform Netlogo to construct the knowledge transfer simulation mode based on small-world net model. Using the average path length and clustering coefficient to stand for AC Frequency and Aggregation Degree among knowledge network nodes and studying the nodes' ability to release and absorb as well as the knowledge transfer effect by trust degree. Operating simulation model means improving the AC Frequency and Aggregation Degree of nodes, enhancing nodes' ability to transfer knowledge can ensure transfer frequency in organization reaching a high level and offer rules and guidance to construct net structure and behavior model suited with knowledge dissemination and transfer.