This paper introduces the PETRI net analysis methods and evolution Internet public opinion events of various factors,analyzing and discussing the evolution regularity Internet public opinion, and public opinion as a form flow analyzes. It proposed the concept of public opinion flow control model. On this basis, it dynamic introduced the associated variable in the network cluster nodes and study the general rules of public opinion between cluster organization associated networkbased mobile. It proposed regulatory network cluster and other opinion between nodes with "public opinion associated with process model". Verification from study of journalism and psychology at phenomenon of learning from each other, then confirmed the reasonableness of model proposed and established. This model can be summed up by public opinion evolution guidance related system design and public opinion guidance and control, provide intelligent support for related industries.