The traditional network public opinion guidance method based on shadow ID posting has the disadvantages of low intelligence level and poor effect. The rapid development of artificial intelligence has laid a solid the-oretical and technical foundation for intelligent guidance of network public opinion. This paper breaks through the thinking stereotype of journalism and communication, and uses the theory of intelligent agent, system modeling and simulation technology to study intelligent guidance of network public opinion. A network public opinion intelligent guiding idea of "situational awareness-simulation deduction-opinion guidance" is put forward, and an intelligent guidance simulation deduction model and the functional module architecture are constructed with the simulation tool defined. On the basis of simulation deduction, network public opinion guidance is carried out, overthrowing the traditional shadow ID posting way and cracking the dilemma of difficulty in trying various guiding strategies directly in the real network environment. This study is of important theoretical significance and reference value for the enriching of related theory in network consensus warfare and the occupation of commanding heights of public opinion in this field.