Situation forecast is a key link in network situation awareness, because this can provide necessary decision support for network managers. In order to realize the big data model of network management, a forecast method based on association rules is proposed aimed at the problem that the current forecast methods fail to take advantages of big data. The method takes the characteristics of big data into full account combined with requirement of situation forecast, and the basic idea and processes of the method are given. The experiment results show that the proposed method reduces time complexity greatly through finding relevance instead of nonlinear registration, and improves accuracy of forecast compared with the traditional methods.