For the purpose of exploring the evolution trend and analyzing the autonomous awareness and control problems, this paper proposes a cognitive awareness-control model for network security situation based on fusion. This model is characterized by the design of the cross-layer architecture and cognitive circle which can improve the interactive and cognitive ability between the different network layers. Based on the analysis of the model components and their functions, this paper uses the fusion algorithm to obtain the accurate decision on the security events made by heterogeneous multi-sensor. Combining with the reasoning of the relation between threat gene and threat level, a hierarchical quantification method is put forward, encompassing service layer, host layer and network layer. This approach has the advantage of overcoming the shortcoming of dealing with the complex memberships among network components and improving the expression ability against network threat. In addition, through establishing the bridge between dispersed computing and the continuous control, the close-up feedback structure is formed and the self-awareness and self-control problems are solved. The simulation experiments prove that the presented model and algorithms can fuse heterogeneous security data, dynamically perceive the evolution trend of network threat and possess the autonomous regulation and control ability. This study meets the research goal of cognitive awareness-control and it provides a new method of monitoring and administrating the networks.