Abstract： A comprehensive overall network security situation awareness framework was proposed, fully considering mul-information sources and multilevel information fusion, which can dynamically produce the current network security situation from the three dimensions, accurately reflect the current network security situation, and easily find the abnormal component. Besides,  a “3σ rule” to discretize continuous random variable was proposed that can establish a Bayesian network suitable for dealing with uncertain information fusion, providing important guidance to theory and practice. Making full use of network instance data, the model and algorithm are verified and the results show that the method is correct.