The network security situation grade assessment(NSGA) pays little attention to experts' trust information, preference information and objective information such as multiple information, thus, the NSGA multi-dimensional method based on the group consensus is established. In the sparse network security situation social network generation problem, the uncertainty degree of trust information is defined, further an imperfect trust network construction model considering the trust information uncertainty degrees and internal differences of single path trust, and the non-compensation trust information of multi-path trusts is proposed, and the weight of experts are computed. The objective ranking based on grade eigenvalue is defined, the degrees of group consistency and group inconsistency of experts' preferences are defined, and NSGA multi-dimensional preference piecewise variable synthesis optimization models including expert trust information, preference information and actual information are further established. This method provides a theoretical basis for the construction and test of NSGA, helping to develop advanced NSGA systems and improving the ability of network security protection.