关键词: 网络安全风险, t-支持向量回归机（SVR）评估模型, 遗传算法, 参数组合寻优

Abstract: In order to improve the accuracy and real-time of network security risk assessment, this paper proposes a model about network security risk assessment based on the Support Vector machine for Regression optimized by t time-delay parameter. It combines and optimizes the key parameters of t-SVR, making use of the global search performance of GA. The simulation result of network security risk data-set indicates that the assessment model of t-SVR evaluation based on GA parameter optimization has solved?the?shortage?of SVR, and the risk assessment is made more accurate, the time more less and the performance more stable.

Key words: network security risk, t-Support Vector machine for Regression（SVR） assessment model, Genetic Algorithm（GA）, optimal parameters combination