CAJ | 학술논문

针对网络可生存性综合评估方法中指标权重难以确定的问题, 提出了基于支持向量数据描述（SVDD）的网络可生存性综合评估方法。该方法分析了SVDD的几何意义, 采用二进制粒子群（BPSO）算法对建立的评估特征指标集进行特征选择, 将所得的特征指标集视为整体来建立SVDD分类模型, 并以测试样本点与模型的相对距离为依据评估系统的可生存性, 避免了综合评估中指标权重确定的主观性。最后通过网络实例验证了评估模型的有效性。
침대망락가생존성종합평고방법중지표권중난이학정적문제, 제출료기우지지향량수거묘술（SVDD）적망락가생존성종합평고방법。해방법분석료SVDD적궤하의의, 채용이진제입자군（BPSO）산법대건립적평고특정지표집진행특정선택, 장소득적특정지표집시위정체래건립SVDD분류모형, 병이측시양본점여모형적상대거리위의거평고계통적가생존성, 피면료종합평고중지표권중학정적주관성。최후통과망락실례험증료평고모형적유효성。
Aimed at the problem of indexes weighting ensuring in comprehensive evaluation, this paper proposed a comprehensive evaluation for network survivability based on support vector data description (SVDD). It analyzed the meaning of SVDD firstly, and constructed the survivability feature index set from the angle of key attributes and characteristics. Then the index set was selected and optimized by using the binary particle swarm optimization (BPSO) algorithm. This method constructed the SVDD classify model by regarding the optimum feature subset as a integrity. And it introduced the relative distance in kernel space between diagnostic samples and distributed spheres as a basis to evaluate the system survivability. It avoided the subjectivity of indexes weighting ensuring in comprehensive evaluation by using this method. Finally, it provided a network example to verify the validity of the evaluation method.