基于模糊贴近度的集群风险评价方法

Risk-Oriented Evaluation Based on Fuzzy Proximity for Clusters

通过对成熟集群周期性风险、结构性风险和网络性风险成因的分析,提取了21个风险特征指标;在此基础上,描述了成熟集群在不同风险水平的运行状态下的特征模式.根据集群系统的复杂性和不确定性的特点,应用多目标决策理论中理想解法的思想,首先确定集群风险特征指标体系对各个运行状态特征模式的非对称贴近度,再采用理想解法确定正负理想等级,根据各个运行状态与正负理想等级的贴近程度确定集群的风险水平.通过模拟算例证明了该方法评价集群风险的可行性.

Analyzing the causes of periodic, structural and network risks for mature clusters, a risk-oriented evaluation system including 21 indices is proposed. Then, the characteristic patterns of operation of a mature cluster at different risk levels are described. According to the complexity and uncertainty of cluster system, a TOPSIS （technique for order preference by similarity to ideal solution） comprehensive evaluation method in multi-object decision making theory is also proposed. The way the asymmetric proximity between the cluster＇s risk index system and actual operation conditions are calculated and, based on TOPSIS, the positive and negative ideal grades are determined. The proximities between each and every operation condition and ideal grades are thus compared to determine the risk level of mature cluster. An example is given to show the feasibility of the method.