基于特征选择验证的故障注入仿真可信度评估

针对故障注入仿真的数据复杂、细节特征多,导致单一指标可信度评估和专家目测评估方法难以得到准确的评估结果和评估专家按照AHP方法给出的是区间互补判断矩阵,导致可信度指标的权重难以确定的问题。提出了一种基于特征选择验证FSV(Feature Selection Validation)的多指标综合可信度评估模型,通过FSV方法计算可信度指标的定量或定性的可信度值;定义了区间数的模糊度和相离度,构建了多目标优化模型计算可信度指标的权重。实例证明了方法的可行性和有效性。

Credibility Assessment of Fault Injection Simulation Based on Feature Selective Validation

Aiming at the complexity and large number feature details of the data of fault injection simulation which lead to the difficulty for the single indicators or expert visual assessment methods to obtain the accurate assessment results, and the evaluation expert giving the interval complementary judgment matrix on the basis of AHP method which lead to the difficulty for the weights of credibility indicators to be determined, a multi-indicator comprehensive credibility assessment model based on Feature Selection Verification (FSV) is proposed. It can calculate the quantitative credibility values and the qualitative credibility values of the credibility indicators. In order to calculate the weights of credibility indicators, the concepts of interval ambiguity and interval deviation are defined, and a multi-objective optimization model is constructed. The results show the feasibility and effectiveness of the method.