CAJ | 학술논문

当系统某次评价不达标，选取怎样的方案使系统整改后达标便是评价后续决策问题。首先类比事故树基本事件重要度，建立能综合反映指标重要度的模型及改善代价的模型；之后对最大重要度所对应指标得分进行固定步长、渐升步长以及渐降步长等3种形式的增值，再求新的系统评价得分，直到系统评价满足阈值，并确定3种迭代模型的取舍策略；最后分析迭代过程及结果，建立指标改变先后度模型，确定指标改变的轻重缓急次序。文中以模糊综合评价方法为对象分析，实例验证切实可行，并可推广应用于灰色评价、可拓学评价以及集对分析等常规评价方法。
당계통모차평개불체표，선취즘양적방안사계통정개후체표편시평개후속결책문제。수선류비사고수기본사건중요도，건립능종합반영지표중요도적모형급개선대개적모형；지후대최대중요도소대응지표득분진행고정보장、점승보장이급점강보장등3충형식적증치，재구신적계통평개득분，직도계통평개만족역치，병학정3충질대모형적취사책략；최후분석질대과정급결과，건립지표개변선후도모형，학정지표개변적경중완급차서。문중이모호종합평개방법위대상분석，실례험증절실가행，병가추엄응용우회색평개、가탁학평개이급집대분석등상규평개방법。
If any evaluation does not reach the standard,which plan should be selected to amend the system is the subsequent decision after evaluation.This article intends to study the subsequent decision for fuzzy com-prehensive evaluation from the perspective of importance degree for indexes.Firstly,by comparing importance degrees of cases based on fault tree analysis,the model for importance degree of indexes representing index structure space,ratio space,modification space,and easiness for modification characteristics is built,mean-while,models of score price,score unit price,and improvement costs are built.Then the importance degrees for indexes are calculated,and the score for the most important index is raised and the new evaluation score for the system is calculated.If the score does not satisfy the threshold,importance degrees for indexes are recalcu-lated and previous steps are repeated,and iterative computation is performed till the system meets standard re-quirements.Three kinds of iterative models,i.e.fixed step,gradually ascending step and gradually descending step,are proposed,and trade-offs strategies for the three kinds of iterative models calculation results are given. Finally,the iteration process and results are analyzed;and the index modification priority model is established. When the budget is tight or the system is difficult to be operated,the model could be used to decide the priority for modifying indexes.The method is feasible by practical examples,and can be extended to the conventional e-valuation methods such as the gray evaluation,extenics assessment and set pair analysis.