系统可靠性决策规则发掘方法研究

为根据已有的经验提炼出对系统可靠性评判的决策准则,基于汪培庄的因素空间理论与笔者提出的空间事故树理论思想,构造了一套从决策经验中提取决策准则的方法.该方法适用于条件属性为域值的多因素影响系统.首先数据归一化构建基础信息决策表Ψ(T).计算x_i与x_j关于属性a_q的相似度S(x_i,x_j,a_q),形成相似性表Θ(U,C,S).根据相似性阈值f将Θ(U,C,S)化简得Θ′(U,C′,S).根据模糊度θ,将Θ′(U,C′,S)改造为模糊二项相似表Ω(Θ′,θ),从而构建区分矩阵Γ(Ω).对U进行决策组划分,形成决策项,组成决策归纳表L(T).利用L(T)和Ψ(T)得到了切合实际使用的决策语义表L(T).从方法处理实际问题的结果看,在只利用条件属性集合与对象集合的条件下,即可判断经验决策的合理性.经过一系列的数据处理最终得到的决策语义表L(T),它可以判断系统工作状态属于哪种系统可靠性的决策,并确定了这种决策的置信度.

The method research on decision criterion discovery of system reliability

In order to discovery decision criterion of system reliability evaluation on the basis of the field experience, with factor space theory and space fault tree from Wang Peizhuang and the authors separately, a set of methods are constructed, which discovered the decision criterion of system reliability on the basis of the field experience. This method is suitable for the system of multi-factor influence and the condition factors are the domain value. Basal information decision table Ψ(T) is constructed by data normalization. The similarities S(x_i,x_j,a_q) are calculated between x_i and x_j on factor a_q to get similarity table Θ(U,C,S). According to the similarity threshold f, Θ(U,C,S) was simplified to get Θ '(U,C',S). According to the fuzzy degree θ, Θ'(U,C',S) is transformed to get the similar fuzzy binomial table Ω(Θ',θ) and then to build distinguish matrix Γ (Ω). Decision items are formed through the division of decisions group U, the items make up the decision table l(T). l(T) and Ψ(T) are used to get the practical use of decision semantic table ℘(T). From the results of the method to deal with the practical problems, it can be used to judge the rationality of the empirical decision-making under the condition of the conditional attribute set and object set. After a series of data processing, the decision semantic table ℘(T) is obtained. ℘(T) can judge which kind of system reliability the system working state belongs to, and determines the confidence level of decision-making.