基于灰色理论的复杂系统多故障模糊诊断

摘要

航天、核电等复杂系统源发故障概率通常难于获取,由此导致基于最大后验概率准则的诊断方法失效.针对上述问题,从测试可靠性先验概率入手,提出了一种基于相关矩阵和灰色系统理论的故障诊断方法,通过对系统建立故障-测试相关矩阵,生成有排序的测试报警概率矩阵,并利用灰色关联度衡量测试结果向量与故障特征向量的接近度,实现了对多发故障的模糊诊断.实验结果表明,该方法在诊断指标权重调整、诊断精细度、重点关注故障检出等方面优势明显,诊断结论正确率满足实用需求.%The failure probability is usually hard to collect for complicated systems in aerospace or nuclear power,which makes fault diagnosis based on Bayesian model inapplicable.By surveying this problem in test space,a novel multi-fault fuzzy diagnosis method based on dependency matrix and grey theory was proposed.A fault-testing dependency matrix was described first,and the fault alarm probability matrix with interest priority was developed.By calculating grey correlation degrees between test vectors and fault feature vectors,suspect faults can be located.Experimental results show that the proposed method can support diagnosis parameters' adjustment and report interested faults first,and the diagnosis conclusion has suitable accuracy,which can meet the practical demand.