结构残差在基于SDG故障分离中的应用

符号有向图（SDG）是大规模复杂系统中故障传播关系的描述模型,但它仅能通过在相容通路上的定性推理进行故障分离,因此有着明显的局限性。而结构残差是定量故障分离的成熟方法,将其引入基于SDG的故障分离中,用以提高系统故障分离的能力,弥补SDG方法的不足。具体的做法是将SDG转化为结构残差的描述形式,再利用结构残差的基本思想,分别就变量是否完全可检测的情况给出了具体的故障分离方法。这样做可以提高系统故障分离的鲁棒性和准确性。应用实例表明,该方法可通过残差计算,准确判断故障源,实现故障分离。

Application of Structural Residuals in SDG-based Fault Isolation

Signed directed graph(SDG)is a model to describe the propagation relationships among faults in large-scale complex systems,but it can just make qualitative inference along the consistent paths to isolate faults.However,structural residual is a mature method to isolate faults quantitatively,so it is introduced into the SDG-based fault isolation to improve the fault isolation ability of the systems and overcome the disadvantages of the SDG method.The SDG is transformed into the form of structural residuals,and the fault isolation methods are proposed using the basic idea of structural residual in the two cases according to whether all variables are detectable.This method improves the robustness and correctness of fault isolation.The application example shows that the method can judge the orignal fault correctly to realize fault isolation by residual caculation.