Precise signed digraph modelling based on causal dependence identification

Abstract

The fault diagnosis approach based on signed digraph is promising, but signed digraph models built by existing methods often contain false causalities and make spurious diagnosis results. In this article, a signed digraph modelling method based on causal dependence identification is proposed. Many equations used to describe mechanism of process system can be used to analyse the cause–effect relation between state variables and build precise signed digraph models. The cause–effect relation hided in system equations is extracted through causal dependence identification of algebraic and differential equations. Signed digraph model is then constructed by merging the analysis results. The method of causal dependence identification and strongly connected components identification of process system is investigated in detail. Algorithm of causal dependence identification is summarized and results in a simple but effective signed digraph construction procedure. The validity of the proposed method was tested by the case study of signed digraph modelling for a series connected liquid storage system, and the efficiency of the algorithm was tested by another case study of a compressor unit system. The comparison result shows that the proposed method can extract precise causal dependence relation between state variables from system equations and build signed digraph model effectively with less resource consumption, which is very important for building signed digraph model of large scale process system.