Progress in Root Cause and Fault Propagation Analysis of Large-Scale Industrial Processes

Abstract

In large-scale industrial processes, a fault can easily propagate between process units due to the interconnections of material and information flows. Thus the problem of fault detection and isolation for these processes is more concerned about the root cause and fault propagation before applying quantitative methods in local models. Process topology and causality, as the key features of the process description, need to be captured from process knowledge and process data. The modelling methods from these two aspects are overviewed in this paper. From process knowledge, structural equation modelling, various causal graphs, rule-based models, and ontological models are summarized. From process data, cross-correlation analysis, Granger causality and its extensions, frequency domain methods, information-theoretical methods, and Bayesian nets are introduced. Based on these models, inference methods are discussed to find root causes and fault propagation paths under abnormal situations. Some future work is proposed in the end.

Towards the development of a tool for visualising plantwide dependencies

“…Methods for capturing connectivity and causality follow two main approaches: qualitative and quantitative. As detailed in [5], topology can be constructed and described based on a priori knowledge (qualitative methods) including first-principles and mathematical models. These methods include the use of structural equations, causal graphs and rulebased models.…”

“…Previous investigations in this topic [4,5] have pointed out that there is a need for tools that can integrate various quantitative and qualitative methods in fault propagation analysis. Yet, to date, there has been little work on creating tools that are able to efficiently integrate and represent plant topology information.…”

Determining the propagation path of a disturbance in multi-rate process and electromechanical systems

“…A common goal in process monitoring and diagnosis is to distinguish the root cause from the propagated disturbances. To that end, a recent topic in the literature is the extraction of the propagation path of the disturbance from measurement data (Yang and Xiao, 2012). The propagation path is a qualitative model of the affected system, and shows the affected measurements in a directed succession according to the order of propagation of the disturbance.…”

Root cause analysis of process fault conditions on an industrial concentrator circuit by use of causality maps and extreme learning machines

“…However, several studies have reported the potential and benefits derived from the use of hybrid systems for fault detection and diagnosis, combining fundamental process knowledge with data-base models (Van Lith et al, 2003;Venkatasubramanian et al, 2003c;Nandi et al, 2004;Van Sprang et al, 2005;Thornhill and Horch, 2007;Yang and Xiao, 2012), especially when considering large-scale industrial situations. From a model-based perspective, it has been shown that the use of hierarchical system decomposition, as a qualitative model-based reasoning tool, can provide effective modularity for structuring large-scale diagnostic knowledge bases (Prasad et al, 1998).…”