SymCure: A Model-Based Approach for Fault Management with Causal Directed Graphs

摘要

SymCure is an object-oriented graphical model-based fault management methodology that integrates automated and interactive fault diagnosis, testing, recovery, and impact prediction. SymCure allows domain experts to define class level fault models over generic events that are represented in the form of causal directed graphs. At run time, SymCure combines these generic fault models with incoming events and a specific domain representation, which describes a particular system configuration and relations among specific components, to diagnose root causes for abnormal system behavior. This methodology can be used for fault management in domains as diverse as enterprise wide communications and manufacturing processes. This paper describes SymCure's architecture, its diagnostic knowledge representation, fault management procedures, and discusses its contributions.