Intelligent Monitoring of Complex Discrete-Event Systems

Abstract

A complex active system is a discrete-event system exhibiting both horizontal and vertical interaction between components. By inspiration of biological systems in nature, which are organized in a hierarchy of subsystems, the horizontal interaction between components at a given hierarchical level gives rise to an emergent behavior at a superior level, which is unpredictable from a knowledge of the behavior of the interacting components only. Since real critical-systems, such as power networks and nuclear plants, can be conveniently modeled as complex active systems, monitoring and diagnosis of complex active systems is of paramount importance to the safety of society. This is why an intelligent diagnosis framework for complex active systems is presented in this paper. Intelligence means that diagnosis does not require the naive reconstruction of the system behavior as a whole, which would be exponential with the number of components. Experiments show the effectiveness of the diagnosis technique.