Clustering intelligent sensor nodes for distributed fault detection & diagnosis

Abstract

Having studied distributed process monitoring using intelligent nodes in a cluster, we now extend the study to include using multiple clusters of nodes to monitor multiple process units interconnected to form a process plant. Our case study is that of a reaction process consisting of two CSTRs and auxiliary equipments such as tanks, heat exchangers etc. The study aims to explore the possibility of the new framework of distributed (and collaborative) process monitoring, in which fault detection and diagnosis is performed at the physical sensor and sensor cluster level. The resulting framework should be a better advisory tool for situation analysis and fault diagnosis by plant operator; while potential extension includes prognosis and the inclusion in control and fault recovery. © 2006 IEEE.