ABSTRACT
In this chapter, an integrated design framework to utilize multi-sensor data fusion techniques is proposed for process fault monitoring. The multi-sensor data fusion technique is presented by frameworks of centralized and decentralized unscented Kalman filter architectures. A set of simulation studies has been conducted to demonstrate the performance of the proposed scheme on a quadruple tank system and an industrial boiler. It is established that the decentralized integrated framework retrieves more effectively the critical information about presence or absence of a fault from the dynamic model with a minimum time delay, and provides accurate unfolding-in-time of the finer details of the fault as compared to the centralized integrated framework, thus completing the overall picture of fault monitoring of the system under test. Experimental results on the quadruple tank system and industrial utility boiler, show that the proposed method is able to correctly identify various faults even when the dynamics of the systems are large.
