Monitoring Processes with Structurally Different Operating Regimes

Abstract This paper presents a concept for monitoring processes with structurally different and changing operating regimes. The monitoring task is to identify the current operating regime and estimate time varying parameters within each regime. A multiple model description of such systems is considered, and the on-line regime- and parameter identification is performed by a bank of Augmented Kalman filters. The estimated regime information is used to supervise the parameter updating in each filter. A simplified fluidized bed model with a bubbling- and a turbulent regime is simulated to demonstrate the concept.