Robust Nonlinear Fault Detection Applied to Chemical Processes

This paper proposes a robust fault detection system that can be applied to nonlinear chemical processes. A nonlinear state estimator which is able to handle both parameter estimation and parameters with uncertainties is designed. The detection process is created by examining changes in the controlled outputs with respect to set point, followed by probing variations in the parameters that are estimated. A steam generator system is used to validate this approach where a process fault is considered.