I. Introduction

Fault detection and isolation (FDI) has become increasingly important in maintaining stable, reliable, and profitable operations in the presence of component malfunctions, drifting parameters, and other abnormal events. Process disturbances, measurement noise, model nonlinearities, and other sources of uncertainty make fault diagnosis a challenging task that is further complicated by the steadily increasing complexity of industrial systems. By now, many methods have been proposed to address these challenges, such as residual-and observer-based methods [1]–[4], set-based approaches [5], [6], and data-based methods [1]. The majority of these methods are passive, meaning that the inputs are not actively changed and that the fault status of the system is deduced only on the basis of measurements obtained during standard operation, compared with model predictions or historical data. However, faults may not be detectable or isolable at the current operating conditions, or when faults are obscured by the corrective action of the control system itself. Then it is necessary to inject a signal into the system to improve fault detectability and isolability, which is an approach known as active fault diagnosis [7]. Although active fault diagnosis can significantly improve fault isolation, the required excitations can have adverse effects on the process that must be minimized.