H∞ Based Fault Detection Observer Design for Three-tank System with Parametric Uncertainty

Early fault detection is essential to prevent machine damage, performance degradation, and even health risks. This study dispenses the design of an observer-based residual generator for fault detection in a linear time-invariant uncertain system with application to the three-tank system benchmark. A robust observer-based fault detection filter based on H∞ optimization has been designed as linear matrix inequality constraints. In the benchmark system, the observer's performance is successfully assessed for a few fault categories, including sensor, actuator, and water leakage issues. The simulation results demonstrated the proposed observer's ability to identify the faults.