Fault Detection in Continuous Stirred Tank Reactor (CSTR) System Using Extended Luenberger Observer

This research proposes fault detection in a Continuous Stirred Tank Reactor (CSTR) system using an Extended Luenberger Observer (ELO). The ELO is chosen due to the non-linearity of the CSTR system. Accurate state estimation is critical for effective fault diagnosis; therefore, the performance of the ELO is initially tested using two indicators: robustness and sensitivity in estimating the level and concentration within the CSTR system. The sensitivity test yields promising results, with the ELO accurately estimating the actual system despite variations in input and initial conditions, and with a fast convergence time of 1 seconds. The robustness test also demonstrates positive outcomes, as the ELO continues to estimate the system accurately even in the presence of noise with standard deviation 2.5% of measurements. Furthermore, faults that can be related to sensor malfunctions or the disturbances in the CSTR process were successfully detected using the ELO. Performance analysis and fault detection in the CSTR system are presented through simulation. The contributions of this research include development of ELO for non-linear dynamics CSTR system and evidence of its effectiveness in detecting fault within the in CSTR system.