Fault diagnosis of a chemical process using identification techniques

The paper presents the application results concerning the fault diagnosis of a chemical process using dynamic system identification and model-based residual generation techniques. The considered approach consists of identifying different families of models for the monitored system. Then, dynamic output observers or Kalman filters are used as residual generators. The proposed fault diagnosis and identification scheme has been tested on a real chemical process in the presence of both sensor, actuator, component faults and disturbance.