:: Volume 8, Issue 1 (9-2021) ::
2021, 8(1): 133-154 Back to browse issues page
Simultaneous Incipient Faults Diagnosis in the Non-Isothermal Continuous Stirred-Tank Reactor
Hossein Safaeipour , Mehdi Forouzanfar * , Amin Ramezani
Assistant Professor, Department of Electrical Engineering, Ahvaz Branch, Islamic Azad University, Ahvaz, Iran. , m.forouzanfar@iauahvaz.ac.ir
Abstract:   (5729 Views)
In chemical processes, thermal reactors are described by nonlinear closed-loop dynamic models. Timely detection of simultaneous fouling phenomena in the heat transfer system is a concern of this art. In this work, a new incipient fault diagnosis approach is proposed for application in the closed-loop non-isothermal continuous stirred-tank reactor (CSTR) system subjected to simultaneous Gaussian and non-Gaussian noises. First, the state vector is estimated by applying the well-known particle filter estimator. Then, the primary residual signal is generated using the system measurements, and the fault vector estimation is obtained. After that, by an adaptive either fixed threshold design applied in the online monitoring devised with the proposed evaluation technique, while the fault detectability is improved, the false detection problem is restricted to the system permitted number. Bank on, preventive maintenance scheduling also incipient fault trend prediction have become possible using the Gauss-Newton identification method. Finally, in order to evaluate the proposed approach, the simultaneous fouling incipient fault diagnosis over the heat transfer unit built-in nonlinear closed-loop CSTR system is considered. Furthermore, the confusion matrix and associated evaluation indices are employed to assess the simulation results quantitatively.
Keywords: exponential moving average, fault diagnosis, incipient fault trend prediction, model-based, particle filter.
Full-Text [PDF 1899 kb]   (1955 Downloads)    
Type of Study: Applicable | Subject: Fault Detection and Isolation (FDI)
Received: 2021/09/5 | Accepted: 2021/11/1 | Published: 2022/01/19


XML   Persian Abstract   Print



Rights and permissions
Creative Commons License This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
Volume 8, Issue 1 (9-2021) Back to browse issues page