ANFIS Based Fault Detection and Identification of Control Valve Failures in Three Tanks Interacting Level Process
S. Nagarajan1, K. Kalaivani2, P. Navaseelan3, T. Kalaiselvi4, S. Sobana5
1S.Nagarajan, Department of Electronics and Instrumentation Engineering, Easwari Engineering College, Chennai (Tamil Nadu), India.
2K.Kalaivani, Department of Electronics and Instrumentation Engineering, Easwari Engineering College, Chennai (Tamil Nadu), India.
3P.Navaseelan, Department of Electronics and Instrumentation Engineering, Easwari Engineering College, Chennai (Tamil Nadu), India.
4T.Kalaiselvi, Department of Electronics and Instrumentation Engineering, Easwari Engineering College, Chennai (Tamil Nadu), India.
5S.Sobana, Department of Electronics and Instrumentation Engineering, Easwari Engineering College, Chennai (Tamil Nadu), India.
Manuscript received on 28 September 2019 | Revised Manuscript received on 10 November 2019 | Manuscript Published on 22 November 2019 | PP: 1389-1397 | Volume-8 Issue-6S3 September 2019 | Retrieval Number: F12470986S319/19©BEIESP | DOI: 10.35940/ijeat.F1247.0986S319
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Fault detection and identification for control valve failures in a three tanks interacting level process (TTILP) and failure accommodation under such failure conditions is presented in this paper.. The proposed Fault Detection and Identification is done based on multiple model observer approach and the observers are developed using Adaptive Neuro-Fuzzy Inference System (ANFIS). Four models are developed with a dedicated observer to represent the three control valve failures ‘struck at minimum flow position’, ‘struck at maximum flow position’, and ‘struck at last maintained position’ and the controller failure respectively. Decision functions are developed from the outputs of these models to identify the failures. Fault Confirmation Logic uses two more multiple model observers to avoid false alarms. The three types of control valve failures are considered in this paper. The results show that the TTLIP is able to control perfectly even under any type of control valve failure.
Keywords: ANFIS, Dedicated Observer Scheme, Fault Detection and Identification, Control Valve Failures, Three Tanks Interacting Level Proces, Observers.
Scope of the Article: Signal Control System & Processing