Adaptive neuro-fuzzy inference system for performance health monitoring of industrial gas turbines

Tahan-Bouria, M. and Muhammad, M. and Abdul Karim, Z.A. (2016) Adaptive neuro-fuzzy inference system for performance health monitoring of industrial gas turbines. In: UNSPECIFIED.

Full text not available from this repository.
Official URL: https://www.scopus.com/inward/record.uri?eid=2-s2....

Abstract

Figuring whether or not a gas turbine is inclined to faults provides useful help for determining the required preventive action before failure happening. System identification is a discipline that learns the behavior of the healthy engine and employs it to predict the fault proneness. This study aims to discuss the performance of the Adaptive Neuro-Fuzzy Inference System (ANFIS) compared to the Artificial Neural Networks (ANNs) for the purpose of gas turbine performance identification. Toward this end, three system identification Bank of Networks (B-Ns), each corresponding to seven variables that are commonly measurable on most twin-shaft industrial gas turbine engines, are developed. The accuracy of the trained B-Ns are analyzed using the healthy performance data of an industrial 18.8 MW open-cycle offshore gas turbine. Making a comparison between the gained results from ANFIS and two various ANNs models revealed that ANFIS model is able to forecast various performance parameters with higher correlation coefficient and smaller MAPE values. © IEOM Society International. © IEOM Society International.

Item Type: Conference or Workshop Item (UNSPECIFIED)
Additional Information: cited By 1; Conference of 6th International Conference on Industrial Engineering and Operations Management in Kuala Lumpur, IEOM 2016 ; Conference Date: 8 March 2016 Through 10 March 2016; Conference Code:135628
Depositing User: Mr Ahmad Suhairi UTP
Date Deposited: 09 Nov 2023 16:19
Last Modified: 09 Nov 2023 16:19
URI: https://khub.utp.edu.my/scholars/id/eprint/7350

Actions (login required)

View Item
View Item

No Altmetrics are currently available for this publication.