Title: Machine learning approaches for modelling a single shaft gas turbine
Authors: Hamid Asgari; Emmanuel Ory
Addresses: VTT Technical Research Centre of Finland Ltd., Espoo, Finland ' VTT Technical Research Centre of Finland Ltd., Espoo, Finland
Abstract: In this study, machine learning-based models of a single shaft gas turbine (GT) are developed. For this purpose, recurrent neural networks (RNN) are employed to train the datasets of the GT variables in both Python programming environment (by using Pyrenn Toolbox) and MATLAB software. Thirteen significant variables of the GT are considered for the modelling processes. The resulting models are validated against the test datasets. The results demonstrate that the RNN models are capable of performance prediction of the system with a high reliability and accuracy. However, in this study, the overall results demonstrate that the RNN model set up in MATLAB has a better performance with a higher accuracy compared to the model developed in Python.
Keywords: gas turbine; machine learning; modelling; simulation; artificial intelligence; recurrent neural network; black-box model.
DOI: 10.1504/IJMIC.2021.121862
International Journal of Modelling, Identification and Control, 2021 Vol.37 No.3/4, pp.275 - 284
Received: 21 Oct 2020
Accepted: 02 Apr 2021
Published online: 07 Apr 2022 *