Development of digital twin data-driven modelling for gas turbine operation behaviour
Digital twins have recently gained attention as digital solutions in "Energy 4.0" that will reshape the future of the power generation industry toward the digital era. It is supported by the rapid advancement of data connectivity and computational power to intensify the potential of digita...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Universiti Putra Malaysia Press
2025
|
| Online Access: | http://psasir.upm.edu.my/id/eprint/120974/ http://psasir.upm.edu.my/id/eprint/120974/1/120974.pdf |
| Summary: | Digital twins have recently gained attention as digital solutions in "Energy 4.0" that will reshape the future of the power generation industry toward the digital era. It is supported by the rapid advancement of data connectivity and computational power to intensify the potential of digital twin technology in addressing the energy trilemma. The energy trilemma has been identified as a global challenge to transform the power generation industry landscape to be more efficient and competitive. Digital twins have been identified as a key enabler to address the impacts of this global challenge on power plants due to several factors such as ageing, performance degradation, and high operating costs. This study will evaluate the concept of the digital twin approach by developing the gas turbine digital twin to provide future insights into operational performance and optimisation. The gas turbine digital twin model is developed through a cutting-edge data-driven approach, utilising an artificial neural network (ANN) to deliver superior performance in advanced monitoring applications. The digital twin model is constructed structurally in four steps: process identification, data collection, pre-processing, and developing the digital twin plant model. The gas turbine operating parameters are analysed for critical parameter verification to emulate the gas turbine operation behaviour environment. The best deep learning structure for data-driven methods is identified based on a lower Mean Squared Error (MSE) and an average error of less than 0.5% of the predicted value. The findings indicate that the digital twin data-driven modelling can be applied to future advanced monitoring of gas turbines in the power generation industry. |
|---|