Predicting compressor mass flow rate using various machine learning approaches
International Journal of Turbo and Jet Engines, 2024 (SCI-Expanded, Scopus)
- Publication Type: Article / Article
- Publication Date: 2024
- Doi Number: 10.1515/tjj-2023-0105
- Journal Name: International Journal of Turbo and Jet Engines
- Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Applied Science & Technology Source, Compendex, INSPEC
- Keywords: compressor map, compressor performance parameter, deep learning, machine learning, mass flow rate
- Eskisehir Osmangazi University Affiliated: Yes
Abstract
A major focus of the present study is to construct high-fidelity models for predicting corrected mass flow rates based on the collected compressor map data. Both traditional machine learning research and modern deep learning techniques have been employed to obtain well-fitted regression models of compressor mass flow rate. As traditional machine learning methods, Multiple Linear Regression and Random Forest, are conducted on compressor maps for prediction of corrected mass flow rate. The time series-based deep learning models are able to capture the overall trend of a given input for specific map data. Therefore, a time series-based deep learning technique, namely Gated Recurrent Unit has been employed to improve regression results. Besides, the prediction capabilities of the models, results also show that the proposed models can be used for the development of dynamic aero-thermal mathematical models of gas turbine engines and mass flow rate models created for dynamic compressors in other disciplines.