Energy, Exergy and Emission Performance Prediction of the Hydrogen-fueled Scimitar Engine With Machine Learning Methods

Abstract

Machine learning methods become popular in recent years for the analysis and optimization of energy systems. In this paper, energy, exergy and emission performance of the hydrogen-fueled Scimitar engine is predicted machine learning approaches. Models are constructed with neural network, nearest neighbors, decision tree, gradient boosted trees, random forest, Gaussian process and linear regression approaches to predict the impacts of hydrogen mass flow rate, air mass flow rate, combustion chamber wall heat flux, cruise speed and cruise altitude on the overall efficiency, exergy efficiency and NOx emission index of the Scimitar engine. The results show that the Gaussian process approach has the best predictive capability for overall and exergy efficiencies while the linear regression provided the best results for the NOx emission index

Keywords: Scimitar Engine, Energy, Exergy, Emission, Machine Learning