Predição de Desempenho Termodinâmico em Projeto Conceitual de Motores Aeronáuticos (Off-Design) com Inteligência Artificial

Abstract

The technological race for competitiveness and efficient, economic, and environmental soluti ons has been expressly transformed by the implementation of artificial intelligence (AI). In the aerospace sector, the topic is widely discussed in academic and industrial research, planning, ethics, cybersecurity, airworthiness, and certification. In this context, AI is identified as an op portunity to explore its usefulness in conceptual phase projects, specifically in turbofan engines. The general objective of this work is to develop an artificial intelligence meta-model capable of predicting the thermodynamic performance of turbofan engines under off-design conditions, considering critical variables such as thrust, specific fuel consumption, efficiencies, nitrogen oxides (NOx) emissions, turbine inlet temperature, among others. The methodology involved three stages: a thermodynamic on-design project inspired by the GE90 engine and historical success trends; followed by off-design simulations in GasTurb 15 (76 scenarios); and finally, the development of a prediction meta-model using the Gradient Boosting Regressor algorithm in Python, with Mach, altitude, and flight phase as input variables. The results showed high predictive accuracy, surpassing Random Forest and demonstrating good generalization, al though with limitations in some flight phases, indicating the need to expand the database. It is concluded that the approach proved promising as decision support in aerospace propulsion, especially regarding the expansion of analyses in the conceptual phase concomitant with design time reduction, with prospects for future applications in other conceptual design scenarios, cer tification, and sustainability.