Modelagem da degradação em turbinas de aeronave sob condições reais de voo

Abstract

The paper performs a modeling of the degradation curves in aircraft turbines under real flight conditions and also a comparison between machine learning techniques based on decision trees. The algorithms used are: Decision Trees (DT), Random Forest (RF) and Gradient Boosting (GB). Coefficient of determination, mean square error and root mean square error are employed as performance evaluation methods. The presented results show the best performance of RF and GB in estimating the values. The coefficients of determination of the algorithms reached average values higher than 0.98, thus showing the efficiency of the proposed models to be used in this application.