Proposta de um modelo preditivo com aprendizado de máquina aplicado em motores de aeronaves

Abstract

The use of Machine Learning and Big Data, based on data collected from aeronautical engine sensors, makes it possible to carry out studies on the predictability of engine maintenance.Therefore, technological advances and the amount of data generated allow the construction of computational models to enable the prediction of engine maintenance, which can reduce operating costs, in addition to increasing flight safety. In this scenario, it seeks to apply Artificial Intelligence techniques (Decision Tree and Random Forest) to find the engine's failure cycle and contribute to the improvement of the database. Therefore, the objective of this research is to build a predictive model based on machine learning that can be applied in aircraft engines. So, this research seeks to apply data science to demonstrate results obtained with the analyzed data and the importance of information storage. As a methodology, a qualitative approach was chosen, using bibliographic and experimental procedures for the construction of the model. Prediction values of the engine failure cycle performed by the three predictive models analyzed were obtained and, when the predictive model performance accuracy was measured, the third obtained 73%, being the best result. Thus, it is concluded that the use of information technology applied to data can bring to the aviation market an increase in air safety, reduce operating costs, in addition to strengthening the use of data in a strategic way. It is recommended as a complement to the research, the application of this predictive model in maintenance of small and medium-sized aircraft in order to verify the performance of this model in practice.