PREDICTION OF FAILURE CURVES IN AIRCRAFT COMPONENTS BASED ON REAL DATA

Efficient maintenance of aeronautical components is vital to ensure flight safety, reduce operational costs, and increase fleet reliability. In this context, failure analysis of such components becomes essential to improve predictive maintenance policies. This work proposes a study based on anonymized real-world failure history data from different types of components of a national airline. The objective is to estimate parameters of classical failure curves, such as Weibull, exponential, and log-normal distributions. After organizing and segmenting the data, statistical model fitting techniques are applied using Python libraries. For the accuracy of the developed tool, factors such as the quantity and quality of data proved to be critical. The results enable the prediction of failure probability, on a scale from 0 to 100% and over a 365-day period, for each type of component analyzed, supporting resource prioritization and the definition of more efficient interventions. The approach proves promising for the application of reliability methods and survival analysis in the context of aeronautical maintenance, reinforcing the role of operations research in data-driven decision-making.