Analysis of machine learning algorithms for predicting the service life of aircraft structures under conditions of uncertainty in material properties

This paper addresses the relevant engineering problem of determining material parameters for fatigue strength calculations and crack growth prediction in structural components of transport category aircraft. This is especially significant under conditions of material property uncertainty, which is characteristic of the post-market operation, modification, and conversion processes of aircraft requiring additional certification. The work presents a classification and comparative analysis of Artificial Intelligence (AI) algorithms suitable for solving this task. Their advantages and disadvantages are evaluated in terms of accuracy, data requirements, and computational complexity. It is substantiated that AI methods are capable of detecting complex dependencies, non-obvious to humans, among a large number of known material parameters. The application of AI will make it possible to obtain reliable input data for durability calculations without performing costly bench tests, thereby accelerating the analysis and certification of aircraft.