Effort estimation for aircraft maintenance using machine learning techniques

Abstract

The maintenance of aircrafts is one of the most demanding, and critical aspects of modern air travel. Airlines carry out their maintenance at aircraft Maintenance, Repair and Overhaul (MRO) centres and are charged according to the amount of man hours used to perform scheduled or preventive maintenance tasks. This work was motivated, and proposed, by a local MRO that showed interest in determining if machine learning techniques could be used for effort estimation. With the cooperation of the local MRO, this study investigated the application of several machine learning techniques, such as support vector machines and decision trees, in order to predict the amount of man hours required for a specific maintenance job. The MRO made available data from its European and non-European MRO centres. Moreover, a visual desktop application was built. This application will be used by the industry partner since it hides the complexities of machine learning algorithms from the user. It was designed to be used by the engineers who prepare man-hour estimates for customers. This study demonstrated to the satisfaction of the industrial partner that machine learning techniques can be used to improve effort estimation for MRO tasks. Consequently, all results obtained were confirmed to be relevant by the industrial partner.