MACHINE LEARNING METHODS AND PREDICTIVE MODELING TO IDENTIFY FAILURES IN THE MILITARY AIRCRAFT

Modern aircraft are costly and require heavy investment. It is the same regardless of industries, such as commercial airlines and militaries. It is primarily about maintaining desired readiness by reducing ground time in the militaries, which is critical to maintaining air superiority and winning the war. There are two types of maintenance activities such as preventive and corrective maintenance. Preventive maintenance requires taking action before failures happen. Meanwhile, corrective maintenance reacts to failures, which takes time to buy parts and repair failed components. If we can predict aircraft failures accurately, we will be able to change corrective maintenance activities to preventive maintenance activities, which will reduce aircraft downtime and, thus, increase aircraft readiness or availability. This paper proposes multiple machine learning tools to minimize aircraft downtime to predict aircraft failures with the highest accuracy possible. This paper validates the usefulness of the proposed machine learning tools by experimenting with the actual data obtained from the maintenance record of 33 aircraft operated by the U.S. Air Forces.