Data Analytics Tools for the Automated Logistics Environment (ALE)

The ALE Analytics Application (ALE AA) is a tool enabling high-end analytics using thousands of hours of in-flight performance data to predict and prevent failures before they happen. We are coordinating with KBRwyle to apply machine learning techniques using the Hawkeye Flight E-2D Integrated Analytics tool. Mosaic will capture requirements through stakeholder engagement, prioritize use cases, define data format, design an application, and demonstrate value. Our approach does not require a deep understanding or documentation for aircraft subsystems. Modern data science techniques allow us to learn directly from the data and discover new relationships without having to incorporate endless amount subject matter expertise. We have already begun evaluating candidate techniques and algorithms. We will begin with exploratory analysis. This alone will support improved decision-making by synthesizing cross-fleet information on system performance and identifying and characterizing high-frequency or high-impact failure patterns. As we progress with the data, we anticipate using anomaly detection approaches to help find unexpected failures and respective precursors. Other appropriate modeling approaches will likely include predictive machine learning models that can flag failure conditions before they occur, Bayesian networks for modeling and understanding failure progressions, and deep learning for mining for complex relationships within and between systems.