Prognostic Real-Time Operation Machine Learning for Engine Health and Usage Supervision

Navy Rotorcraft operate in harsh environments that pose unique wear on engine health. Saltwater from the ocean, sand from deserts, and other factors can cause buildup or erosion on the internal components of turboshaft engines. If not maintained, this can cause decreased power output and a multitude of other issues on the platform that could lead to the rotorcraft not being available for operations when needed.  The objective of this project is to utilize advancements in Artificial Intelligence and Machine Learning algorithms to monitor various sensor outputs and determine when rotorcraft engines need maintenance. Currently, the Navy follows traditional predictive maintenance schedules to maintain engine health which may result in unnecessary vehicle downtime or catastrophic failures if issues are missed.  CRG is developing the Prognostic Real-Time Operation Machine Learning for Engine Health and Usage Supervision (PROMETHEUS). PROMETHEUS will evaluate core engine health performance in real-time throughout all stages of flight and highlight and monitor degradation caused by corrosion and turbine and compressor blade erosion. PROMETHEUS’s ensemble machine learning (ML) model will be trained on real-world Navy datasets combined with simulation data from our research partner’s 2-Spool Gas Engine Dynamics Model. This will generate training data for both healthy and damaged engines across a wide variety of operating and environmental conditions. CRG’s expertise in machine learning for aerospace and our partner’s expertise in turbine engine modeling will form the basis for the PROMETHEUS solution.