Abstract:
The US Army has improved a method for accelerating the maturity of vibration-based mechanical diagnostics, by measuring the Frequency Response Functions (FRFs) between potential failure locations and sensor locations within epicyclic gearboxes, and by building Condition Indicators (CIs) using these FRFs. The previous FRF methodology has been expanded to include frequencies up to 100 kHz, using the piezo-exciters, aircraft-installed Health and Usage Monitoring Systems (HUMS), and custom data acquisition hardware described herein. Previous CI development methodology has been improved by filtering captured vibration data with the FRFs. Using a recent process for generating diagnostic algorithms using machine learning, these FRF-based CIs outperform conventional CIs, and meet Aeronautical Design Standard 79D diagnostic classification criteria for use on board aircraft.