Diagnostic enhancements for air vehicle HUMS to increase prognostic system effectiveness

A major objective of health and usage monitoring systems (HUMS) is to transition from time based part replacement to performing maintenance actions based on evidence of need. While existing HUMS capability has demonstrated progress, the ability to diagnose component faults in their early stages is limited. This is due in part to sensitivity to signal noise, variations in environmental and operating conditions, and underutilization of prognostic techniques. Using the representative example of the fan support bearing in the oil cooler of the UH-60 helicopter, this paper discusses key areas to improve fault detection methods for health monitoring of a damaged helicopter transmission component. These include: (1) sensing and data processing tools, (2) selection and extraction of optimum condition indicators/features, (3) fusion of data at the sensor and feature levels, and (4) incipient fault detection using a Bayesian estimation framework. Results illustrating the effectiveness of these techniques are presented for fielded UH-60 bearing vibration data and laboratory test results.