Refinements to Mechanical Health Monitoring Algorithms 2012-01-2096
This paper discusses recent improvements made by Honeywell's Condition-Based Maintenance (CBM) Center of Excellence (COE) to Mechanical Health Management (MHM) algorithms. The Honeywell approach fuses Condition Indicators (CIs) from vibration monitoring and oil debris monitoring. This paper focuses on using MHM algorithms for monitoring gas turbine engines. First an overview is given that explains the general MHM approach, and then specific examples of how the algorithms are being refined are presented. One of the improvements discussed involves how to detect a fault earlier in the fault progression, while continuing to avoid false alarms. The second improvement discussed is how to make end of life thresholds more robust: rather than relying solely on the cumulative mass of oil debris, the end of life indication is supplemented with indicators that consider the rate of debris generation. Another improvement discussed is an approach for allowing rapid modifications to the logic that fuses vibration and oil debris indicators. This involves a development process in which fuzzy logic is used initially to create a three dimensional surface that defines the relationship between the vibration/oil debris inputs and the fused output. Once that surface is created, minor changes can quickly be made to that surface by replacing the fuzzy logic with a simple lookup table. Finally, the paper discusses lessons learned in applying existing HUMS vibration indicators, primarily used for helicopter gearbox monitoring, to high speed turbine engine components.