Development and Validation of Bearing Diagnostic and Prognostic Tools using HUMS Condition Indicators

Health and usage monitoring systems (HUMS) are currently used in military and civil service helicopters for health monitoring of flight critical components. Typically, vibration data recorded during a flight is processed to generate condition indicators (CIs). CIs from healthy components are normally used to set thresholds such that there is a small probability of the CIs of nominal components exceeding the thresholds. If a CI exceeds the threshold, the component is declared bad. The limitation of these CI thresholds is that they don't quantitatively correlate to the health condition of the components and therefore cannot be used for accurate diagnosis and prognosis in the implementation of condition-based maintenance. In this paper, we present an experience in developing bearing diagnostic and prognostic tools using HUMS condition indicators. Data mining models are investigated to correlate the CIs with the physical damage of the bearings and determine the minimum set of CIs with the maximum correlation coefficient. Further, data mining models for bearing prognostics are investigated. These data mining models are validated using real rolling element bearing test data with intermediate inspection. The significance of the work presented in this paper is that it could be used not only be used to set CI thresholds in HUMS for reliable diagnostics, but also potentially, to enhance the prognostic capability of HUMS.