Adaptive Prognostics for Rotary Machineries

Abstract

The success of condition-based-maintenance (CBM) of mechanical components, in terms of cost and reliability, hinges closely on the ability to accurately estimate the remaining life of the components at point. In the case of rotary machineries, special attention for remaining life estimation is warranted in view of the dynamic behaviors associated with repetitive contact motions. This paper discusses the rotary machinery prognostics using rolling element bearing – one of the foremost causes of breakdown in rotary machinery – as an example to illustrate the treatments of self-leaning prediction of remaining life. To date bearing remaining life prediction issue has not been fully addressed, as a result of the highly random nature of the bearing defect growth behavior in fatigue cycles. This paper addresses the lack of current bearing condition monitoring techniques by proposing a remaining life adaptation methodology based on the approaches of mechanistic modeling of vibration and self-training of parameter adaptation, thus the instantaneous rate of defect propagation can be apprehended, even when the initial health condition is unknown and the defect growth behavior is time-varying, to deliver a reliable prediction of the remaining life.