Applying Adaptive Prognostics to Rolling Element Bearings

Rolling element bearing failure can cause problems for industries ranging from mild inconveniences such as simple replacement to catastrophic damage such as large production-line equipment failure. Rolling element bearing failure has plagued industries for many years. Bearings are currently monitored to determine whether or not there is a defect in the bearing, but the remaining lifetime of the bearing remains unknown. This research estimates the bearings remaining lifetime through digital signal processing in conjunction with a modified version of Pariss equationa fatigue-failure equation well known in rotating machinery prognostics. An energy quantity, coined the Power Spectrum Value (PSV), is the maximum amplitude of the frequencies within a relatively small band around the resonant frequency of the system. The current PSV is estimated and updated using a chronologically weighted least squares algorithm. It is this PSV which is implemented in the modified Paris equation to determine the remaining lifetime of the bearing. This research presents a non-intrusive method of determining the lifetime of the bearing so that the bearings utility is maximized and reactive maintenance procedures are minimized.M.S