Outcome: Researchers from the Center for Advanced Communications of Villanova University and Bucknell University working with the Vibration Specialty Corporation, Philadelphia, PA have developed advanced signal processing algorithms to detect bearing fault defect signatures from the complex vibration signals observed for condition monitoring of rotating machines.
Impact/Benefit: Advanced signal processing algorithms reveal the weak fault signatures buried under the strong interference signals. Such algorithms will pave the way for detecting faults in early stages as well as identifying the types of faults, thereby informing operators to take measures before costly failures in rotating machinery occurs.
Explanation: Vibration signal analysis is a practical and cost effective technique for fault detection in bearing assemblies. The industry standard envelope spectrum technique based on spectral analysis is unable to reveal the incipient fault signatures in the presence of strong bearing interference signal that dominates the vibration data. Researchers have developed a vibration signal measurement model composed of bearing interference signal, fault signal, and measurement noise, each of which has its own characteristics. Exploiting these characteristics, they have devised an adaptive interference cancellation algorithm to enhance the fault signatures. Furthermore, upon cancellation of interference, optimum techniques based on constrained minimizations are utilized to recover the fault pulsing events. The characteristics of the pulsing events indicate the type of the fault which is essential in selecting a cost-effective solution for the replacement of the faulty part.
Detection of fault pulsing events from vibration data measured from a faulty bearing: a) The vibration signal is dominated by the strong bearing interference signal, b) The fault signal is revealed after bearing interference cancellation, c) Fault pulsing events are clearly identified using optimum constrained minimization technique.
