Similarity‐based modeling of vibration features for fault detection and identification

Abstract

Purpose

To provide an overview of the similarity‐based modeling (SBM) technology and review its application to condition monitoring of rotating equipment using features calculated from vibration sensor signals.

Design/methodology/approach

Concentrates on the practical capabilities and underlying technology of SBM. Examines the effectiveness of it as an approach to detect and diagnose faults in an electric motor‐driven shaft during variable speed operating conditions.

Findings

The SBM is a non‐parametric pattern recognition technology developed by SmartSignal that is applied generally to multivariate condition monitoring problems. A vibration sensor is monitored by first transforming the digitized time domain sensor signal into relevant features over time. These features are monitored continuously in real time to detect any discernable differences from normality. The deviations in turn, produce fault signatures in time‐feature space that aid in fault diagnosis.

Originality/value

Gives information on an approach that employs a multivariate similarity‐based modeling technique to characterize the expected behavior of vibration signal features which enables the detection of incipient faults in rotating machinery.

Citation

Wegerich, S. (2005), "Similarity‐based modeling of vibration features for fault detection and identification", Sensor Review, Vol. 25 No. 2, pp. 114-122. https://doi.org/10.1108/02602280510585691