Modified self-organising map for automated novelty detection applied to vibration signal monitoring

This paper proposes a novelty detection-based method for machine condition monitoring (MCM) using vibration signals and a new feature extraction method based on higher-order statistics of the power spectral density. This novel MCM method is based on Kohonen's self-organising map and adopts a multidimensional dissimilarity measure for dual class classification. The approach is designed to be highly modular and scale well for a multi-sensor condition monitoring environment. Experiments using real-world vibration data sets with upto eight sensors have shown high accuracy in classification and robustness across different condition monitoring applications.