Fault Severity Estimation in Rotating Mechanical Systems Using Feature Based Fusion and Self-Organizing Maps

The capability of Self-Organizing Maps (SOM) to visualize high-dimensional data is well known. The presented work concerns a SOM based diagnostic system architecture for the monitoring of fault evolution in bearings. Bearings form an essential part of rotating machinery and their failure is one of the most common causes of machine breakdowns. A SOM based approach has been used to map time series of feature data produced by acceleration sensors in order to capture the process dynamics. The fusion of specific features and the introduction of new features related to fault severity can enable the monitoring of fault evolution. The evolution of system states showing the bearing health trend has been shown to warn of impeding failure.