Rotating machine condition monitoring using neural networks

Artificial Neural Networks can provide a method for classifying the condition of rotating machinery if provided with input features which depend upon the faults the system is trying to detect. The cumulants of the magnitude of the vibrations provide a useful set of features for the detection of imbalance and rub faults. Pre-processing of the vibration signal can amplify relevant spectral features improving the classification success. Artificial neural networks provide one way of classifying the condition based upon many features each of which gives some indication of a possible fault.