Neural Network Detection of Machine Faults
We introduce an approach to detect machine faults using vibration sensor data. Convolutional neural network supervised learning models, trained on Eastway collected and labelled data, demonstrate high levels of performance. The modelling approach is flexible to incorporate further data sources such as machine ambient data, however vibration data may be sufficient given the preliminary results presented here. Furthermore, the approach is adaptable to fault classification upon training with appropriately labelled data.