Swarmed Neuro-Artificial Features from Vibration Data for Fault Detection and Isolation

This paper presents a neural network-based approach to create artificial features, defined as computer crafted, data-driven, and possibly without physical interpretation, expanding the number of tools that can be implemented to construct such features. In this work, we applied the approach to the problem of fault detection and isolation. A neural network artificial feature was extracted from vibration data of an accelerometer sensor to monitor, detect, and isolate a crack fault seeded in an intermediate gearbox of a helicopter´s main transmission. Classification accuracies for the artificial feature constructed from raw data were 100% for the training and independent validation sets.