Predicting faults in technical systems using convolutional neural networks

Demonstrates how to predict the remaining useful life (RUL) of technical systems using deep convolutional neural networks (CNN). The advantage of the deep learning approach is that it eliminates the need to manually extract or select features for the model used to predict the RUL. In addition, no prior knowledge of machine condition prediction or signal processing is needed to develop a deep learning-based RUL prediction model. The method was tested in a MatLab demo program implementing this method for predicting the occurrence of faults in technical systems¹. The program used the Predictive Maintenance Toolbox and Deep Learning Toolbox of the MatLab environment.

Authors: Yu. А. Korablev, D. М. Loseva, М. Yu. Shestopalov

Direction: Informatics, Computer Technologies And Control

Keywords: faults, diagnostics of the state of a technical system, prediction of the occurrence of faults, residual service life of a technical device, convolutional neural networks, training of a neural network

---

View full article