Severe Anomaly Prognosis Using Multi-Class Imbalanced Deep Learning for Smart Factories

The project investigates multi-class imbalanced deep learning for severe anomaly prognosis in Industry 4.0 smart factories. We apply convolutional neural network (CNN) along with recurrent neural networks (RNN) of the long short-term memory (LSTM) model to determine if severe anomalies may appear in a smart factory machine. The so-called severe anomaly happens rarely, but once it happens, it usually has a significant impact on the system. Therefore, accurate methods are needed to predict its happening. We rely on a number of solutions to multi-class imbalanced deep learning and expect that the developed severe anomaly prognosis technology has good performance. We also plan to apply the developed technology to the wire electrical discharging machine (WEDM) smart factories to validate the performance of the developed technology.