Fault Diagnosis and Remaining Useful Life Prediction in Manufacturing using Deep Learning Approaches

The fourth Industrial Revolution has transformed the manufacturing paradigm into smart manufacturing through the development of digital technologies such as Artificial Intelligence (AI), the Internet of Things (IoT), and so on. The rapid progress of Industry 4.0 and IoT results in the proliferation of sensing technologies, generating a sheer volume of industrial data. This allows real-time monitoring of the health state of manufacturing equipment on the factory floor. Meanwhile, an increasingly competitive and changing market emphasizes the significance of maintenance activities in the production. Predictive maintenance has substantial potential as an effective maintenance strategy for assisting enterprises in attaining a competitive advantage by reducing operational costs. Nevertheless, there are still some variations in maintenance management due to complexity of manufacturing environment. Recently, the tremendous advancement of digital applications has resulted in the emergence and development of the semiconductor industry, allowing it to establish itself as a deemed essential industry. Hence, a project is proposed to apply Deep Learning methods on a real-world industrial dataset for diagnosing faults and predicting the remaining useful life of an ion mill etching system used in semiconductor manufacturing.