Performance Degradation Prediction of Industrial Equipment using Machine Learning
Min-Seop So, Jong-Ho Shin · Korean Journal of Computational Design and Engineering · 2019
Unexpected facility failures of heavy industry derive large cost so that it is important to prevent it beforehand by estimating machine health status, predicting remaining useful life (RUL), and determining proper time schedule of facility for inspection and replacement. Accordingly, active researches have been proceeding recently, such as prognostics and health management (PHM), condition based maintenance (CBM) method for failure prediction, and so on. The PHM and CBM are technologies that collect operating and environment data related to the failure/lifetime of the equipment and predict failure in advance before failure. In accordance with this purpose, this study proposes a method to predict remaining useful life by collected operating and environment data. The methodology proposed in this study includes signal processing techniques to remove noise signal from sensor data, and clustering, support vector machine, and deep learning, which are used to calculate and predicting reduction rate according to usage environment.