A study on data augmentation optimization for data-centric health prognostics of industrial systems

Abstract

Factory of the Future and Industry 4.0 bring new challenges for manufacturing systems. At the same time, new maintenance strategies have become a major pillar to keep manufacturing systems performances and competitiveness. Such strategies allow to anticipate failure thanks to prognostics process. As such, promising prognostics approaches use data-driven machine learning techniques, however the recorded data set for learning is often “small” as failure occurrences are rare. Therefore, this study investigates data augmentation methods for improving prognostics by adding new samples generated from existing ones. This paper use a public prognostics data set to assess the performances of data transformation methods based on transformation hyper-parameters optimization using cross-validation.