Mixup domain adaptations for dynamic remaining useful life predictions

Muhammad Furqon, Mahardhika Pratama, Lin Liu, Habibullah Habibullah, Kutluyil Dogancay

Abstract: Highlights•We propose mix-up domain adaptation for time-series unsupervised domain adaptation.•MDAN is applied to dynamic remaining useful life predictions and fault diagnosis.•We propose a self-supervised learning method via a controlled reconstruction learning.