Novel approach to estimate remaining useful life in condition based maintenance
No Thumbnail Available
Date
2021
Journal Title
Journal ISSN
Volume Title
Publisher
University of Oum El Bouaghi
Abstract
Deep learning is an efficient tool for Remaining Useful Life (RUL) estimation, which is crucial for intelligent prognosis and Condition-Based Maintenance (CBM) strategies. To achieve this task, Bidirectional long short-term memories have been preferred for their ability to identify patterns of temporal sequences independently, and Convolutional Autoencoder is performed in extracting features. To benefit from the advantages of these two deep learning models, this paper proposes their hybridization. We investigate the best configuration by varying the values of the hyperparameters and evaluating their impact on the new model’s performance. Finally, it is compared with other similar models in order to study the effectiveness of the
approach.
Description
Keywords
Condition-based maintenance; Remaining Useful Life (RUL); Bidirectional Long-Short Term Memory neural network; Convolutional auto-encoder neural network; C-MAPSS dataset