Towards Distribution Clustering-Based Deep LSTM Models for RUL Prediction
| Affiliation auteurs | !!!! Error affiliation !!!! |
| Titre | Towards Distribution Clustering-Based Deep LSTM Models for RUL Prediction |
| Type de publication | Conference Paper |
| Year of Publication | 2020 |
| Auteurs | Sayah M, Guebli D, Zerhouni N, Masry ZAl |
| Editor | Long J, Pu Z, Ding P |
| Conference Name | 2020 PROGNOSTICS AND SYSTEM HEALTH MANAGEMENT CONFERENCE (PHM-BESANCON 2020) |
| Publisher | Fento St; Le Cnam; Univ Paris Saclay; Alstop; IEEE France Sect; IEEE Reliabil Soc; UBFC; L2S; Geeps; Int Soc Measurement, Management & Maintenance; Chongqing Technol & Business Univ; Chinese Journal Aeronaut |
| Conference Location | 10662 LOS VAQUEROS CIRCLE, PO BOX 3014, LOS ALAMITOS, CA 90720-1264 USA |
| ISBN Number | 978-1-7281-5675-0 |
| Mots-clés | C-MAPSS, clustering, Distribution-Based Clustering, LSTM Network, RUL prediction |
| Résumé | This paper proposes a clustering based deep learning network to predict Remaining Useful Life or RUL for a system component. This RUL means length from current time to end of component useful life time. The objective of our approach is to perform in prior, a distribution based clustering of all collected sensors data and operational monitoring information about the system component. Then, deep LSTM is generated using the constructed clusters to predict component system RUL. Indeed, numerical features from data transactions are cleaned and then organized in clusters by regards to different level or standard deviation accuracy theta epsilon [0, 1]. In our experiments, results on NASA C-MAPSS (Commercial Modular AeroPropulsion System Simulation) training/testing data-sets shows the relevance of distribution based clustering use in the setting of RUL prediction Deep LSTM model. |
| DOI | 10.1109/PHM-Besancon49106.2020.00049 |