Skip to Main content Skip to Navigation
Theses

Contributions to data-driven model-based identification using small-size datasets for estimating bearings remaining useful life

Abstract : Remaining useful life (RUL) estimation for bearings degradation monitoring is an important metric for decision making in condition based maintenance of rotating mechanics. RUL estimation involves generally two steps: degradation indicator extraction and model identification. Common vibration signal based features for bearings degradation monitoring are sensible on the last stage of the degradation process. In this thesis, we propose new bearing degradation monitoring indicators that are monotonic and incorporate historical degradation information. To overcome the drawback of a small size training datasets for model identification, we elaborated a mixture distribution analysis based fuzzy model identification method for RUL estimation. Furthermore, we proposed a method to tune the parameters of the fuzzy models for bearings RUL estimation when new knowledge becomes available. The aim is to improve the accuracy of the RUL estimation through a knowledge accumulation process.
Document type :
Theses
Complete list of metadata

Cited literature [240 references]  Display  Hide  Download

https://hal.univ-lorraine.fr/tel-02877609
Contributor : Thèses Ul Connect in order to contact the contributor
Submitted on : Monday, June 22, 2020 - 3:36:42 PM
Last modification on : Tuesday, June 23, 2020 - 3:18:41 AM

File

DDOC_T_2020_0019_HUANG.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : tel-02877609, version 1

Collections

Citation

Fei Huang. Contributions to data-driven model-based identification using small-size datasets for estimating bearings remaining useful life. Engineering Sciences [physics]. Université de Lorraine, 2020. English. ⟨NNT : 2020LORR0019⟩. ⟨tel-02877609⟩

Share

Metrics

Record views

156

Files downloads

362