Loading...

Prediction of Rolling Element Bearings Degradation Trend Using Limited Data

Tajdini, Jalal | 2024

0 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 57017 (08)
  4. University: Sharif University of Technology
  5. Department: Mechanical Engineering
  6. Advisor(s): Behzad, Mehdi; Arghand, Hesam Al-din
  7. Abstract:
  8. Condition monitoring of machinery is of significant economic importance to mitigate production losses resulting from downtimes. Unforeseen failure of roller element bearings is the most common issue observed in industrial units. However, detecting and tracking the progression of these failures through machine vibration monitoring and predicting the deterioration of these rotating components are viable solutions. Numerous studies have focused on using laboratory accelerated life test data for fault detection and remaining useful life prediction of these components. While online monitoring of all equipment in the industry may not be feasible, and conditions in the field differ from laboratory settings, only a limited number of measurements are available at various time intervals. In this study, the impact of increasing measurement intervals and data shortage on the performance of predictive models is investigated. A proposed model for predicting the degradation trend of roller bearings using limited data is presented and evaluated. Six baseline models are utilized to fit degradation curves and predict future trends through extrapolation. Combining these simple models into an ensemble model based on voting regression has resulted in a more powerful predictive tool. The performance of these models is then assessed with reduced laboratory and industrial training data. Ultimately, it was observed that in limited datasets (less than 13 measurements), linear models demonstrated better performance compared to non-linear models, and the proposed voting regressor model outperformed other models in predicting the degradation trend
  9. Keywords:
  10. Preventive Maintenance ; Rolling Element Bearings (REBs) ; Support Vector Machine (SVM) ; Ensemble Model ; Limited Data ; Degradation Trend Prediction

 Digital Object List

 Bookmark

No TOC