Abstract
The timely and quantitative evaluation of the degradation is crucial for traction inverter systems in railway applications. The implementation in the industry is impeded by two major challenges including the varying operational profiles and the scalability for system-level applications. This paper proposes a deep recurrent autoencoder-based degradation evaluation method, to assess the degradation level of the traction power module online. The recurrent structure is embedded for processing multivariate time series condition monitoring data stream, in order to exploit the inherent time dependence to improve the accuracy and robustness. The autoencoder-based framework enables the scalability of the proposed method to system-level applications and can be applied under varying operating conditions. The method is experimentally demonstrated on an FPGA-based hardware platform.
Original language | English |
---|---|
Title of host publication | 2023 IEEE Energy Conversion Congress and Exposition, ECCE 2023 |
Number of pages | 7 |
Publisher | IEEE |
Publication date | 2023 |
Pages | 1542-1548 |
Article number | 10362793 |
ISBN (Print) | 979-8-3503-1645-2 |
ISBN (Electronic) | 979-8-3503-1644-5 |
DOIs | |
Publication status | Published - 2023 |
Event | 2023 IEEE Energy Conversion Congress and Exposition, ECCE 2023 - Nashville, United States Duration: 29 Oct 2023 → 2 Nov 2023 |
Conference
Conference | 2023 IEEE Energy Conversion Congress and Exposition, ECCE 2023 |
---|---|
Country/Territory | United States |
City | Nashville |
Period | 29/10/2023 → 02/11/2023 |
Sponsor | COMSOL, DELTA, et al., Hitachi, John Deere, Oak Ridge National Laboratory |
Series | IEEE Energy Conversion Congress and Exposition (ECCE) |
---|---|
ISSN | 2329-3748 |
Bibliographical note
Publisher Copyright:© 2023 IEEE.
Keywords
- autoencoder
- deep learning
- degradation
- inverter system
- Power module
- prognostics and health management