Data‐Driven Monitoring and Early Warning of Valve Plate Loss in Compressor Valves
Chemical Engineering & Technology · 2026-11 · 49 (11)
DOI: 10.1002/ceat.70344 ↗Kurz gefasst
Fachartikel in der Zeitschrift Chemical Engineering & Technology, online erschienen am 04.11.2026. Verfasst von Feng Wang und 4 weiteren Autor:innen. Die Arbeit nennt 20 Literaturangaben. Gefördert von: National Natural Science Foundation of China.
Abstract (Originalsprache)
This article addresses weak signals of valve plate loss in reciprocating compressor valves and limitations of traditional methods. A data‐driven early‐warning approach combining multidimensional distance‐metric‐based dimensionality reduction and a transformer model is proposed. Multi‐sensor signals are used to construct a four‐dimensional metric to evaluate feature importance, reducing 74 features to 10. Results show temperature rise in the faulty valve and an inverse relation with the opposite valve and validate vibration energy transmission. The method improves transformer performance under small samples, alleviates overfitting, and achieves accurate fault identification and reliable early warning.
Förderung
- National Natural Science Foundation of China
Zitieren
APA
Wang, F., Zhang, K., Shao, H., Wang, T., & Zhao, Q. (2026). Data‐Driven Monitoring and Early Warning of Valve Plate Loss in Compressor Valves. Chemical Engineering & Technology, 49(11). https://doi.org/10.1002/ceat.70344
Vancouver
Wang F, Zhang K, Shao H, Wang T, Zhao Q. Data‐Driven Monitoring and Early Warning of Valve Plate Loss in Compressor Valves. Chemical Engineering & Technology. 2026;49(11). doi:10.1002/ceat.70344
BibTeX
@article{Wang2026,
author = {Wang, Feng and Zhang, Kaidi and Shao, Huajin and Wang, Tianyu and Zhao, Qingliang},
title = {Data‐Driven Monitoring and Early Warning of Valve Plate Loss in Compressor Valves},
journal = {Chemical Engineering & Technology},
year = {2026},
volume = {49},
number = {11},
doi = {10.1002/ceat.70344}
}
EUVIMED