Fault diagnosis for supercharged boiler based on self-improving few-shot learning

Guolong Li, Yanjun Li, Jian Su, Haotong Wang, Shengdi Sun, Jiarui Zhao, Guolei Zhang and Jianxin Shi

Energy, 2025, vol. 316, issue C

Abstract: This study proposes a fault diagnosis method, SNN-MSU(H), to address the issue that the diagnosis model and support set cannot be updated and improved during the supercharged boiler fault diagnosis process. The proposed method sends samples that are not correctly diagnosed by the model to experts for re-diagnosis. These expert-re-diagnosed samples are then collected for model re-training. The results show that as the number of test samples increases or noise decreases, the model's accuracy increases and the need for expert intervention decreases. The SNN-MSU's accuracy is greater than 95 %, and the number of expert interventions in every 10 test samples does not exceed 1 when there are no samples in the initial support set. When the model is pre-trained with a few supercharged boiler fault samples, the SNN-MSU's accuracy exceeds 97 %, and that of SNN-MSUH exceeds 99 %. Finally, actual system tests demonstrate that SNN-MSU(H) can accurately detect new fault categories, such as valve disturbances, and request expert intervention in time. After expert intervention, the accuracy of the model diagnosis can be improved based on the knowledge provided by the experts.

Keywords: Supercharged boiler; Few-shot learning; Confidence threshold; Self-improving training; Model-expert cooperative (search for similar items in EconPapers)
Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
http://www.sciencedirect.com/science/article/pii/S0360544225001495
Full text for ScienceDirect subscribers only

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:eee:energy:v:316:y:2025:i:c:s0360544225001495

DOI: 10.1016/j.energy.2025.134507

Access Statistics for this article

Energy is currently edited by Henrik Lund and Mark J. Kaiser

More articles in Energy from Elsevier
Bibliographic data for series maintained by Catherine Liu ().