Transparent fault diagnosis for magnetic control circuit Breakers: A ViT-Based unsupervised approach

IF 7.5 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Expert Systems with Applications Pub Date : 2025-10-15 DOI:10.1016/j.eswa.2025.130026

Kunquan Chen , Haoqing Wang , Fengchao Wang , Haiming Gao , Yiran Xia , Shude Zhao , Yakui Liu

{"title":"Transparent fault diagnosis for magnetic control circuit Breakers: A ViT-Based unsupervised approach","authors":"Kunquan Chen , Haoqing Wang , Fengchao Wang , Haiming Gao , Yiran Xia , Shude Zhao , Yakui Liu","doi":"10.1016/j.eswa.2025.130026","DOIUrl":null,"url":null,"abstract":"<div><div>The traditional fault diagnosis method of circuit breakers is always hindered by the scarcity of labeled fault data and the lack of transparency in model decision-making, thereby compromising its practical applicability. To address these challenges, this study proposes an unsupervised fault diagnosis framework for magnetic control circuit breakers, integrating a Vision Transformer (ViT) autoencoder, HDBSCAN clustering, and explainable AI (XAI) techniques. Acoustic and vibration signals are fused into mixed-feature Mel-spectrograms, enabling the ViT autoencoder to detect anomalies through reconstruction errors by learning normal-state distributions. HDBSCAN clusters latent features and attention scores to generate pseudo-labels, which are mapped to fault types using attention attribution and Integrated Gradients heatmaps, guided by expert knowledge. A classifier achieves 100% accuracy in fault detection and diagnosis. Experimental validation demonstrates the framework’s robustness and transparency, providing an effective solution for intelligent fault diagnosis in industrial settings.</div><div>© 2017 Elsevier Inc. All rights reserved.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"299 ","pages":"Article 130026"},"PeriodicalIF":7.5000,"publicationDate":"2025-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Expert Systems with Applications","FirstCategoryId":"94","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0957417425036425","RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE","Score":null,"Total":0}

引用次数: 0

Abstract

The traditional fault diagnosis method of circuit breakers is always hindered by the scarcity of labeled fault data and the lack of transparency in model decision-making, thereby compromising its practical applicability. To address these challenges, this study proposes an unsupervised fault diagnosis framework for magnetic control circuit breakers, integrating a Vision Transformer (ViT) autoencoder, HDBSCAN clustering, and explainable AI (XAI) techniques. Acoustic and vibration signals are fused into mixed-feature Mel-spectrograms, enabling the ViT autoencoder to detect anomalies through reconstruction errors by learning normal-state distributions. HDBSCAN clusters latent features and attention scores to generate pseudo-labels, which are mapped to fault types using attention attribution and Integrated Gradients heatmaps, guided by expert knowledge. A classifier achieves 100% accuracy in fault detection and diagnosis. Experimental validation demonstrates the framework’s robustness and transparency, providing an effective solution for intelligent fault diagnosis in industrial settings.

查看原文本刊更多论文

磁控断路器的透明故障诊断：一种基于vit的无监督方法

传统的断路器故障诊断方法由于故障标记数据的稀缺性和模型决策的不透明性，影响了其实际适用性。为了解决这些挑战，本研究提出了一种磁控制断路器的无监督故障诊断框架，该框架集成了视觉变压器（ViT）自动编码器、HDBSCAN聚类和可解释人工智能（XAI）技术。声学和振动信号被融合到混合特征mel谱图中，使ViT自动编码器能够通过学习正态分布的重建误差来检测异常。HDBSCAN对潜在特征和注意力得分进行聚类生成伪标签，在专家知识的指导下，利用注意力归因和集成梯度热图将伪标签映射到故障类型。该分类器在故障检测和诊断方面达到100%的准确率。实验验证了该框架的鲁棒性和透明性，为工业环境下的智能故障诊断提供了有效的解决方案。©2017 Elsevier Inc.版权所有。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Expert Systems with Applications 工程技术-工程：电子与电气

CiteScore

13.80

自引率

10.60%

发文量

2045

审稿时长

8.7 months

期刊介绍： Expert Systems With Applications is an international journal dedicated to the exchange of information on expert and intelligent systems used globally in industry, government, and universities. The journal emphasizes original papers covering the design, development, testing, implementation, and management of these systems, offering practical guidelines. It spans various sectors such as finance, engineering, marketing, law, project management, information management, medicine, and more. The journal also welcomes papers on multi-agent systems, knowledge management, neural networks, knowledge discovery, data mining, and other related areas, excluding applications to military/defense systems.