{"title":"Robust open-set partial discharge diagnosis based on hybrid supervised contrastive learning and SVM framework","authors":"H.P.D.Shiran Madhuranga, Wong Jee Keen Raymond, Hazlee Azil Illias, Nurulafiqah Nadzirah Binti Mansor","doi":"10.1016/j.asej.2025.103762","DOIUrl":null,"url":null,"abstract":"<div><div>Automated partial discharge (PD) diagnosis using machine learning models is useful for high-voltage equipment (HVE) insulation condition monitoring. However, without a mechanism to identify unknown PD classes (defined as new classes not present in the training data), models will misclassify unknown classes as one of the known classes. To address this, a novel hybrid open-set recognition (OSR) framework based on Supervised Contrastive Learning (SupCon) is proposed to address a previously unexplored direction in the PD diagnosis domain. The framework integrates discriminative representation learning with both unified and per-class rejection strategies using one-class classification, enabling effective separation of known and unknown PD classes. The main contribution is the synergistic integration of SupCon for constructing structured latent spaces, SVM for precise closed-set classification, and dual OCSVMs for adaptive unknown rejection, together forming a unified pipeline that achieves both fine-grained discrimination and robust unknown detection. To evaluate the effectiveness of the proposed framework, comprehensive experiments are conducted across 30 OSR tasks, covering 12 PD classes from three types of high-voltage equipment under varying openness levels. The proposed framework is benchmarked against five state-of-the-art approaches, including ArcFace, GAN-Flow, a convolutional neural network (CNN), Autoencoder, and Vision Transformer. Experimental results demonstrate that the proposed framework achieved the best performance, with a mean normalized accuracy of 97.66 % and a Youden’s index of 0.953, confirming its robustness, generalization capability, and potential to advance open-set PD diagnostic methodologies.</div></div>","PeriodicalId":48648,"journal":{"name":"Ain Shams Engineering Journal","volume":"16 12","pages":"Article 103762"},"PeriodicalIF":5.9000,"publicationDate":"2025-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Ain Shams Engineering Journal","FirstCategoryId":"5","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S2090447925005039","RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"ENGINEERING, MULTIDISCIPLINARY","Score":null,"Total":0}
引用次数: 0
Abstract
Automated partial discharge (PD) diagnosis using machine learning models is useful for high-voltage equipment (HVE) insulation condition monitoring. However, without a mechanism to identify unknown PD classes (defined as new classes not present in the training data), models will misclassify unknown classes as one of the known classes. To address this, a novel hybrid open-set recognition (OSR) framework based on Supervised Contrastive Learning (SupCon) is proposed to address a previously unexplored direction in the PD diagnosis domain. The framework integrates discriminative representation learning with both unified and per-class rejection strategies using one-class classification, enabling effective separation of known and unknown PD classes. The main contribution is the synergistic integration of SupCon for constructing structured latent spaces, SVM for precise closed-set classification, and dual OCSVMs for adaptive unknown rejection, together forming a unified pipeline that achieves both fine-grained discrimination and robust unknown detection. To evaluate the effectiveness of the proposed framework, comprehensive experiments are conducted across 30 OSR tasks, covering 12 PD classes from three types of high-voltage equipment under varying openness levels. The proposed framework is benchmarked against five state-of-the-art approaches, including ArcFace, GAN-Flow, a convolutional neural network (CNN), Autoencoder, and Vision Transformer. Experimental results demonstrate that the proposed framework achieved the best performance, with a mean normalized accuracy of 97.66 % and a Youden’s index of 0.953, confirming its robustness, generalization capability, and potential to advance open-set PD diagnostic methodologies.
期刊介绍:
in Shams Engineering Journal is an international journal devoted to publication of peer reviewed original high-quality research papers and review papers in both traditional topics and those of emerging science and technology. Areas of both theoretical and fundamental interest as well as those concerning industrial applications, emerging instrumental techniques and those which have some practical application to an aspect of human endeavor, such as the preservation of the environment, health, waste disposal are welcome. The overall focus is on original and rigorous scientific research results which have generic significance.
Ain Shams Engineering Journal focuses upon aspects of mechanical engineering, electrical engineering, civil engineering, chemical engineering, petroleum engineering, environmental engineering, architectural and urban planning engineering. Papers in which knowledge from other disciplines is integrated with engineering are especially welcome like nanotechnology, material sciences, and computational methods as well as applied basic sciences: engineering mathematics, physics and chemistry.