FSR‐SSL: A fault sample rebalancing framework based on semi‐supervised learning for PV fault diagnosis
Abstract Photovoltaics face the threat of many potential faults in daily operation, which calls for accurate fault diagnosis to avoid huge economical losses. This paper investigates practical and troublesome scenarios, where the photovoltaics station has a large number of unlabeled samples and only...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2022-09-01
|
| Series: | IET Renewable Power Generation |
| Online Access: | https://doi.org/10.1049/rpg2.12458 |
| _version_ | 1828287686565167104 |
|---|---|
| author | Qi Liu Xinyi Wang Bo Yang Zhaojian Wang Yuxiang Liu Xinping Guan |
| author_facet | Qi Liu Xinyi Wang Bo Yang Zhaojian Wang Yuxiang Liu Xinping Guan |
| author_sort | Qi Liu |
| collection | DOAJ |
| description | Abstract Photovoltaics face the threat of many potential faults in daily operation, which calls for accurate fault diagnosis to avoid huge economical losses. This paper investigates practical and troublesome scenarios, where the photovoltaics station has a large number of unlabeled samples and only a few are labelled. In the labelled set, the sample sizes of different types are unbalanced. To this end, a new fault sample rebalancing framework based on semi‐supervised learning (FSR‐SSL) is proposed. Specifically, a dual‐threshold selection mechanism is proposed to choose trusted pseudo‐label samples from unlabeled data to expand the training set. Moreover, a fault sample rebalancing strategy is designed to further filter the obtained trusted pseudo‐label samples, thereby flexibly adding different amounts of pseudo‐label data to various types. As the training rounds increase, the fault samples are gradually rebalanced and the model learning bias caused by type imbalance is well overcome. The extensive numerical experiments show that the proposed FSR‐SSL method reaches 99% accuracy. Compared with existing methods, the accuracy is increased by up to 33%. |
| first_indexed | 2024-04-13T09:52:48Z |
| format | Article |
| id | doaj.art-406c7847cc43471688d47e2ba7a43f13 |
| institution | Directory Open Access Journal |
| issn | 1752-1416 1752-1424 |
| language | English |
| last_indexed | 2024-04-13T09:52:48Z |
| publishDate | 2022-09-01 |
| publisher | Wiley |
| record_format | Article |
| series | IET Renewable Power Generation |
| spelling | doaj.art-406c7847cc43471688d47e2ba7a43f132022-12-22T02:51:32ZengWileyIET Renewable Power Generation1752-14161752-14242022-09-0116122667268110.1049/rpg2.12458FSR‐SSL: A fault sample rebalancing framework based on semi‐supervised learning for PV fault diagnosisQi Liu0Xinyi Wang1Bo Yang2Zhaojian Wang3Yuxiang Liu4Xinping Guan5Department of Automation Shanghai Jiao Tong University Shanghai ChinaDepartment of Automation Shanghai Jiao Tong University Shanghai ChinaDepartment of Automation Shanghai Jiao Tong University Shanghai ChinaDepartment of Automation Shanghai Jiao Tong University Shanghai ChinaDepartment of Automation Shanghai Jiao Tong University Shanghai ChinaDepartment of Automation Shanghai Jiao Tong University Shanghai ChinaAbstract Photovoltaics face the threat of many potential faults in daily operation, which calls for accurate fault diagnosis to avoid huge economical losses. This paper investigates practical and troublesome scenarios, where the photovoltaics station has a large number of unlabeled samples and only a few are labelled. In the labelled set, the sample sizes of different types are unbalanced. To this end, a new fault sample rebalancing framework based on semi‐supervised learning (FSR‐SSL) is proposed. Specifically, a dual‐threshold selection mechanism is proposed to choose trusted pseudo‐label samples from unlabeled data to expand the training set. Moreover, a fault sample rebalancing strategy is designed to further filter the obtained trusted pseudo‐label samples, thereby flexibly adding different amounts of pseudo‐label data to various types. As the training rounds increase, the fault samples are gradually rebalanced and the model learning bias caused by type imbalance is well overcome. The extensive numerical experiments show that the proposed FSR‐SSL method reaches 99% accuracy. Compared with existing methods, the accuracy is increased by up to 33%.https://doi.org/10.1049/rpg2.12458 |
| spellingShingle | Qi Liu Xinyi Wang Bo Yang Zhaojian Wang Yuxiang Liu Xinping Guan FSR‐SSL: A fault sample rebalancing framework based on semi‐supervised learning for PV fault diagnosis IET Renewable Power Generation |
| title | FSR‐SSL: A fault sample rebalancing framework based on semi‐supervised learning for PV fault diagnosis |
| title_full | FSR‐SSL: A fault sample rebalancing framework based on semi‐supervised learning for PV fault diagnosis |
| title_fullStr | FSR‐SSL: A fault sample rebalancing framework based on semi‐supervised learning for PV fault diagnosis |
| title_full_unstemmed | FSR‐SSL: A fault sample rebalancing framework based on semi‐supervised learning for PV fault diagnosis |
| title_short | FSR‐SSL: A fault sample rebalancing framework based on semi‐supervised learning for PV fault diagnosis |
| title_sort | fsr ssl a fault sample rebalancing framework based on semi supervised learning for pv fault diagnosis |
| url | https://doi.org/10.1049/rpg2.12458 |
| work_keys_str_mv | AT qiliu fsrsslafaultsamplerebalancingframeworkbasedonsemisupervisedlearningforpvfaultdiagnosis AT xinyiwang fsrsslafaultsamplerebalancingframeworkbasedonsemisupervisedlearningforpvfaultdiagnosis AT boyang fsrsslafaultsamplerebalancingframeworkbasedonsemisupervisedlearningforpvfaultdiagnosis AT zhaojianwang fsrsslafaultsamplerebalancingframeworkbasedonsemisupervisedlearningforpvfaultdiagnosis AT yuxiangliu fsrsslafaultsamplerebalancingframeworkbasedonsemisupervisedlearningforpvfaultdiagnosis AT xinpingguan fsrsslafaultsamplerebalancingframeworkbasedonsemisupervisedlearningforpvfaultdiagnosis |