| Autor: |
Hualiang Zhou, Lu Lu, Gaoming Wang, Zhantao Su |
| Jazyk: |
angličtina |
| Rok vydání: |
2024 |
| Předmět: |
|
| Zdroj: |
IEEE Access, Vol 12, Pp 16095-16104 (2024) |
| Druh dokumentu: |
article |
| ISSN: |
2169-3536 |
| DOI: |
10.1109/ACCESS.2024.3358399 |
| Popis: |
With the rapid development of digital operation and maintenance of transformers, the interference of abnormal data caused by the low reliability of online monitoring sensors has become increasingly evident, directly affecting the accuracy of transformer fault diagnosis and impacting the development of digital operation and maintenance of transformers. Therefore, detecting abnormal data and distinguishing whether abnormal data are caused by equipment faults is an important prerequisite for the accurate diagnosis of transformers. When the difference between normal and abnormal data from online monitoring is insignificant, the existing methods are not ideal for detecting abnormal transformer data and cannot determine invalid data caused by sensor failure. To accurately identify valid data and remove invalid data, this study proposes a combined association rule and improved density clustering method to detect abnormalities in the online monitoring data of transformers. First, association rules mimic the association relationships between transformer data sequences. Then, the association rules are combined with an improved density clustering method based on the number of categories–neighborhood radius threshold curve–to detect anomalies in the sequences. Finally, the accurate identification of abnormal data is achieved, and invalid data caused by sensor reasons are effectively screened, thereby screening valuable data generated by transformer faults. The test results indicate that the proposed method can effectively improve the accuracy and stability of abnormal data detection in online monitoring of transformers. |
| Databáze: |
Directory of Open Access Journals |
| Externí odkaz: |
|
