Two‐layer coordination architecture HIF detection with µPMU data. Issue 15 (31st August 2018)

• Record Type: Journal Article
• Authors: Wang, Xiaojun; Zhang, Yongjie; Luo, Yiping; He, Jinghan; Ling, Ping; Fang, Chen
• Is Part Of: Journal of engineering. Volume 2018:Issue 15(2018)
• Page Start: 1033
• Page End: 1037
• Publication Date: 2018-08-31
• Subjects: pattern clustering -- support vector machines -- feature extraction -- principal component analysis -- phasor measurement -- fault diagnosis -- power distribution faults -- power engineering computing; detection scheme -- two‐layer coordination architecture HIF detection -- μPMU data -- data‐driven method -- master station -- synchronous data -- principal component analysis -- data categories -- HIF fault -- 34 nodes distribution network -- high‐impedance fault detection -- single‐ended microphasor measurement unit -- feature extraction -- k‐means clustering algorithm -- silhouette coefficient -- support vector machine -- PSCAD‐EMTDC; Engineering -- Periodicals; Engineering; Electronic journals; Periodicals; 620.005
• DOI: 10.1049/joe.2018.0258
• Languages: English
• ISSNs: 2051-3305

Abstract

Abstract : The detection of high‐impedance fault (HIF) on distribution network has been one of the most difficult problems. This study presents a data‐driven method for HIF detection by using single‐ended micro‐phasor measurement unit (µPMU). This approach is based on the two‐layer coordination architecture, local side with μPMUs and master station for further analysis. At the local side, the μPMUs gather the synchronous data and achieve feature extraction with k ‐means clustering algorithm and principal component analysis. For determining the amounts of data categories, the authors adopt a method based on silhouette coefficient. Meanwhile, send the characteristics to the master station, then detect the HIF fault through the support vector machine. Finally, the method was tested on a 34 nodes distribution network in PSCAD/EMTDC. The results justify the effectiveness and the proposed detection scheme has >85% accuracy.