A Hybrid Framework for Fault Detection, Classification, and Location—Part I: Concept, Structure, and Methodology
| Autor: | Chih-Hung Hung, Ying-Tung Hsiao, Joe-Air Jiang, Cheng-Long Chuang, Yung-Chung Wang, Jiing-Yi Wang, Chien-Hsing Lee |
|---|---|
| Rok vydání: | 2011 |
| Předmět: |
Engineering
Artificial neural network business.industry Energy Engineering and Power Technology Wavelet transform computer.software_genre Fault detection and isolation Fault indicator Stuck-at fault Support vector machine Fault coverage Electronic engineering Data mining Electrical and Electronic Engineering Fault model business computer |
| Zdroj: | IEEE Transactions on Power Delivery. 26:1988-1998 |
| ISSN: | 1937-4208 0885-8977 |
| DOI: | 10.1109/tpwrd.2011.2141157 |
| Popis: | Bridging the gap between the theoretical modeling and the practical implementation is always essential for fault detection, classification, and location methods in a power transmission-line network. In this paper, a novel hybrid framework that is able to rapidly detect and locate a fault on power transmission lines is presented. The proposed algorithm presents a fault discrimination method based on the three-phase current and voltage waveforms measured when fault events occur in the power transmission-line network. Negative-sequence components of the three-phase current and voltage quantities are applied to achieve fast online fault detection. Subsequently, the fault detection method triggers the fault classification and fault-location methods to become active. A variety of methods-including multilevel wavelet transform, principal component analysis, support vector machines, and adaptive structure neural networks-are incorporated into the framework to identify fault type and location at the same time. This paper lays out the fundamental concept of the proposed framework and introduces the methodology of the analytical techniques, a pattern-recognition approach via neural networks and a joint decision-making mechanism. Using a well-trained framework, the tasks of fault detection, classification, and location are accomplished in 1.28 cycles, significantly shorter than the critical fault clearing time. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|