Active islanding detection method via current injection disturbance using Elman neural network

Faa-Jeng Lin,Yi-Sheng Huang,Kuang-Hsiung Tan,Yung-Ruei Chang
DOI: https://doi.org/10.1080/02533839.2014.981934
2014-12-17
Journal of the Chinese Institute of Engineers
Abstract:Islanding detection is an essential protection requirement for distributed generators for personnel and equipment safety. This study presents a new active islanding detection method via current injection disturbance using intelligent tracking control of the active and reactive power commands. First, an active islanding detection method based on injecting a disturbance signal into the system through the d-axis current which leads to a frequency deviation at the terminal of the RLC load when the grid is disconnected is proposed. The feasibility of the proposed method is evaluated under the UL1741 anti-islanding test configuration. Then, to further improve the performance of the islanding detection method, an Elman neural network intelligent controller is adopted to replace the proportional-integral controller used in the traditional injection method for islanding detection. Moreover, the network structure and the online learning algorithm are described in detail. Finally, the feasibility and effectiveness of the proposed current injection disturbance method are verified with experimental results.
engineering, multidisciplinary
What problem does this paper attempt to address?