Adaptive protection for series-compensated transmission lines using neural networks

Document Type : Original Article

Authors

1 Student member, IEEE., A. Hosny is a Doctoral Candidate at the State University of New York at Buffalo, Amherst, NY 14260 USA. Phone: 716 645 3115 ext. 1204; Fax: 716 645 365.

2 Fellow, IEEE., M. Safiuddin is with the Department of Electrical Engineering, University at Buffalo, Amherst, NY 14260 USA.

Abstract

Abstract:
This paper presents an adaptive protection approach for classifying and locating faults
in Thyristor Controlled Series-Compensated (TCSC) transmission lines. The proposed
scheme is based on Multilayer Feedforward Neural Networks (MFNNs). Levenberg-
Marquardt (LM) training algorithm is employed. The LM algorithm appears to be the
fastest training algorithm and highly nominated for better generalized models. Threephase
power system currents and voltages at the relay location are used as inputs to
MFNN-based relay. Two neural networks are trained to address fault classification and
location. Feasibility and reliability of the proposed scheme are investigated using fault
data set of a typical 500 kV power system simulated in EMTP-ATP package. Studied
system is subjected to all possible shunt faults at different operating conditions,
including fault location, fault inception angle and fault resistance. Simulation results
demonstrate that MFNN-based relay system is very robust, fault tolerant, and highly
accurate in protecting Flexible AC Transmission Systems (FACTS), such as
transmission lines with TCSC.

Keywords

Files

Share

How to cite

Statistics