Phasor estimation in power systems using a neural network with online training for numerical relays purposes

• Published in: IET science, measurement & technology Vol. 9; no. 7; pp. 836 - 841
• Main Authors: da Silva, Chrystian Dalla Lana, Cardoso Junior, Ghendy, Mariotto, Lenois, Marchesan, Gustavo
• Format: Journal Article
• Language: English
• Published: The Institution of Engineering and Technology 01.10.2015
• Subjects: Algorithms; amplitude estimation; artificial neural network; current measurement; DC offset decaying; Devices; electric current measurement; learning (artificial intelligence); Mathematical models; multilayer perceptron structure; Multilayer perceptrons; Neural networks; numerical analysis; numerical protective relay; Online; online training process; performance evaluation; phase angle estimation; phase estimation; phasor estimation method; phasor measurement; Phasors; power engineering computing; power system measurement; power system protection; relay protection; Training; multilayer perceptrons; artificial neural network; phasor estimation method; power system measurement; power system protection; numerical protective relay; amplitude estimation; performance evaluation; power engineering computing; phase estimation; relay protection; numerical analysis; online training process; electric current measurement; multilayer perceptron structure; current measurement; DC offset decaying; phasor measurement; learning (artificial intelligence); phase angle estimation
• ISSN: 1751-8822; 1751-8830; 1751-8830
• DOI: 10.1049/iet-smt.2014.0312

There are a few components of the current signal that may lead to inaccurate current measurement in power systems, and therefore, may cause malfunction on numerical protective relays and control devices. Some of these components include harmonics, the decaying DC offset, and noises. In this study, a phasor estimation method based on artificial neural networks is proposed, which will provide fast response time and accuracy. The method uses the multilayer perceptron structure to precisely estimate the amplitude and phase angle of the current waveform by determining its input weights during an online training process. The proposed algorithm is tested and compared with other reliable and well-known methods for a performance evaluation.