Artificial neural network applications for power system protection

The most commonly used systems for protecting transmission and subtransmission lines belong to the family of distance relays. Over the past eighty years, successful designs based on electromechanical, solid-state and digital electronics technologies have been produced and marketed. These relays implement various characteristics, such as impedance, offset-impedance, admittance, reactance and blinders. The artificial neural network based designs of distance relays proposed so far work well for ideal fault conditions but are not able to maintain the integrity of the boundaries of the relay characteristics of generic designs. This paper reviews ANN models that have been proposed in the past for protecting components of power systems and presents a methodology that fully exploits the potential of ANNs in designing generic distance relays that retain the integrity of the boundaries of their characteristics

1 Introduction

Artificial neural network based technology, which is inspired by biological neural networks, has developed rapidly in the previous decade and has been applied in power system protection applications. Specific applications include direction discrimination for protecting transmission lines [1] [2], fault classification for faults on double circuit lines [3], ANN based distance relays [4], differential protection of three phase power transformers [7] and faults on generator windings [8]. The ANN based designs of generic protection systems proposed so far work well only for ideal fault conditions but do not maintain the integrity of the boundaries of the relay characteristics. This deficiency exists even if the networks are trained to identify the operating states in the neighborhood of the boundaries of the characteristics. This is because of the fact that none of the designs fully exploit the potential of ANNs in implementing generic relay characteristics for maintaining the integrity of their boundaries. Investigations of the fundamental drawbacks of ANNs and development of new strategies for designing ANNs, which would work as generic distance relays with clearly defined operating boundaries, are needed.