Laboratory investigation of a digital recurrent network for transmission line directional protection

This paper presents a recurrent neural network-based technique to identify the direction of a fault on a transmission line. The network uses samples o f voltage and current signals to learn the hidden relationship existing in the input patterns. The proposed small size recurrent network has been impl emented on a digital signal processor board and its behavior is investigate d on a physical power system model. Details of implementation and the exper imental studies are given and analyzed in the paper. Performance studies re sults show that the proposed approach is able to detect the direction of a fault on a transmission line rapidly and correctly. It shows that the propo sed network is robust, fast and accurate. (C) 1998 Elsevier Science B.V. Al l rights reserved.