複素数を用いたニューラルネットワークによる送電線の故障区間標定

概要

論文の詳細を見る
Distribution patterns of grand wire current along transmission lines have been analyzed by a experts in order to find the fault location. Various types of fault location methods using multi-layered feed forward type neural network models have been proposed and these efficiencies have been shown in several papers. The amplitudes and the phase angles are used as the input data of the neural network. If we use complex number for the neural network, we can relate the amplitude and the phase angle to each other. In this paper, we propose a fault location method using the multi-layered feed forward type neural network model. In the proposed method, the neurons are expressed by the sigmoid functions which are extend to the complex number and we use the complex back propagation learning algorithm.