人工神经网络应用于输电线路故障测距研究

Study of Applying ANN to Locating Fault on HV Transmission Lines

随着人工神经网络理论的发展，人工神经网络用于输电线路距离保护和故障测距的研究日益受到关注。在有关此项研究的大量文献中，笔者发现所建立的训练样本均未考虑本端等值系统阻抗及电势变化对故障测距或距离保护的影响。研究表明，上述二个因素对故障测距或距离保护有很大影响。此外由于人工神经网络用于故障测距时应当考虑的因素极多，所需训练样本巨大，且训练很难收敛。大量数字仿真表明，人工神经网络实用于故障测距尚需进一步研究。

With the development of ANN theory, ANN has been used in locating fault and in distance relaying for transmission line. The author no tices that, in the most of the articles about ANN used in fault location and dis tance relaying, the equivalent system impedance and electromotive force variatio n at near terminal are not taken into account in selecting the train samples. B ut the two factors described as above have the considerable effects upon the res ults of locating fault and distance relaying, which are not taken into account i n the present algorithm of fault location. Moreover, many factors should be cons idered in fault location using ANN, which may encounter to a large scale of samp lesset and the difficulty in convergence. Then, the conclusion that it is not exactly feasible to apply ANN to locating fault is drawn.