基于神经网络原理的高压架空输电线路故障测距模型的研究

THE MODEL STUDY OF FAULT LOCATION SYSTEM FOR HIGH VOLTAGE OVER-HEAD TRANSMISSION LINE BASED ON HIERARCHICALDISTRIBUTED NERUAL NETWORKS

根据目前广泛应用的单端电气量的基频分量实现单回高压架空输电线路故障测距的其各种算法尚存在的不足之处，如求解出现伪根或可能存在迭代不收敛的问题；测距结果受对端系统运行方式变化的影响等。本文运用人工神经网络的理论开辟了故障测距的新途径，提出用基于分布分层式神经网络系统来实现高压架空输电线路故障测距的新方法。经研究和仿真表明：该方法能有效地实现输电线路故障点的精确定位，测距结果不受故障过渡电阻和对端运行方式的影响，且可获得全线基本均匀的测距绝对误差。

At present,all of algorithms used widely in our country to fault location of the single-circle high voltage over-head transmission line,based on fundamental components of one-terminal information,have some disadvantages.And moreover,artificial neural network has powerful function at aspects of classification,nonlinear-approaching et..So this paper applies artificial neural network to fault location of transmission line.For the problem of fault location ,the Hierarchical Distributed NN model is constructed on the basis of the analysis of NN models,and a lot of training and testing to this NN model structure have been carried on .The simulation results show that this model can accomplish accurate fault location of transmission line,and the result cannot be affected by the factors of fault transitional resistance,the other terminal system impedance,and fault angle et.The result in this paper opens up a new direction for the study and application of fault location.