【摘要】 提出一种新的基于自谐振神经网络结构的自适应故障选相元件。采用故障分量作为网络输入量。由于故障分量能够很好地反映故障特征,而自谐振神经网络结构具有训练快速、不存在局部极小值点以及网络权值稳定等特点,因此自适应选相元件不受线路运行方式变化的影响。大量EMTP仿真和实际故障录波数据验证了其可靠性和正确性。更多还原

【Abstract】 This paper presents a new adaptive phase selector with an adaptive resonance theory (ART) based neural network. The fault component that can represent the essential fault characteristic is selected as the network input. Because the ART based neural network is characterized by fast weight training, stable network weight and absence of partial minimum point, the proposed method can dynamically adapt to the varying operation conditions of the power system. Numerous EMTP simulations and experimental field data tests show that the new phase selector is reliable and effective under various operation conditions.更多还原