摘要： 针对神经网络分类器在不同类样本间距离较近时训练速度较慢的缺点，提出了基于粗集理论的归一化方法。利用粗集理论对样本进行归一化处理后，用处理后的样本对神经网络进行训练。并以配电网故障选线为例，对该方法进行了分析。仿真实验结果表明，样本处理后的神经网络训练时间明显缩短。

关键词: 归一化, 粗集理论, 神经网络, 故障选线

Abstract: To overcome the disadvantage of the longtime training of neural network classifier when the distance between samples of different classes is small, the normalization method based on rough set theory is proposed. The samples are normalized using rough set theory and then the normalized samples are used to train neural network. The method is analyzed with an example of faulty line detection for distribution network. The simulation results show that the training time of neural network with processed samples is shorter.

Key words: Normalization, Rough set theory, Neural network, Faulty line detection