中文摘要：

通过对局部放电的模式识别可以了解放电类型及严重程度,并在此基础上确定维护方案。为了对局部放电进行识别,建立了油纸绝缘中的5种典型缺陷模型;运用K-W检验从相间局部放电（PRPD）统计算子中提取出分类能力最强的11个特征;基于提取的特征,在小样本训练集的前提下,利用层次分析法对典型放电模型进行识别,同时和同种情况下使用人工神经网络的识别效果进行了比较。实验结果表明,在小样本训练集下,运用层次分析法得到了较好的识别效果,正判率均大于85%,优于人工神经网络,这为小样本训练集的情况下局部放电的快速识别奠定了基础。

英文摘要：

The pattern recognition of partial discharge（PD） can facilitate estimations of PD mode and grade,and design of protection scheme.To recognize partial discharges,five kinds of typical defects in oil-paper insulation are set up.Through K-W test,11 features with the strongest ability of classification are chosen from the statistical operators of Phase Resolved Partial Discharges（PRPD）.In the case of being trained by small sample,the analytic hierarchy process（AHP） is employed to recognize these typical partial discharges（PD） based on the chosen features.Compared with the artificial neural network（ANN） in the same condition,AHP achieves higher recognition accuracy of over 85%.This study may lay a foundation for rapid recognition of typical partial discharges in the case of small training sample.