基于聚类-小波神经网络的油纸绝缘气隙放电发展阶段识别方法

Method to Identify Developing Stages of Air-Gap Discharge in Oil-Paper Insulation Based on Cluster-Wavelet Neural Network

基于油纸绝缘气隙放电模型,在实验室搭建了气隙放电及其发展特性研究试验平台;采用恒压法,对其进行局部放电发展特性实验;提取了局部放电最大放电量相位分布、平均放电量相位分布、放电次数相位分布以及局部放电幅值分布中的29个特征参量,通过核主成分分析,采用系统聚类对放电不同的发展阶段进行划分。建立了基于聚类-小波神经网络的放电发展阶段识别方法,识别结果表明:所建立的识别方法能很好地根据放电有效特征量识别放电所处阶段,与系统聚类分析结果基本吻合。

Based on air-gap within oil-paper insulation,a test platform was set up in laboratory to research air-gap discharge and its development stages.Using constant voltage method the experiments for developing characteristics of partial discharge were performed and 29 parameters of maximum discharge amplitudes,average discharge amplitudes and discharge times of phase resolved partial discharge（PRPD） as well as amplitude distributions of discharge were extracted;utilizing kernel principal component analysis,the different developing stages of partial discharge were divided by system clustering.A method based on cluster-wavelet neural network to identify developing stages of discharge was established.Identification results show that according to effective discharge characteristic quantities the proposed method can identify discharging stages well and the obtained results are basically conform to the results of system clustering analysis for the same data.