摘要
随着电网的扩大,变压器在电网中的作用日益提高,从而变压器在线故障诊断技术也显得异常重要。分析了变压器常见的故障类型以及变压器故障与变压器油产生特征气体的对应关系;选出具有代表性的特征气体作为最优诊断特征量;将故障诊断过程分为三个层次:正常-故障层、过热-放电层和故障细化层(高温过热、中低温过热、低能量放电、高能量放电);最后用互信息方法与神经网络法实现基于特征气体分层特性的变压器故障诊断方案。
With the expansion of power grid, the role of the transformer in the power grid is increasing day by day, thus transformer on-line fault diagnosis technology also becomes extremely important. This paper analysis the common faults of transformer and the corresponding relation between transformer faults and the characteristics gas produced from transformer oil. Select the characteristics of typical gas as the optimal diagnostic characteristic features. Fault diagnosis process is divided into three levels： normal-fault layer, overheating fault and the layer-discharge refined layer （high temperature, low temperature overheating in overheating, low energy discharge, high-energy discharge）. Use mutual information method and neural network to realize the scheme of Power Transformer Fault Diagnosis Based on Layered Characteristics of Fault Feature Gases.
出处
《电气技术》
2012年第11期10-13,共4页
Electrical Engineering
