基于自适应免疫算法的变压器故障诊断  

Transformer Fault Diagnosis Based on Adaptive Immune Algorithm

作　　者：邢涛 李志军 曹玲燕 丁立华 

机构地区：[1]华电国际电力股份有限公司山东分公司,山东济南250014 [2]江苏国电南自海吉科技有限公司,江苏南京211153 [3]华电青岛发电有限公司,山东青岛266000

出　　处：《发电与空调》2017年第6期42-45,共4页

基　　金：华电集团科技项目"火电厂电气设备智能运维技术研究与示范应用"

摘　　要：变压器油中溶解气体分析法DGA(Dissolved Gas Analysis)是变压器故障诊断的重要方法,针对变压器故障的特点,提出了一种自适应免疫遗传算法进行在线故障诊断。该算法将传统的免疫算法与约束独立成分分析c ICA相结合,将研究对象的先验信息作为约束条件,使新的算法仅仅收敛于感兴趣的故障信号,更快速有效的进行分类;加速抗体种群数量和促进更新概率,使抗体库更加有效的进化。改进后的算法应用于DGA电力变压器数据分析,实现了变压器的故障在线诊断。实验数据表明:该算法收敛速度快,能有效地对样本进行识别,提高了故障诊断的针对性和效果。Transformer oil Dissolved Gas Analysis in the DGA(Dissolved Gas Analysis) is an important method for transformer fault diagnosis.According to the characteristics of transformer faults,an adaptive immune genetic algorithm is proposed for on-line fault diagnosis.The algorithm combines the traditional immune algorithm with the constraint independent component analysis c ICA,and takes the prior information of the object as a constraint condition,so that the new algorithm only converges to the fault signal of interest,faster and more efficient classification.Promote the number of antibody populations and update probability to make the antibody library more effective.The algorithm is applied to DGA power transformer data analysis,and the transformer fault on-line diagnosis is realized.Experimental data shows that this algorithm can effectively identify the samples and improve the pertinence and effect of fault diagnosis.

关 键 词：故障诊断 免疫遗传算法 约束独立成分分析 变压器油中溶解气体 

分 类 号：TM41[电气工程—电器]