基于气体成分分析的变压器过热隐患预警方法  

An early warning method for transformer overheating based on gas composition analysis

作　　者：谭翼坤 陈明 黄腾 乔苏朋 李志军[2] TAN Yikun;CHEN Ming;HUANG Teng;QIAO Supeng;LI Zhijun(Fujian Huadian Kemen Power Generation Co.,Ltd,Fuzhou 350500;Guodian Nanjing Automation Co.,Ltd,Nanjing 211106)

机构地区：[1]福建华电可门发电有限公司,福州350500 [2]国电南京自动化股份有限公司,南京211106

出　　处：《电气技术》2022年第10期51-58,共8页Electrical Engineering

摘　　要：目前,油中溶解气体分析是油浸式变压器状态分析最为可靠的方法之一。本文针对变压器过热隐患,提出一种基于气体成分分析的变压器过热隐患预警方法。首先,采用K级差分法及一分类支持向量机建立能够识别变压器早期隐患的预警模型;其次,遵循特定的告警触发机制,综合判断当前变压器是否存在过热隐患;最后,经过变压器真实数据验证表明,本文所述方法具有较高的辨识灵敏度。At present, the analysis of dissolved gas in oil is one of the most reliable methods for the condition analysis of oil-immersed transformers. Aiming at the hidden danger of transformer overheating, this paper proposes an early warning method for the hidden danger of transformer overheating based on gas composition analysis. Firstly, the K-level difference method and one-class support vector machine are used to establish an early warning model that can identify the early hidden dangers of transformers. Secondly, following a unique alarm triggering mechanism, it is comprehensively judged whether the current transformer has hidden dangers of overheating. Finally, the verification of real transformer data shows that the proposed method has high identification sensitivity.

关 键 词：油中溶解气体分析(DGA) 一分类支持向量机 预警 

分 类 号：TM411[电气工程—电器]