基于耦合隐马尔可夫模型的变压器故障诊断研究  

Research on Transformer Fault Diagnosis Based on Coupled Hidden Markov Model

作　　者：过浩[1] 王智博 赵丽莉[3] 朱超[1] 

机构地区：[1]可再生能源发电技术教育部工程研究中心(河海大学),江苏南京210098 [2]宁夏石嘴山供电公司,宁夏石嘴山753000 [3]南瑞集团公司(国网电力科学研究院),江苏南京211106

出　　处：《广东电力》2015年第12期104-108,共5页Guangdong Electric Power

基　　金：国家高技术研究发展计划(863计划)项目资助(2013AA050601)

摘　　要：针对油中溶解气体分析(dissolved gas analysis,DGA)技术存在判断标准过于绝对和编码不全的不足,以及隐马尔可夫模型(hidden Markov model,HMM)的单状态序列结构限制了多通道信息融合的能力,提出基于耦合隐马尔可夫模型(coupled hidden Markov model,CHMM)的故障诊断方法。采用变压器油色谱数据进行模型训练和状态评估测试,验证了该方法的可行性和有效性。算例结果表明,该故障诊断方法的准确率优于传统基于HMM的故障诊断方法,有效降低了诊断误判的风险。In allusion to problems of criteria for dissolved gas analysis （DGA） technology being too absolute and coding being incomplete as well as single-state sequence structure of hidden Markov model （HMM） constraining capacity of multi-channel information fusion, a kind of method for fault diagnosis based on coupled HMM is proposed. Transformer oil spectrum data is used for model training and state evaluation testing that verifies feasibility and effectiveness of this method. Example results indicate that accuracy rate of this diagnosis method is superior to that of the method based on HMM which means effectiveness to reduce risks of erroneous judegement.

关 键 词：电力变压器 故障诊断 隐马尔可夫模型 耦合隐马尔可夫模型 油中溶解气体分析 

分 类 号：TM862[电气工程—高电压与绝缘技术]