| 摘要: |
| 电力变压器故障诊断中的测试数据信息不完备、有偏差，而贝叶斯网络处理不确定性问题能力强。提出了一种基于选择性贝叶斯分类器的、溶解气体分析结合其他电气试验结果的变压器故障诊断方法，并建立了变压器选择性贝叶斯故障诊断模型。详细阐述并验证了该方法解决信息不完备问题的优越性。该模型还可以通过不断积累完善训练样本，自动修正网络结构参数和概率分布参数。实验表明提出的选择性贝叶斯分类器适于变压器故障诊断。 |
| 关键词: 变压器 故障诊断 贝叶斯网络 选择性分类器 |
| DOI： |
| 分类号: |
| 基金项目:河北省自然科学基金项目（E2009001392)；中央高校基本科研业务费专项资金资助项目（09QG33） |
|
| Transformer fault diagnosis based on selective Bayes classifier |
|
ZHAO Wenqing
|
|
North China Electric Power University，Baoding 071003，China
|
| Abstract: |
| As the test data of electric power transformer fault diagnosis are incomplete and biased，a transformer fault diagnostic method is proposed based on selective Bayes classifier，which，with the ability to process uncertain information，combines ＤＧＡ（Dissolved Gas Analysis） with other electrical test results. The transformer fault diagnosis model is built based on selective Bayes classifier. Its superiority in uncertain information processing is elaborated in detail. With the accumulated and improved training samples，it automatically modifies the parameters of network structure and probability distribution. Experimental results show that Bayes classifier is suitable for the transformer fault diagnosis. |
| Key words: transformer fault diagnosis Bayes network selective classifier |
