摘要
针对变压器的故障诊断,很多解决方法已被提出,但都有各种缺陷。为了提高变压器的故障诊断判正率,保证得到较高的精确度,提出了一种基于纠错编码和支持向量机相结合的多分类算法。介绍了纠错编码的原理应用并分析了编码长度、码间汉明距离与支持向量机多分类算法的推广性关系。运用VS2008对变压器中油中溶解气体(DGA)数据进行了仿真,结果表明该算法适合于变压器故障诊断。
For transformer fault diagnosis, a lot of solutions have been made, but they have many defects. In order to improve the correct judging ratio of transformer fault diagnosis and guarantee high accuracy, A muhiclass classifi- cation algorithm based on error-correcting codes and support vector machine is proposed. The application of the principle of error-correcting codes is described and the relation between generalization performance of muhi-classifi- cation algorithm of support vector machine and the code length, Hamming distance between codes is studied. U- sing VS2008 dissolved gases in transformer oil data is simulated and diagnosis shows that the algorithm has higher classification accuracy.
出处
《电力科学与工程》
2012年第11期39-43,共5页
Electric Power Science and Engineering
关键词
故障诊断
纠错编码
支持向量机
多分类
变压器
fauh diagnosis
error-correcting codes
Support Vector Machine
multiclass classification
transform-el
