基于K均值聚类和混合核函数相关向量机的变压器故障诊断方法

Power Transformer Fault Diagnosis Method Based on K-Means Clustering and Relevance Vector Machine with the Optimized Combined Kernel Function

油中溶解气体分析是电力变压器故障诊断中常用的方法,论文提出了一种基于K均值聚类和混合核函数相关向量机的变压器故障诊断方法。论文首先对样本进行K均值聚类,然后再利用相关向量机的二叉树结构进行分类划分,相关向量机的核函数采用高斯核函数和二项式核函数混合组合的方法。实验结果表明,相比于三比值法、BP神经网络算法,论文所提方法具有较高的变压器故障诊断正确率。

Dissolved gas analysis in oil is a common method in power transformer fault diagnosis.This paper proposes a power transformer fault diagnosis method based on K-means clustering and relevance vector machine with the optimized combined kernel function.Firstly,the paper clusters the samples with K-means,and then uses the binary tree architecture of the relevance vector machine to classify and divide.The kernel function of the relevance vector machine adopts the method of combining Gaussian kernel function and polynomial kernel function.The experimental results show that,compared with the three-ratio method and BP neural network algorithm,the method proposed in the paper has a high accuracy rate of power transformer fault diagnosis.