基于KPCA和LS-SVM的变压器故障诊断研究  

A Fault Diagnosis Method for Transformer Based on Least Squares Support Vector Machine Optimized by Kernel Principal Component Analysis

作　　者：高国磊 李英娜[1] 王昕 段效琛[1] 李川 

机构地区：[1]昆明理工大学信息与自动化学院,云南昆明650500 [2]云南电网有限责任公司电力科学研究院,云南昆明650000 [3]中国南方电网公司电能计量重点实验室,云南昆明650217

出　　处：《软件》2017年第10期158-161,共4页Software

摘　　要：电力变压器故障机理复杂,具有不确定性,难以进行准确的状态评估,提出核主成分分析和最小二乘支持向量机结合的变压器诊断方法。首先对样本数据进行非线性映射到高维空间,对映射后的特征向量进行信号重构,其次利用特征空间信号重构的最小误差准则对数据进行离群判断,找出异常特征样本并剔除,最后将核主元分析方法提取特征的数据输入最小二乘支持向量机中分类,识别数据是否存在故障及故障的类型。结果证明本方法的可行性和有效性。Power transformer failure mechanism is complex, with uncertainty, it is difficult to carry out accurate state assessment.A method of transformer diagnosis based on kernel principal component analysis and least squares support vector machine was proposed. Firstly, the sample data were mapped to high-dimensional space, and the re-constructed feature vector was reconstructed. Secondly, the minimum error criterion of the reconstructed feature space signal was used to judge the outliers, to find out the abnormal feature samples and remove them. Finally, the data of the kernel principal component analysis method were input into the least squares support vector machine, and identify whether the data is faulty and the type of fault. The results show that the method is feasible and effective.

关 键 词：变压器 故障诊断 核主成分分析 最小支持二乘向量机 

分 类 号：TM403[电气工程—电器]