摘要
提出一种基于因子分析和基因表达式编程(GEP)的互感器故障诊断方法。先用因子分析对原数据进行一系列变换,可降低特征向量的维数,并消除向量间的相关性,从而减小了故障分类器的计算复杂度,提高训练及测试的精度。然后将得到的新样本数据用基因表达式编程算法进行训练,并构建互感器故障诊断模型。试验表明,该模型比单独使用遗传规划(GP)或遗传算法(GA)得到的结果具有更高的诊断精度和稳定性。
An approach of mutual inductor fault diagnosis based on factor analysis(FA) and gene expression programming algorithm(GEP) is proposed in this paper.Firstly,the original data is transformed with FA,which can decrease the dimensions of eigenvector and eliminate the relevance between the vectors,and then the accuracy of the data training and testing is increased while the calculation complexity of fault sorter minishes.Then,trainss the new sample data with GEP and structures the model of mutual inductor fault diagnosis.The test shows that the higher accuracy and stability are gained by this model than genetic algorithm or genetic programming solely.
出处
《电力科学与工程》
2011年第10期26-30,56,共6页
Electric Power Science and Engineering
关键词
互感器
故障诊断
基因表达式程序设计
因子分析
mutual inductor
fault diagnosis
gene expression programming
factor analysis.