粗糙集约简算法在变压器故障诊断中的应用研究  

Application of Rough Set Reduction Algorithm in Fault Diagnosis of Power Transformer

作　　者：吕宏丽[1] 

机构地区：[1]唐山学院信息工程系,河北唐山063000

出　　处：《唐山学院学报》2015年第3期29-31,86,共4页Journal of Tangshan University

基　　金：2014年度河北省高等学校科学技术研究项目(Z2014037)

摘　　要：将粗糙集理论和神经网络技术应用于变压器故障诊断中,粗糙集约简作为神经网络的前置单元,采用基于属性重要性的约简算法。详细阐述了基于属性重要性的约简算法和实现方法,经实际数据训练和测试结果表明,该算法减少了输入样本数,提高了训练速度效率和故障诊断准确率,验证了该算法应用于变压器故障诊断系统的可行性和有效性。In this paper, the rough set theory and neural network technology are applied to transformer fault diagnosis, through reduction algorithm with rough set reduction as the pre-unit of neural network. Reduction algorithm and its implementation are discussed in detail. The actual training data and test results show that rough set reduction algorithm reduces the number of input samples, improves the training speed and efficiency and accuracy of fault diagnosis. In conclusion, the application of rough set reduction algorithm to fault diagnosis system of power transformers is feasible and effective.

关 键 词：故障诊断 神经网络 粗糙集 属性约简 属性重要性 

分 类 号：TP299[自动化与计算机技术—检测技术与自动化装置]