FCM属性约简方法在汽轮机故障诊断中的应用  

An approach of attribute reduction based on FCM and its application in fault diagnosis on turbo-generator unit

作　　者：齐晓轩[1,2] 纪建伟[1] 原忠虎[2] 

机构地区：[1]沈阳农业大学信息与电气工程学院,辽宁沈阳110161 [2]沈阳大学信息工程学院,辽宁沈阳110044

出　　处：《合肥工业大学学报（自然科学版）》2011年第2期232-236,共5页Journal of Hefei University of Technology：Natural Science

基　　金：国家自然科学基金资助项目(60974113);沈阳市科技计划资助项目(1091038-4-00)

摘　　要：文章通过汽轮机故障样本数据的空间分布状态分析,基于模糊c均值(FCM)聚类方法实现故障征兆参数属性的离散化处理;采用粗糙集(RS)属性约简方法对原始特征向量进行优化,去除冗余特征,提取特征向量的本质信息;基于约简后的特征向量建立支持向量机(SVM)故障诊断模型。实验结果表明,该文所提出的方法不仅可以提高故障诊断的准确性和实时性,还可以减少故障检测项目,降低故障诊断成本。Fuzzy c-means（FCM） clustering method is applied to discretizing continuous attributes of turbo-generator unit faults by the analysis of spatial distribution of pattern datasets.The attribute reduction algorithm for rough set（RS） is utilized to optimize original feature vectors of turbo-generator unit faults,eliminate redundant information and extract essential information of feature vectors.Fault diagnosis model based on support vector machine（SVM） is built up according to the reduced feature vectors.Experimental results demonstrate that the proposed method can improve the accuracy and real-time performance of fault diagnosis,decrease the number of test items and accordingly cut down the cost of fault diagnosis on turbo-generator unit.

关 键 词：故障诊断 粗糙集 模糊C均值聚类 属性约简 支持向量机 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]