摘要
提出一种面向能量数据、基于主元信息熵初始分类策略的发动机故障模式模糊识别方法.该方法首先运用主元能量特征提取方法对相关性强的多维能量数据降维,按可能分类数对能量保持率最大的第一主元进行聚类,同时基于核密度估计并按最大熵原理确定最佳分类数和初始聚类中心,然后仅面向主元能量数据进行模糊聚类,得出最佳聚类中心,再通过计算最大贴近度进行故障模式识别.实验结果表明,因采取了独立的初始分类算法,该方法有效避免了随机选取初值的敏感问题,聚类精度优于传统算法,并可有效降低运算开销,提高故障识别效果.
Proposed in this paper is a fuzzy recognition method of engine fault modes based on the initial classifica-tion strategy with principal component entropy and oriented to energy data.In this method,first,the principal com-ponent data extraction method is used to reduce the highly-relevant multi-dimension data dimension.Next,the first principal component data that maintain the maximum energy are clustered according to the number of possible classi-fications,and the optimal category number as well as the initial cluster centers is determined based on the kernel density estimation and the maximum entropy principle.Then,the best cluster center is produced via the fuzzy clus-tering of the principal component data only.Finally,the fault mode is recognized by calculating the maximum near-ness.Test results show that the proposed method effectively avoids the random selection of initial data due to the adoption of an independent initial classification algorithm,and that it is superior to the traditional algorithm due to its high classification accuracy,low computational overhead and excellent recognition performance.
出处
《华南理工大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2013年第11期137-142,共6页
Journal of South China University of Technology(Natural Science Edition)
基金
广东省自然科学基金资助项目(S2011010002118)
广东省教育部科技部产学研结合项目(2010B090400496)
关键词
故障识别
主元
降维
能量数据
最大熵
模糊k均值
fault recognition
principal component
dimension reduction
energy data
maximum entropy
fuzzy k-means
