摘要
为更好识别转子系统故障,将正交思想引入到局部Fisher判别分析(LFDA)中,提出了一种基于Schur正交的局部Fisher判别(SOLFD)监督流形学习算法。算法以转子故障训练样本为监督信息,通过局部加权邻接矩阵重新定义类内散度和类间散度,构建局部Fisher判别函数。以判别函数值最大化为目标,通过Schur正交分解方式求解最优正交投影向量。将新增测试数据投影到该向量上,获取新数据故障类别信息。转子故障诊断实验表明,相对其他流形学习算法,SOLFD算法有更好的诊断效果。
In order to better identify the fault of a rotor system,a new method of fault diagnosis based on Schur orthogonal local Fisher discriminant（SOLFD） is proposed to solve the fault diagnosis of a rotor system.In this method,the training data as supervision information was used to compute the local with-class scatter and between-class scatter,and to construct the local fisher discriminant function.To maximize the value of discriminant function,optimal orthogonal projection vector was acquired.New test data was mapped to the projection vector and its fault information was known.Experiment for rotor fault diagnosis shows that SOLFD is better than other manifold learning algorithms.
出处
《机械科学与技术》
CSCD
北大核心
2011年第1期62-65,共4页
Mechanical Science and Technology for Aerospace Engineering
基金
国家自然科学基金项目(51075140)
湖南省自然科学基金重点项目(09JJ8005)
湖南省科技创新团队资助
