基于pLSA模型的滚动轴承故障检测  

Fault detection of rolling bearings based on pLSA model

作　　者：朱韶平[1] 樊晓平[1,2] 

机构地区：[1]湖南财政经济学院信息管理系,长沙410205 [2]中南大学信息科学与工程学院,长沙410075

出　　处：《计算机应用研究》2015年第10期2999-3003,共5页Application Research of Computers

基　　金：湖南省科技计划资助项目(2014FJ3057);湖南省教育科学"十二五"规划课题(XJK015BGD007);湖南省普通高等学校教学改革研究资助课题(湘教通【2012】401号544);湖南省教育厅资助科研项目(15B040)

摘　　要：针对滚动轴承故障检测过程中训练样本收集难的问题,研究了视觉词袋模型和p LSA算法,提出了一种基于概率潜在语义模型的滚动轴承故障检测新方法。为了减少计算复杂度,降低特征的维数,在利用小波包变换提取滚动轴承故障特征后,引入视觉词袋模型将故障特征表示为视觉词袋特征的直方图;为了减少训练样本收集的难度,解决小样本问题,进而运用p LSA模型对滚动轴承故障进行检测。实验结果表明,该方法缩短了训练时间,提高了检测精度,具有一定的实用价值。Consdering the difficulty of the training sample collection in fault detection of rolling bearings,this paper proposed a new method for fault detection of rolling bearings based on probabilistic latent semantic（ pLSA） model after studying the bag-of-visual-words model and pLSA algorithm. To reduce the computational complexity and the dimensions of the feature,it introduced the bag-of-visual-words model after extracting fault features of the rolling bearings using wavelet packet transform,which put the fault features into the bag-of-visual-words histogram. At last it used pLSA model to detect faults of rolling bearings,which reduced the difficulty of the training sample collection and solved the small sample problem. The experimental results show that the proposed method based on pLSA model shortens training time and improves detection rates and has practical values.

关 键 词：滚动轴承故障检测 小波包变换 视觉词袋模型 PLSA模型 

分 类 号：TP207[自动化与计算机技术—检测技术与自动化装置]