当前位置:主页 > 科技论文 > 自动化论文 >

基于相对变换的ICA故障检测方法

发布时间:2018-01-12 00:20

  本文关键词:基于相对变换的ICA故障检测方法 出处:《电子测量与仪器学报》2017年07期  论文类型:期刊论文


  更多相关文章: 电主轴 故障检测 相对变换 欧氏距离 独立主元分析


【摘要】:针对传统独立主元分析方法(independent component analysis,ICA)在标准化处理后导致特征值大小近似相等,难以提取有代表性变量等问题,提出了一种基于相对变换的独立主元分析(relative transformation ICA,RTICA)故障检测方法。该方法引入欧氏距离相对变换理论,将原始空间数据变换得到相对空间,然后在相对空间进行独立主元分析,降低相对空间的数据维数,使提取的独立主元特征具有更大的适应性,建立故障检测模型,最终实现在线故障检测。该方法通过田纳西-伊斯曼过程仿真加以验证,并应用到电主轴裂纹故障的状态监测中,实验结果表明该方法能有效减少独立主元个数,简化故障检测模型的复杂度,增强状态检测性能。
[Abstract]:In view of the traditional independent component analysis method (independent component analysis, ICA) in the size of approximately equal eigenvalues in the standardized treatment, it is difficult to extract representative variables and other issues, put forward a kind of independent component analysis based on relative transformation (relative transformation ICA, RTICA) fault detection method. The method is based on Euclidean distance relative transform theory, transform the original data to obtain the relative space space, then independent component analysis in relative space, reduce the dimension of the space is relatively independent, the principal component extraction feature has more adaptability, establish the model of fault detection, finally realizes the online fault detection. The method by Tennessee Eastman process simulation examples, and applied to condition monitoring of electric spindle crack fault. The experimental results show that this method can effectively reduce the number of independent component, simplified fault detection The complexity of the model, enhanced state detection performance.

【作者单位】: 沈阳建筑大学国家地方联合工程实验室;
【基金】:沈阳市科技计划(17-231-1-28) 辽宁省自然科学基金(2016010623) 中国博士后科学基金(2016M601335)资助项目
【分类号】:TG659
【正文快照】: 1引言随着高速加工技术的不断进步和发展,尤其是在航空、航天、汽车、轮船等高端技术行业的广泛应用,以及机械、电子等产品的需求不断增加,使得数控机床技术越来越受到重视。数控机床是将高效率、高精度以及高柔性集为一体,不仅提高精度,而且提高生产效率,而高速电主轴又是数

本文编号:1411875

资料下载
论文发表

本文链接:https://www.wllwen.com/kejilunwen/zidonghuakongzhilunwen/1411875.html


Copyright(c)文论论文网All Rights Reserved | 网站地图 |

版权申明:资料由用户fff78***提供,本站仅收录摘要或目录,作者需要删除请E-mail邮箱bigeng88@qq.com