期刊文献+

基于PCA的汽车涂装线设备信号特征提取 预览 被引量:13

Signal feature extraction for automobile coating equipment based on PCA
在线阅读 下载PDF
收藏 分享 导出
摘要 应用监测系统对设备信号进行采集、处理可以得到设备运行参数空间。为解决参数空间信息重叠多、维数大及对目标状态区分度小的问题,提出了一种基于主成分分析的特征提取方法并加以改进。首先,在建立烘房设备监测结构图的基础上,进行监测参数PCA建模,给出原数据空间降维改进具体化、层次化方法;然后,通过比较设备信号特征对各主成分影响程度高低、各主成分贡献率大小的方法,得到能真正反映设备运行状态的核心特征集,并对降维效果进行比较,依此确定最佳维数;通过烘房燃烧加热系统验证了该方法能够有效地提取复杂系统的信号特征。 Running parameter space could be obtained by collecting and processing the signals of the equipment using a monitoring system,but the parameter space information is overlapped,has large dimensions and poor distinguishing ability for target states.Aiming at these problems,a new method of extracting features based on principal component analysis was proposed and improved.Firstly,a PCA model of monitoring parameters is set up on the basis of the structure chart for the monitoring equipment in oven system and a hierarchical method for reducing the dimensions of initial data space is proposed.Secondly,the core feature sets are acquired through comparing the influence degree of initial signal features of automobile coating line equipment on each principal component and the variances explained of different principal components;and the dimension reduction effects are compared and the best monitoring space dimension is achieved.Finally,a case study on an oven heating system is provided,and test results prove that the proposed method can extract features effectively in complicated systems.
作者 叶永伟 刘志浩 黄利群 Ye Yongwei,Liu Zhihao,Huang Liqun(1 Key Laboratory of E & M,Ministry of Education & Zhejiang Province,Zhejiang University of Technology, Hangzhou 310014,China;2 Dongfeng Nissan Diesel Motor Co.,Ltd,Hangzhou 310015,China)
出处 《仪器仪表学报》 EI CAS CSCD 北大核心 2011年第10期 2363-2370,共8页 Chinese Journal of Scientific Instrument
基金 浙江省科技厅项目(2008C21160)资助
关键词 PCA 降维 特征提取 加热系统 PCA reduce dimension feature extraction heating system
  • 相关文献

参考文献17

二级参考文献87

共引文献138

同被引文献124

引证文献13

二级引证文献58

投稿分析
职称考试

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部 意见反馈