| 一种间歇过程异常数据剔除的主元分析方法 |
| |
| 引用本文： | 刘强,陈亚秋.一种间歇过程异常数据剔除的主元分析方法[J].计算机工程与应用,2003,39(29):222-224,230. |
| |
| 作者姓名： | 刘强 陈亚秋 |
| |
| 作者单位： | 浙江大学化工系,杭州,310027 |
| |
| 基金项目： | 国家自然科学基金项目:复杂化学分类模式的特征提取与分类器构造方法研究(编号:69975017) |
| |
| 摘 要： | 提出一种用于间歇生产过程中异常数据控制的方法。这种方法将原始的三维间歇生产数据集合展开成一个二维数据矩阵,进行中心化和规格化后再转化成另一个按照时间序列排列的二维数据矩阵。这种方法可以克服Wold方法在对数据进行中心化时引起的原始信息失真问题。通过对聚合反应釜过程数据进行分析,表明该方法能有效地对生产数据剔除异常。
|
| 关 键 词： | 间歇过程 主元分析 |
| 文章编号： | 1002-8331-(2003)29-0222-03 |
A PCA Method for Excluding Outliers in Batch Process Data |
| |
| Liu Qiang Chen Yaqiu.A PCA Method for Excluding Outliers in Batch Process Data[J].Computer Engineering and Applications,2003,39(29):222-224,230. |
| |
| Authors: | Liu Qiang Chen Yaqiu |
| |
| Abstract: | In this paper,an new approach for excluding outliers in batch process data is presented.The method uses historical batch process data set by arranging the original three-way matrix X into an unfolding two-way matrix,on which mean centering and auto-scaling is operated.Then this matrix is refold to the other two-way matrix,which is consistent with the natural data collection sequence.The new method is developed to overcome the distortion of original information caused by improper mean centering in Wold's method(1998).A case from industrial polymerization processes illustrated that it can efficiently exclude outliers in original batch process data. |
| |
| Keywords: | batch processes PCA(principal component analysis) |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
