摘要: 为提高软测量的模型精度,剔除建模数据中的过失误差,提出采用Bagging-PCA方法进行误差侦破。利用Bagging算法的集成思想,改善单变量大误差对经典PCA的影响,提高算法稳定性,实现数据的过失误差侦破。用该方法对丙烯浓度的软测量进行过失误差侦破,取得了良好的效果。
关键词:
Bagging-PCA方法,
软测量,
过失误差侦破
Abstract: o improve the precision of soft sensing model, a method of Bagging-PCA is proposed to detect gross error. Using the integrated idea of Bagging algorithm, it improves the stability of classical PCA and detects gross errors effectively by weakening the effect of the big errors in part variable. The method is used to detect gross errors in modeling data for a propylene concentration soft sensing and good results are obtained.
Key words:
Bagging-PCA method,
soft sensing,
gross error detection
中图分类号:
胡云苹, 赵英凯, 李丽娟. 基于改进PCA的软测量数据过失误差侦破[J]. 计算机工程, 2010, 36(18): 282-284.
HU Yun-Peng, DIAO Yang-Kai, LI Li-Juan. Gross Error Detection of Soft Sensing Data Based on Improved PCA[J]. Computer Engineering, 2010, 36(18): 282-284.