基于主元分析（PCA）的显著误差检测与校正及应用

在风粉浓度软测量的测量数据中出现显著误差,将会严重恶化测量数据品质,破坏数据统计特性,导致软测量失败,因此显著误差检验和校正是误差处理的首要任务。本文讨论了基于主元分析(PCA)的显著误差检测与校正原理,运用Q统计方法,结合贡献图对某电厂热风送粉系统风粉浓度软测量中可能出现的显著误差进行了仿真分析,结果表明,基于PCA主元分析的显著误差检验和校正方法在风粉浓度软测量工业应用中可行。

Gross Error Detection and Proofreading Concentration Based on PCA In the Measurement of Pulverized-Coal Concentration

If errors appear in the soft-sensing of the pulverized-coal concentration., it would seriously worsen the quality of measurement data and destroy the character of data statistics, which leads to the failure of the soft-sensing . Therefore error detection and error proofreading are the primary task of dealing with errors. The theory of error detection and error proofreading based on the principal components analysis (PCA) are discussed in the pulverized-coal concentration. The simulation analysis combined with the contribution charts is made by means of Q statistical methods to check the possible errors. The results show that PCA is an effective approach to error detection and error proofreading in the industry of the soft-sensing measurement in the pulverized-coal concentration.