Aplikasi Analisis Komponen Utama Dalam Deteksi Gross Error Pada Pengukuran Multivariable
| Main Author: | Perpustakaan UGM, i-lib |
|---|---|
| Format: | Article NonPeerReviewed |
| Terbitan: |
[Yogyakarta] : Program Pascasarjana Universitas Gadjah Mada
, 2000
|
| Subjects: | |
| Online Access: |
https://repository.ugm.ac.id/20796/ http://i-lib.ugm.ac.id/jurnal/download.php?dataId=3653 |
Daftar Isi:
- Principle Component Analysis (PCA) is a tool of many multivariate statistical analysis based on a linear transformation from a set of correlated variables to a new set of uncorrelated variables. This transformation will lead to data reduction and interpretation via extraction of variance-covariance structure of p original variables into just k principle component. This k principle components, generally only small number of k, can explain as much information as there is in the p original variables. Automatic measurements made on the modern computer controlled chemical plant often result many inter-correlated variables. So, the need of a tool to analyze these variables (e.g. for error detecting) is real. The aim of this research is to promote Principle Component Analysis as a tool to detect gross error occurred in the chemical process network. The computer simulation of measurements made on the chemical process network is made to test the capability of PCA in detecting and identifying gross error. This paper shows that PCA is capable of detecting gross errors of small magnitudes and has substantial power to correctly identify the variables in error when the other methods (univariate methods) fail.