Implementing PCA Based on Fault Detection System Based on Selected Important Variables for Continuous Process

摘要

Multivariate Statistical Process Control (MSPC) is known generally as an upgraded technique, from which, it was emerged as a result of reformation in conventional Statistical Process Control (SPC) method where MSPC technique has been widely used for fault detection and diagnosis. Currently, contribution plots are used in MSPC method as basic tools for fault diagnosis. This plot does not exactly diagnose the fault but it just provides greater insight into possible causes and thereby narrow down the search. Therefore, this research is conducted to introduce a new approach and method for detecting and diagnosing fault via correlation technique. The correlation coefficient is determined using multivariate analysis techniques that could use less number of newly formed variables to represent the original data variations without losing significant information, namely Principal Component Analysis (PCA). In order to solve these problems, the objective of this research is to develop new approaches, which can improve the performance of the present conventional MSPC methods. The new approaches have been developed, the Outline Analysis Approach for examining the distribution of Principal Component Analysis (PCA) score. The result from the conventional method and ne approach were compared based on their accuracy and sensitivity. Based on the results of the study, the new approaches generally performed better compared to the conventional approaches.