Monitoring semi-batch reactor using principal component analysis

This work is aimed for the detection of fault occurred in the semi-batch reactor, which treats chromium sludge, at high sludge flow rate. Multivariate Statistical Process Control (MSPC) techniques and Principal component analysis (PCA) were applied to the simulated data for monitoring. In this work, an attempt is made by employing Lanczos symmetric tridiagonalization means for the determination of largest principal components instead of classical methods. Using established PCA model from normal operating condition batches, semi-batch reactor is monitored for specified time period in online fashion. T2 statistic was computed for each sample in order to identify abnormal scenario. The results have shown that the fault is successfully detected.