Faults detection and isolation based on PCA: an industrial reheating furnace case study

The fault detection and identification based upon multivariate statistical projection methods (such as principal component analysis, PCA) have attracted more and more interest in academic research and engineering practice. In this paper, PCA and statistical control chart (SCC) have been used to detect and isolate process operating faults on an industrial rolling mill reheating furnace. The Q statistic (also referred as squared prediction error, SPE) and Hotelling T² statistic are used calculating the control limits of SCC. The diagnosing results to single fault (fuel-gas pipe control valve failure or furnace temperature sensor failure alone) and multiple faults (control valve failure and temperature sensor failure simultaneously) are presented after establishing the operating PCA model. The simulation results indicate that the method is effective and available.