A nonlinear process monitoring strategy for a Metal Forming process

Abstract

The article deals with the development of a nonlinear statistical process monitoring strategy for a metal forming process. The developed strategy is based on Kernel Principal Component Analysis (KPCA) and has been adopted to address the issue of inherent nonlinearity amongst the measured characteristics. Fault Detection has been carried out by the employment of KPCA score based Hotelling T2 chart. The devised strategy was applied over a Rolling Mill Unit (RMU) which comes under Bulk Metal Forming process. Observations pertaining to the Process and Feedstock Characteristics (PFC) of the RMU have been considered for the analysis. The devised monitoring strategy has been found to be efficient in carrying out the fault detection.