Subspace decomposition approach of fault deviations and its application to fault reconstruction

In the present work, a new subspace decomposition approach of fault deviation is developed for fault diagnosis via reconstruction for principle component analysis (PCA) based monitoring system. It focuses on decomposing the fault effects in different monitoring subspaces, principal subspace (PCS) and residual subspace (RS), and finding the significant fault deviations that are responsible for the concerned alarming monitoring statistic. This is achieved by analyzing the relative changes of fault process in comparison with the normal status and designing a two-step feature decomposition procedure. Each fault data space is decomposed into two different parts for the purpose of fault reconstruction in PCS and RS of monitoring system respectively. One is composed of the concerned fault deviations that contribute to alarming monitoring statistics which are thus significant to remove the out-of-control signals. The other is composed of general variations that are deemed to follow normal rules or insignificant to remove alarming monitoring statistics. Its feasibility and performance are illustrated with experimental data.