Combined Use of Parameters Identification and Principal Component Analysis in Quality and Process Monitoring. Application to Cutting Machine of Wire Rolling Process

摘要

We consider in this paper a combined use of Principal Component Analysis (PCA) and Recursive Least square (RLS) for quality and process monitoring. PCA approach operates generally on the system inputs - outputs which are affected by the noise; this also affects the sensitivity of Q statistic indicator. To overcome this inconvenient, we consider a coupling between RLS and PCA algorithms, i.e. the PCA computing procedure operates on the estimated parameters data but not on the inputs - outputs. Comparatively to the conventional PCA, this approach is particularly efficient in dynamic regime. Using the process parameters behavior as inputs variables in the PCA computing procedure improve the detect ability by reducing the wrong status generating by the noise effects. Application on different cutting sequences in wire rolling process shows that the combined method can be easily extended to qualify the cutting operations.