Multivariate statistical process monitoring and control of machining process using principal component-based Hotelling T 2 charts: a machine vision approach

Machine vision offers image-based inspection and quality control. Principal component-based multivariate statistical process monitoring (MSPM) and control facilitates monitoring of production typically involves several quality characteristics with a single control chart that identifies and diagnoses faults by signal decomposition. The paper presents principal component-based MSPM and control of the machining process using machine vision for industrial components manufactured on conventional lathe machines. It involves extraction of critical dimensions and surface characteristics using image-processing techniques, data dimensionality reduction using principal component analysis (PCA), process monitoring, and control using principal components based Hotelling T2 chart. Fault diagnosis involves decomposition of T2 statistic into contribution by individual principal components and their combinations, identification of out-of-control scenarios using decision tree and their physical interpretation to detect possible causes of errors for further analysis and control. The approach potentially offers an industry-ready solution to automated, economic and 100% process monitoring and control.