Diagnostic feature extraction from stamping tonnage signals based on design of experiments

Diagnostic feature extraction with consideration of interactions between va riables is very important, but has been neglected in most diagnostic resear ch. In this paper, a new feature extraction methodology is developed to con sider variable interactions by using a fractional factorial design of exper iments (DOE). In this methodology, features are extra ted by using principa l component analysis (PCA) to represent variation patterns of tonnage signa ls. Regression analyses are performed to model the relationship between fea tures and process variables. Hierarchical classifiers and the cross-validat ion method are used for root-cause determination and diagnostic performance evaluation. A real-world example is used to illustrate the new methodology .