A virtual-metrology-based machining state conjecture system

This study proposes a virtual-metrology-based control system for conjecturing machining states and suggesting controller operating modes. The two machining state conjecture models were integrated to validate the virtual metrology results using a dual-stage featuring scheme and a two-phase modeling procedure. The featuring scheme is to extract significant features from collecting data for modeling and to minimize the modeling features by using a genetic algorithm. The modeling procedure adopts two quality indicators to evaluate model effectiveness. Two case studies indicate that the system achieves a mean MAPE of precision estimation less than 8.5% and suggests the controller with operating modes in 1 s.