Partial Least-Squares Regressive Analysis and Modeling for Tool Wear

The algorithm of partial least-squares regression(PLSR) is briefed firstly. The PLSR analysis is applied to the sample data sets of cutting tool wear under different machining conditions. Six independent variables for modeling including cutting speed V, cutting force components F_x, F_y and F_z, as well as force ratios F_y/F_x and F_z/F_x, are screened from eight original variables based upon the variable important projection and the factor loading. The model with the six independent variables and the flank wear of cutting tool as the dependent variable is built up by using PLSR approach. Two sample data sets, one under the cutting conditions covered in the modeling data and the other under new different cutting conditions, are used to verify the model respectively. The results demonstrate that the variable screening is reasonable and the satisfied values of the flank wear of cutting tools can be obtained from the PLSR model.