Tool wear condition monitoring by combining spindle motor current signal analysis and machined surface image processing

For improving online tool condition monitoring (TCM) reliability and convenience, a novel TCM method based on current signal analysis and machined surface image processing is presented. In this paper, the correlation mechanism of spindle motor current, machined surface image, and the tool wear condition during cutting process was analyzed in detail. The theoretical analysis results show that the reliability of TCM can be improved by the fusion of the above two information. Cutting experiments in multiple conditions had been performed on the FANUC vertical computerized numerical control. The FANUC SERVO GUIDE software was used for obtaining the current data in built-in current sensor from machine tool, and an industrial CCD camera was used in the collection of machined surface image, which is convenient and feasible. Time domain analysis, frequency domain analysis, and variational mode decomposition (VMD) were performed on current signals for sensitive feature extraction closely related to tool wear; rotation-invariant uniform local binary patterns (RULBP) and_Gray-Level_o-occurrence Matrix (GLCM) were performed on machined surface image for sensitive feature extraction that was in close contact with tool wear. The obtained sensitive features were fed into random forest (RF) to establish the correlation between sensitive features and tool wear condition. According to the experimental result, in contrast to available TCM methods, the proposed method shows significantly better performance in improving the accuracy of tool wear condition recognition.