Tool Condition Monitoring in Turning by Applying Machine Vision

In this paper, a method for predicting progressive tool flank wear using extracted features from turned surface images has been proposed. Acquired turned surface images are analyzed by using texture analyses, viz., gray level co-occurrence matrix (GLCM), Voronoi tessellation (VT), and discrete wavelet transform (DWT) based methods to obtain information about waviness, feed marks, and roughness from machined surface images for describing tool flank wear. Two features from each texture analyses are extracted and fed into support vector machine (SVM) based regression models for predicting progressive tool flank wear. Mean correlation coefficient between the measured and predicted tool flank wear is found as 0.991.