Principal component analysis for feature extraction and NN pattern recognition in sensor monitoring of chip form during turning

Abstract

Experimental cutting tests on C45 carbon steel turning were performed for sensor fusion based monitoring of chip form through cutting force components and radial displacement measurement. A Principal Component Analysis algorithm was implemented to extract characteristic features from acquired sensor signals. A pattern recognition decision making support system was performed by inputting the extracted features into feed-forward back-propagation neural networks aimed at single chip form classification and favourable/unfavourable chip type identification. Different neural network training algorithms were adopted and a comparison was proposed.