Abstract Prediction of surface roughness is always considered important in the manufacturing field. A product may require a particular roughness that may be specified by the designer for various reasons, either functional requirement or aesthetic appeal. While modern manufacturing systems and machines have always contributed towards better control of surface quality, better computational facilities and the availability of newer algorithms attract researchers to understand the prediction of quality in a better manner. In this paper, prediction of surface roughness by multiple regression analysis is presented. The predictors are cutting parameters, tool wear and the statistical parameters extracted from the vibration signals of a turning centre. The contribution of various statistical parameters in prediction of surface roughness is studied. A Machine learning approach using feature reduction using principle component analysis is attempted to achieve higher predictability and low computational effort.
Dr. Elangovan M., Dr. Sakthivel N.R., Saravanamurugan, S., Dr. Binoy B. Nair, and Sugumaran, V., “Machine Learning Approach to the Prediction of Surface Roughness Using Statistical Features of Vibration Signal Acquired in Turning”, in Procedia Computer Science, 2015, vol. 50, pp. 282–288.