Hierarchical Clustering of Surface Roughness Using Acoustic Emission Signals

Abstract

The interaction among cutting parameters during the turning process is complex and non-linear, hence making linear predicting methods unsuitable for use. This study is a presentation of hierarchical clustering of surface roughness using acoustic emission signals during single point diamond turning of RSA-443. The intention of the study is to establish if clusters generated by portioning algorithms can be reliably used to understand the internal structure of data. Acquisition of acoustic emission signals has been achieved by a Kistler Piezotron sensor. AE root mean square, prominent frequency and peak rate are extracted from the processed captured AE signals while surface roughness is physically measured using a Kistler Taylor Hobson Profilometer. Validation of the generated clusters has been achieved by using the Purity measure. The computed purity is 1, which is an exhibition of the high quality of the hierarchical clustering result. Hence, clustering can be reliably utilized to understand the internal structure of surface roughness data during single point diamond turning.