Estimation of surface roughness in a turning operation using industrial big data

Abstract

Surface roughness prediction in a turning process is of paramount importance. However, there is hardly any physics-based model that can predict it accurately. Recently, thanks to advancements in information technology, there are an ample amount of data in the industry. This article proposes a methodology to estimate surface roughness in turning based on industrial big data. An attempt has been made to extract and preserve the concise, useful information to reduce the burden on data storage. The proposed methodology predicts the lower, upper and most likely estimates of the surface roughness. A case study containing 35,000 datasets is simulated using a virtual lathe to demonstrate the efficacy of the methodology. The whole region of data is divided into 81 cells, and model fitting is carried out in each cell. The developed model based on industrial big data provides reasonable prediction of surface roughness.