Machine Learning-Based Optimization of Diamond Burnishing Parameters in Terms of Energy Efficiency and Quality Indexes

Diamond burnishing is a prominent solution to boost surface properties with cost savings. In this study, a cold air-cryogenic CO2 internal diamond burnishing is developed, in which the distance to the nozzle (D), the inlet pressure of the compressed air (P), and CO2 flow rate (Q) are optimized. Minimizing the average roughness over a measurement area (Sa) as well as energy consumed (Eb) and improving power factor (PF) as well as Vickers hardness (VH) are primary considerations. The artificial neural network (ANN) and the entropy method are applied to present the burnishing responses and compute the weights. The modified gray wolf optimizer (MQWO) and TOPSIS are utilized to produce a set of solutions and determine the best solution. As a result, the optimal D, P, and Q were 15 mm, 3.0 Bar, and 12.0 L/min, respectively. The Sa and Eb were reduced by 34.1 and 1.5, respectively, while the PF and VH were improved by 13.2 and 9.5, at the selected solution. The developed burnishing process could be utilized to enhance the surface characteristics of internal holes. The optimal data could help to improve quality indicators and reduce energy consumed for the practical diamond burnishing. The ANN-MQWO could be used to present the nonlinear data and obtain optimal global results, as compared to the NSGA-II. The work is expected as a significant contribution to make the diamond burnishing process greener and more efficient. © King Fahd University of Petroleum & Minerals 2024.