Energy Consumption Analysis Of Machining Centers Using Bayesian Analysis And Genetic Optimization

TL;DR

This study analyzes how machining parameters affect energy use and surface quality in CNC machining centers, using Bayesian regression and genetic algorithms to optimize energy efficiency while maintaining surface finish.

Contribution

It introduces a combined Bayesian regression and genetic optimization approach to identify optimal machining conditions for energy efficiency and surface quality.

Findings

Feed rate is the most influential factor on power consumption.

Depth of cut significantly affects surface roughness.

Optimal conditions achieve less than 0.03% predictive error.

Abstract

Responding to the current urgent need for low carbon emissions and high efficiency in manufacturing processes, the relationships between three different machining factors (depth of cut, feed rate, and spindle rate) on power consumption and surface finish (roughness) were analysed by applying a Bayesian seemingly unrelated regressions (SUR) model. For the analysis, an optimization criterion was established and minimized by using an optimization algorithm that combines evolutionary algorithm methods with a derivative-based (quasi-Newton) method to find the optimal conditions for energy consumption that obtains a good surface finish quality. A Bayesian ANOVA was also performed to identify the most important factors in terms of variance explanation of the observed outcomes. The data were obtained from a factorial experimental design performed in two computerized numerical control (CNC) vertical machining centers (Haas UMC-750 and Leadwell V-40iT). Some results from this study show that the feed rate is the most influential factor in power consumption, and the depth of cut is the factor with the stronger influence on roughness values. An optimal operational point is found for the three factors with a predictive error of less than 0.01% and 0.03% for the Leadwell V-40iT machine and the Haas UMC-750 machine, respectively.