Predictive Modelling of Surface Roughness in Grinding Operations Using Machine Learning Techniques

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F61989100%3A27230%2F25%3A10258038" target="_blank" >RIV/61989100:27230/25:10258038 - isvavai.cz</a>

Výsledek na webu

<a href="https://www.webofscience.com/wos/woscc/full-record/WOS:001484847600002" target="_blank" >https://www.webofscience.com/wos/woscc/full-record/WOS:001484847600002</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.21062/mft.2025.006" target="_blank" >10.21062/mft.2025.006</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Predictive Modelling of Surface Roughness in Grinding Operations Using Machine Learning Techniques

Popis výsledku v původním jazyce

This paper details a systematic machine learning workflow designed for the prediction of surface roughness in grinding operations using key machining parameters. Those parameters are Depth of Cut, Feed Rate, Work Speed, and Wheel Speed. The model was trained and validated on a data set which comprised experimental measurements of those parameters and their corresponding values of surface roughness. Three machine learning models, Random Forest, Gradient Boosting, and LightGBM, were developed and evaluated based on accuracy of prediction of the surface roughness. The validation of all three models was performed using performance metrics like Mean Squared Error (MSE), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and R-squared (R2). Among the models, LightGBM exhibited the highest value of performance with the lowest error observed MSE 0.0047, MAE 0.064, and RMSE 0.09 respectively, while an R-squared value closest to zero. (-0.02). The moderate performance was shown by the Random Forest which presented an MSE of 0.0063, MAE of 0.085, and RMSE of 0.10, while the Gradient Boosting recorded the highest error rates which may indicate that it is the least effective model. It is an effective application of machine learning in predicting surface roughness and gives an insight into machining process optimization through predictive modelling.

Název v anglickém jazyce

Predictive Modelling of Surface Roughness in Grinding Operations Using Machine Learning Techniques

Popis výsledku anglicky

This paper details a systematic machine learning workflow designed for the prediction of surface roughness in grinding operations using key machining parameters. Those parameters are Depth of Cut, Feed Rate, Work Speed, and Wheel Speed. The model was trained and validated on a data set which comprised experimental measurements of those parameters and their corresponding values of surface roughness. Three machine learning models, Random Forest, Gradient Boosting, and LightGBM, were developed and evaluated based on accuracy of prediction of the surface roughness. The validation of all three models was performed using performance metrics like Mean Squared Error (MSE), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and R-squared (R2). Among the models, LightGBM exhibited the highest value of performance with the lowest error observed MSE 0.0047, MAE 0.064, and RMSE 0.09 respectively, while an R-squared value closest to zero. (-0.02). The moderate performance was shown by the Random Forest which presented an MSE of 0.0063, MAE of 0.085, and RMSE of 0.10, while the Gradient Boosting recorded the highest error rates which may indicate that it is the least effective model. It is an effective application of machine learning in predicting surface roughness and gives an insight into machining process optimization through predictive modelling.

Druh

J_imp - Článek v periodiku v databázi Web of Science

CEP obor

—

OECD FORD obor

20300 - Mechanical engineering

Projekt

—

Návaznosti

S - Specificky vyzkum na vysokych skolach

Rok uplatnění

2025

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název periodika

Manufacturing Technology

ISSN

1213-2489

e-ISSN

2787-9402

Svazek periodika

25

Číslo periodika v rámci svazku

1

Stát vydavatele periodika

CZ - Česká republika

Počet stran výsledku

151

Strana od-do

nestránkováno

Kód UT WoS článku

001484847600002

EID výsledku v databázi Scopus

—