← Back to Issue 123 contents Diamond Tools

Back propagation neural network-based prediction model of marble surface quality cut by diamond wire saw

Published in Diamante A&T Issue 123 · March 2026 · page 53 · Leggi in italiano

by Hui Dong1,2, Fan Cui1, Zhipu Huo1, Yufei Gao1,3,4
1 Key Laboratory of High Efficiency and Mechanical Manufacture of Ministry of Education, Shandong University, Jinan, China
2 School of Construction Machinery, Shandong Jiaotong University, Jinan, China
3 State Key Laboratory of Advanced Equipment and Technology for Metal Forming, Shandong University, Jinan, China
4 Shandong Key Laboratory of High Performance Tools and System, Shandong University, Jinan, China
Diamante A&T Issue 123 cover – March 2026

Abstract

Marble is widely used in the field of construction and home decoration because of its high strength, high hardness and good wear resistance. Diamond wire sawing has been applied in marble cutting in industry due to its features such as low material loss, high cutting accuracy and low noise. The sawing surface quality directly affects the subsequent processing efficiency and economic benefit of marble products. The surface quality is affected by multiple parameters such as process parameters and workpiece sizes, making it difficult to accurately predict through traditional empirical equations or linear models. To improve prediction accuracy, this paper proposes a prediction model based on backpropagation (BP) neural network. Firstly, through the experiments of sawing marbles with the diamond wire saw, the datasets of surface roughness and waviness under different process parameters were obtained. Secondly, a BP neural network model was established, and the mixed-strategy-improved whale optimization algorithm (IWOA) was used to optimize the initial weight and threshold of the network, and established the IWOA-BP neural network model. Finally, the performance of the model was verified by comparison with the traditional models. The results showed that the IWOA-BP neural network model demonstrated the optimal prediction performance in both the surface roughness Ra and waviness Wa. The minimum predicted values of the root mean square error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE) were 0.0342%, 0.0284% and 1.5614%, respectively, which proved that the model had higher prediction accuracy. This study provides experimental basis and technical support for the prediction of the surface quality of marble material cut by diamond wire saw.

Keywords: marble, diamond wire sawing, surface quality, BP neural network, improved whale optimization algorithm.

Original article: H. Dong, F. Cui, Z. Huo, Y. Gao, Micromachines 2025, 16, 971, MDPI, Basel – doi.org/10.3390/mi16090971. Licensed under CC BY 4.0. Adapted by the editorial staff of Diamante A&T.

Further reading

More from Issue 123

← Back to Issue 123 contents