Backpropagation Neural Network-Based Prediction Model of Marble Surface Quality Cut by Diamond Wire Saw

At a Glance

Metadata	Details
Publication Date	2025-08-23
Journal	Micromachines
Authors	Dong Hui, Feng Cui, Zhipu Huo, Yufei Gao
Institutions	Shandong University, Shandong Jiaotong University
Analysis	Full AI Review Included

Executive Summary

Core Value: Established a highly accurate prediction model (IWOA-BP neural network) for marble surface quality (Roughness Ra and Waviness Wa) cut by diamond wire saws.
Methodology: Optimized the Backpropagation (BP) neural network using a hybrid-strategy Improved Whale Optimization Algorithm (IWOA) to enhance global search capability and avoid local extremum stagnation.
Performance (Ra): Achieved optimal prediction performance for Roughness (Ra) with a Root Mean Square Error (RMSE) of 0.0342 and Mean Absolute Percentage Error (MAPE) of 1.5614%.
Performance (Wa): Achieved optimal prediction performance for Waviness (Wa) with RMSE of 0.0570 and MAPE of 1.7028%, significantly outperforming traditional BP and other optimized models.
Model Structure: The optimal IWOA-BP network utilized a 3-6-1 structure (3 input nodes, 6 hidden nodes, 1 output node) demonstrating a high correlation coefficient (R > 0.99) across training and test sets.
Practical Impact: Provides reliable theoretical support and technical reference for the intelligent optimization and automated control of diamond wire sawing process parameters in industrial settings.

Technical Specifications

Parameter	Value	Unit	Context
Workpiece Material	Calcium Magnesium Carbonate (CaMg(CO₃)₂)	N/A	Experimental marble sample
Saw Wire Type	Nickel-plated diamond wire	N/A	Cutting tool specification
Saw Wire Core Diameter	220	µm	Tool specification
Abrasive Particle Size	25-35	µm	Tool specification
Particle Distribution Density	70-85	grits/mm²	Tool specification
Slice Thickness	2	mm	Output geometry
Workpiece Thickness (H)	20	mm	Sample dimension
Optimal BP Network Structure	3-6-1	N/A	Input (Vf, Vs, L) - Hidden - Output (Ra or Wa) layers
BP Learning Rate	0.01	N/A	Network training parameter
Ra Prediction RMSE (IWOA-BP)	0.0342	N/A	Optimal prediction error
Ra Prediction MAPE (IWOA-BP)	1.5614	%	Optimal prediction error
Wa Prediction RMSE (IWOA-BP)	0.0570	N/A	Optimal prediction error
Wa Prediction MAPE (IWOA-BP)	1.7028	%	Optimal prediction error

Key Methodologies

The study involved systematic experimentation and advanced neural network modeling:

Experimental Setup:
- Cutting was performed on marble workpieces using an SH300 diamond wire sawing machine.
- The cutting process involved the wire saw moving reciprocally at Wire Speed (Vs) while the workpiece was fed at a constant Feed Speed (Vf).
Parameter Variation and Data Generation:
- A comprehensive dataset (47 combinations) was generated by varying three key factors:
  - Feed Speed (Vf): 0.6 to 3.0 mm/min
  - Wire Speed (Vs): 600 to 1400 m/min
  - Sawing Length (L): 10 to 50 mm
- Experimental designs included single-factor, orthogonal, and Box-Behnken design (BBD) response surface methods.
Surface Quality Measurement:
- After cutting and cleaning, surface quality indicators—Roughness (Ra) and Waviness (Wa)—were measured.
- Measurements were taken at five random central positions on each slice using a VK-X200K laser confocal microscope, and the average was used as the characteristic index.
BP Neural Network Configuration:
- A three-layer BP neural network was selected.
- Input layer nodes: 3 (Vf, Vs, L). Output layer nodes: 1 (Ra or Wa).
- Optimal Hidden Layer Nodes: 6 (determined by minimizing RMSE).
- Activation Function: Sigmoid function was used for the hidden layer.
- Error Objective Function: Mean Square Error (MSE).
Improved Whale Optimization Algorithm (IWOA) Strategy:
- IWOA was used to optimize the initial weights and thresholds of the BP network.
- Initialization: Population initialized using sine chaotic mapping combined with quasi-reverse learning for enhanced diversity.
- Convergence Factor: An improved nonlinear convergence factor was introduced to better balance global search (early stage) and local development (later stage).
- Position Update: The original logarithmic spiral was replaced with the Archimedean spiral to expand the search range and reduce premature convergence.
- Adaptive Weighting: An adaptive weight (w) strategy (w_min = 0.4, w_max = 0.9) was applied to optimize the search process during development.
- Mutation: A random differential mutation strategy was used to disturb the population and prevent search stagnation.

Commercial Applications

Stone Processing and Quarrying: Direct application in optimizing diamond wire sawing parameters for marble, granite, and other decorative stone materials to ensure consistent, high-quality surface finish (Ra, Wa).
Precision Hard Material Cutting: The methodology is transferable to the precision cutting of other hard and brittle engineering materials, including silicon wafers, ceramics, and magnetic materials (e.g., NdFeB magnets).
Automated Quality Control (AQC): Integrating the IWOA-BP model into CNC systems allows for real-time prediction of surface quality based on current machine settings, enabling automated parameter adjustment and quality assurance.
Process Optimization and Cost Reduction: By accurately predicting surface quality, the model minimizes the need for repeated trial cutting and reduces material waste, leading to improved production efficiency and lower operational costs.
Tool Wear Management: The predictive capability supports optimizing cutting speeds and feeds to balance throughput against diamond wire wear, extending tool life while maintaining product quality standards.

View Original 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.

Tech Support

Original Source

DOI: https://doi.org/10.3390/mi16090971

References

2022 - The use of waste marble for cleaner production of structural concrete: A comprehensive experimental study [Crossref]
2023 - A review of dimension stone extraction methods [Crossref]
2025 - A review of silicosis and other silica-related diseases in the engineered stone countertop processing industry [Crossref]
2025 - An AHP and BEES model-based sustainability assessment of small-scale marble processing industries: Indian prospects [Crossref]
2025 - Effects of marble waste on shear strength and pullout resistance of sand reinforced with Glass Fiber-Reinforced Polymer (GFRP) bars
2025 - Mixed-mode fracture behavior of a marble exposed to acidic environments [Crossref]
2021 - The mechanics of sawing granite with diamond wire [Crossref]
2022 - Research progress of wire saw machining technology [Crossref]