A Novel Design Concept of Cemented Paste Backfill (CPB) Materials: Biobjective Optimization Approach by Applying an...

Area of Science:

• Mining Engineering
• Materials Science
• Computational Intelligence

Background:

• Uniaxial compressive strength (UCS) is critical for cemented paste backfill (CPB) safety in stope construction.
• CPB cost is a significant factor in overall mining expenses.
• Existing design methods lack integrated UCS and cost optimization.

Purpose of the Study:

• To develop a biobjective optimization approach for CPB design.
• To integrate UCS prediction and cost modeling for improved mine safety and economy.
• To address the limitations of current CPB design methodologies.

Main Methods:

• Construction of an evolved random forest (RF) model, enhanced by the beetle search algorithm (BAS), for UCS prediction.
• Development of a mathematical cost model based on the linear relationship between CPB components and costs.
• Application of the weighted sum method for biobjective optimization of UCS and cost.

Main Results:

• The evolved RF model demonstrated high accuracy in predicting CPB UCS.
• The biobjective optimization approach successfully generated Pareto front optimal solutions.
• Optimal solutions consistently offered improved UCS or reduced cost compared to the actual dataset.

Conclusions:

• The proposed hybrid machine learning model effectively predicts CPB UCS.
• The biobjective optimization approach provides a viable strategy for balancing CPB UCS and cost.
• This method enhances CPB design for safer and more cost-effective mining.