Intelligent Optimization of Slope Step Parameters in Open Pit Mines Containing Weak Interbedded Layers Based on Machine Learning and Multi-objective Optimization Methods

Tao Deng^1*rong Jing He¹wei Ji Sun²xing Shou Peng^2*Xin Pang²Tao Chen²qiang Xiao Zhang¹

• ¹Kunming University of Science and Technology School of Land and Resources Engineering, Kunming, China
• ²Pangang Group Mining Company Limited, Panzhihua, China

The stability of open-pit slopes with weak interlayers is a key concern in mining, as the proper design of bench parameters directly influences both slope safety and economic outcomes. Traditional optimization techniques are time-consuming and often fail to find global optimal solutions. In contrast, machine learning (ML) technologies, with their advanced data processing and pattern recognition capabilities, provide promising alternatives for optimizing bench parameters. This study introduces an intelligent optimization approach for open-pit bench design, combining machine learning and multi-objective optimization methods. A numerical model representing slopes with weak interlayers is developed, and a dataset linking bench control parameters with slope safety and economic performance is created. Machine learning algorithms are employed to develop predictive models that assess how different parameter combinations affect stability and economic performance, ultimately defining an optimized design space. The SVM-NSGAⅡ and SVM-BO algorithms are used for optimization analysis. Results indicate that SVM-NSGAⅡ predicts a safety factor of 1.258 with a 3.8% error, outperforming the 5.6% error of SVM-BO. For economic benefits, SVM-NSGAⅡ achieves a relative error of 1.2%, which is significantly lower than the 9.0% error observed in SVM-BO. Comparative analysis reveals that SVM-NSGAⅡ excels in solving high-dimensional, nonlinear optimization problems, producing parameter combinations that optimize both stability and economic efficiency. A software system for bench parameter optimization, built on the Python platform and utilizing the SVM-NSGAⅡ algorithm, was developed with an intuitive graphical user interface (GUI). This system significantly improves slope stability and mining profitability, offering scientific support for mine design in complex geological environments and showcasing both theoretical importance and practical application potential.