Abstract
Optimizing mining operations simultaneously—from extraction to processing and delivery—is complex due to vast scales, nonlinear processes, and uncertainties in geological attributes and commodity prices. Traditional optimization methods often struggle computationally with this complexity, leading to inefficiencies, increased operational risks, and reduced profitability.
This work introduces a novel framework that integrates machine learning with advanced optimization to enhance mining complex optimization under uncertainty. By analyzing historical operational data, we train predictive models to learn connectivity patterns between mining blocks, effectively capturing the spatial distribution of ore blocks to be mined within the same period. This provides valuable insights into how different sections of the mine interact and influence overall performance. These patterns (probabilities) are incorporated into flow problems to generate initial production schedules that respect operational constraints and capture the inherent uncertainties of mining operations.
Unlike traditional warm-starting methods that rely on deterministic estimates, our probabilistic approach accommodates uncertainties in both the optimization problem and the predictive models, generating a distribution that conveys confidence levels rather than a single-point prediction.
Key benefits include (1) enhanced operational efficiency—by providing better initial solutions, our framework accelerates optimization, saving time and computational resources; (2) improved risk management—accounting for uncertainties leads to resilient plans that mitigate risks from variability in ore grades and market prices; and (3) increased flexibility—the method generates diverse solutions for different scenarios or a single risk-aware solution, allowing adaptation to changing conditions and strategies.
This framework contributes to the development of robust, reasoning, and responsible decision support systems for large-scale industrial applications under uncertainty. It supports the sustainable growth and competitiveness of the mining sector, particularly in meeting the demands of the global energy transition.