Improving Confidence in Evolutionary Mine Scheduling via Uncertainty Discounting
This work addresses uncertainty in mine planning for the mining industry, but it is incremental as it builds on existing evolutionary algorithms by incorporating uncertainty discounting.
The paper tackled the problem of mine scheduling under geological uncertainty by introducing a method that discounts profits based on uncertainty within an evolutionary algorithm, resulting in improved downside risk over an ensemble of models, as demonstrated using Maptek's Evolution software.
Mine planning is a complex task that involves many uncertainties. During early stage feasibility, available mineral resources can only be estimated based on limited sampling of ore grades from sparse drilling, leading to large uncertainty in under-sampled parts of the deposit. Planning the extraction schedule of ore over the life of a mine is crucial for its economic viability. We introduce a new approach for determining an "optimal schedule under uncertainty" that provides probabilistic bounds on the profits obtained in each period. This treatment of uncertainty within an economic framework reduces previously difficult-to-use models of variability into actionable insights. The new method discounts profits based on uncertainty within an evolutionary algorithm, sacrificing economic optimality of a single geological model for improving the downside risk over an ensemble of equally likely models. We provide experimental studies using Maptek's mine planning software Evolution. Our results show that our new approach is successful for effectively making use of uncertainty information in the mine planning process.