Mining operation decision support: intelligent truck scheduling for open-pit mines with modelling and optimization collaboration

Considering time-varying factors like weather and road conditions is critical for open-pit mine route planning, yet it has rarely been addressed in existing studies. This gap leaves transportation solutions unable to guide actual ore production, so data-driven truck scheduling optimisation for open-pit mines is developed in this study. First, a multi-objective truck routing optimisation model with time windows is presented according to a real application scenario. Subsequently, a data-driven optimisation method is introduced to construct a dynamic random forest (RF) surrogate model using truck trajectory data to assess the quality of the new solutions. Moreover, a novel multi-objective evolutionary algorithm is proposed to obtain a set of Pareto-optimal alternatives. It is composed of an improved multi-objective ant colony optimiser and a fitness mechanism, which can strike a balance between diversity and convergence of the solutions. The proposed algorithm is simulated for routing optimisation of an open-pit mine in China. Under the first decision objective, Optimal Truck Route Scheme (OTRS)’s LowerBound outperforms Multi-Objective Ant Colony Optimisation (MOACO) by 15%–40% in most instances and averages better than Non-dominated Sorting Genetic Algorithm II (NSGA-II) and Strength Pareto Evolutionary Algorithm 2 (SPEA2); under the second, its upper-lower bound range is approximately 3.7 times that of MOACO and about 4.27% larger than that of NSGA-II, while SPEA2 has outliers in many cases. The experimental results confirm that this approach has robust performance, with faster hauling efficiency and lower cost.