Real-time scheduling method for unmanned mining trucks in open-pit mines considering dynamic demand

Most existing real-time scheduling methods for open-pit mines estimate the expected waiting time of mining trucks to be scheduled by considering only the influence of trucks en route on loading and unloading points, while ignoring the influence of subsequent trucks to be scheduled on current scheduling decisions, resulting in a lack of global foresight. Meanwhile, most methods lack systematic rescheduling strategies for emergencies such as mining truck failures. To address these problems, a real-time scheduling method for unmanned mining trucks in open-pit mines considering dynamic demand was proposed. This method calculated the arrival times of subsequent trucks to be scheduled at loading and unloading points and simultaneously assigned destinations to the current truck to be scheduled and subsequent trucks that could affect its decision. The expected waiting time of trucks to be scheduled was calculated according to the differentiated service rules of loading points, ore unloading points, and waste unloading points. A mining truck failure state determination rule was established based on the Poisson distribution, and when a truck failed, a truck of the same type was dispatched from the parking lot as a replacement. A real-time scheduling model was constructed with the objectives of minimizing total expected waiting time, total travel distance, total travel cost, and total shovel completion rate deviation, while comprehensively considering constraints such as rated truck load, empty and loaded speeds, loading and unloading times, and fuel consumption. Simulation results based on an actual mining area road network showed that, compared with the multi-objective model, shortest-queue model, and random model, the proposed model achieved the highest mining area output, the lowest transportation cost per ton of ore, and the highest shovel utilization rate. The average truck waiting time under the proposed model was slightly higher than that of the shortest-queue model overall, but significantly lower than those of the multi-objective model and random model. Under truck failure conditions, the output of the comparison models decreased to varying degrees, whereas the output of the proposed model remained basically stable. As the proportion of small trucks increased, the transportation cost per ton of ore and average truck waiting time both decreased, while mining area output first increased and then decreased. Therefore, the proportion of small trucks in the fleet can be appropriately increased when economic benefits are pursued.