Abstract:

More than half of the direct greenhouse gas (GHG) emissions in mining operations are attributed to fuel consumption by haulage fleets. Conventional dispatching approaches in the mining industry that rely on heuristics or manual dispatching often cause long idle and waiting times in the fleet that waste resources and increase GHG emissions. Recent technological advancements in artificial intelligence and autonomous driving show promising applications in smart fleet management systems. However, existing studies have not investigated the GHG reduction potential of the AI-powered dispatching system in mining scenarios. In this research, we show that a smarter fleet has the potential to significantly reduce GHG emissions at mine sites by intelligently choosing routes and optimizing waiting times while achieving the same production levels. Specifically, a smart dispatching system is developed using deep Q-learning (DQN) to manage autonomous haul trucks at a hypothetical mine site. Deep learning algorithms improve fleet efficiency and reduce emissions by training each haul truck to determine the route based on loaded material, road traffic, estimated queueing, and maintenance needs. The GHG reduction potential was investigated by calculating and comparing the emissions of different dispatching approaches. The proposed approach also shows advantages in determining the optimal fleet size and production strategy to better balance productivity and emission targets. This study demonstrates that smart fleet management systems can bring both economic and environmental benefits to mining operations and contribute to the green mining transition alongside other parties in society to tackle climate change.

Menu

Dr. Da Huo

Biography