A study on deep reinforcement learning-based crane scheduling model for uncertainty tasks

Abstract Aiming at the crane scheduling problem for uncertainty tasks in multi-crane situation, this article proposes a deep reinforcement learning-based modeling method that is not dependent on mathematical planning and has certain generality. First, process integrated into learning framework which orbit space of transportation task environmental information intelligent agent. Second, interactive mode between algorithm environment adjusted to adapt combined model. Last, model constructed by optimizing reward discount factor, rate, function intensive mode. Testing carried out based practical one steelmaking workshop. Scheduling proposal generated all are completed within planned time, verifies feasibility Results show compared with manual plan, new reduces total completion time 11.52%, collision routes decreases 57.14%, negative distance shortens 55.26%. The high efficiency therefore verified.