DQN and MCTS-based inter-box multi-field bridge dynamic scheduling method

Abstract

The invention discloses a DQN and MCTS-based inter-container multi-field bridge dynamic scheduling method. The method comprises the steps: training a multi-field bridge dynamic scheduling model through a DQN to obtain a deep Q network which can make an optimal decision-making capability under random interference, guiding the MCTS through the deep Q network to carry out online decision-making of an inter-container multi-field bridge of a storage yard, and giving a next scheduling decision of the field bridge. According to the method, a multi-field-bridge dynamic scheduling model autonomously explores the yard environment through a reinforcement learning method, a large number of learning samples are generated, the problem that real field-bridge scheduling samples are lacked and difficult to obtain is solved, meanwhile, the MCTS and the DQN are combined to solve the multi-field-bridge dynamic scheduling problem between box intervals, the optimal scheduling decision of each step is given online, and the robustness of the decision scheme under uncertain interference is improved.

Description

technical field [0001] The invention relates to the field of scheduling of loading and unloading equipment in container terminals, in particular to a dynamic scheduling method for multi-field bridges in container intervals based on DQN and MCTS. Background technique [0002] The container throughput is increasing year by year, and the large-scale and high-speed container ships pose a higher challenge to the loading and unloading efficiency of the terminal. At present, the key to port operations has shifted from the shore to the storage yard. As the core loading and unloading equipment of the storage yard, the yard bridge is the core loading and unloading equipment, and its scheduling efficiency is related to the port's operating costs and operating cycles. [0003] The main content of the yard bridge operation in the yard includes the stacking operation and unloading operation of the container task. Under the given yard scheduling resources and container task information, d...

AI Technical Summary

Problems solved by technology

[0006] In view of the time-consuming meta-heuristic processing of disturbances in the prior art, the poor stability of the scheduling scheme due to frequent adjustments after rescheduling, and the limitations of traditional reinforcement learning in the face of large-scale state space input, the present invention provides a method based on DQN and MCTS Dynamic Scheduling Method for Multi-field Bridges Between Containers

Images