Regulating the imbalance for the container relocation problem: A deep reinforcement learning approach

Yin Tang, Zengjian Ye, Yongjian Chen, Jie Lu, Shuqiang Huang, Jian Zhang

Abstract: Highlights•We study a 3D container relocation problem with stochastic container retrieval dates.•We propose a prediction-optimization approach to address the irregular state space.•We predict retrieval dates based on historical data by support vector regression.•We use predictions as inputs to train our relocation policy by reinforcement learning.•Our policy outperforms the other reinforcement learning policies without predictions.