基于强化学习的集装箱码头卡车调度策略研究  

Vehicle Scheduling in Container Terminal Based on Reinforcement Learning

作　　者：尚晶[1,2] 徐长生[1] 

机构地区：[1]武汉理工大学物流工程学院,武汉430063 [2]武汉科技大学学报编辑部,武汉430081

出　　处：《武汉理工大学学报》2011年第3期72-76,共5页Journal of Wuhan University of Technology

基　　金：国家自然科学基金(70801047);中国博士后科研基金(20090450769)

摘　　要：研究同时服务于装船和卸船作业的集卡全场调度策略,调度优化目标包括减少岸桥等待集卡的时间以及减少集卡的空载行程。提出了基于Q学习算法的集卡调度强化学习模型,对其系统状态、动作策略、报酬函数进行分析,并结合小脑模型关节控制器(CMAC)神经网络对Q函数进行泛化和逼近。仿真结果表明,与其他集卡调度策略相比,Q学习算法的优化效果比较明显,其在保证岸桥连续作业的同时,还能有效减少集卡的空载行程。Scheduling strategies of container vehicles which serve loading and unloading simultaneously were researched.Scheduling optimizing object was reducing the waiting time of quay cranes for vehicles and empty travel distance of vehicles.A reinforcement learning model based on Q-learning was designed,and its system states,action rules and reward function were defined.The Q-function was generalized and approximated using the cerebella-model-articulation-controller（CMAC） neuron network.Simulation results showed that Q-learning algorithm could optimize the vehicle scheduling strategies,which guaranteed the quay cranes＇ continuous working and reducing the empty travel distance of vehicles efficiently.

关 键 词：集装箱码头 强化学习 Q学习 集卡调度 小脑模型关节控制器 

分 类 号：U691[交通运输工程—港口、海岸及近海工程]