摘要
在高动态、高风险舰面保障环境下,极易发生不确定性干扰事件,这使得构建具备快速响应能力的舰面重调度模型来提升大型舰船舰载机的出动能力变得至关重要。通过分析舰面保障调度任务特点,利用变作业窗的单步推进方式,构建多作业并行执行的舰面保障马尔科夫决策过程(MDP),并基于改进的深度Q值网络(Deep Q-Network,DQN)高效生成多机并行保障重调度方案。经过仿真试验验证,所提出的动态调度策略符合舰面动态调度优化需求。
Uncertain interference events are easily to occur in the highly dynamic,high-risk aircraft supporting environment on flight deck,which makes it critical to build a fast-response re-scheduling model to improve shipboard launch capability of large carrier.Through the analysis of task characteristics of shipboard supporting,a multi-task parallel execution Markov decision process(MDP)is constructed by using a single-step progression of variable operation window,and the multi-aircraft parallel scheduling schemes are generated based on the improved deep Q network(DQN).Through the verification of simulation experiments,it is shown that the proposed scheduling strategy satisfies the requirement of dynamic aircraft scheduling optimization.
作者
白天
罗永亮
刘敬
常智超
王泽
BAI Tian;LUO Yongliang;LIU Jing;CHANG Zhichao;WANG Ze(Systems Engineering Research Institute,CSSC,Beijing 100094,China)
出处
《船舶工程》
CSCD
北大核心
2021年第S02期117-123,共7页
Ship Engineering
基金
国防科技创新特区资助项目(No.18-163-11-ZT-004-030-01)
关键词
深度Q值网络
强化学习
舰面保障
动态调度
deep Q network
reinforcement learning
aircraft support
dynamic scheduling
作者简介
白天(1995-),男,工程师,研究方向:动态调度、建模仿真、人工智能等。
