上海交通大学学报(自然版) ›› 2013, Vol. 47 ›› Issue (02): 284-289.

• 交通运输 • 上一篇    下一篇

舰船冷却水系统多智能体自治状态重构方法

 秦久峰, 曾凡明, 陈于涛   

  1. (海军工程大学 动力工程学院, 武汉 430033)
  • 收稿日期:2012-12-12 出版日期:2013-02-28 发布日期:2013-02-28
  • 基金资助:

    中国博士后科学基金(201150M1547)

Method of Multi-agent Autonomous State Reconfiguration of Shipboard Chilled Water System

 QIN  Jiu-Feng, ZENG  Fan-Ming, CHEN  Yu-Tao   

  1. (College of Power Engineering, Naval University of Engineering, Wuhan 430033, China)
  • Received:2012-12-12 Online:2013-02-28 Published:2013-02-28

摘要: 针对现代舰船运行环境的不确定性和动力平台集成管理智能化的需求,以舰船冷却水系统为研究对象,研究了其在不确定性条件下基于多智能体(Agent)自治决策模型的状态重构方法.通过将该系统的状态重构问题看作松耦合的马尔可夫决策过程(MDP),建立了其基于MDP的多Agent自治决策模型.使用遗传算法对决策空间进行全面搜索得到最优策略,并通过数值计算验证了其性能.在获得的最优策略的指导下,实现了冷却水系统的自动状态重构.该方法能够提高冷却水系统的自主决策水平和任务效率.  

关键词: 多智能体, 舰船冷却水系统, 自主决策, 状态重构, 马尔可夫决策过程, 遗传算法

Abstract: In consideration of the uncertainty of the modern ship environment and the requirements for the intelligence of integrated management of power platform, taking the shipboard chilled water system as the research object, the method of state reconfiguration based on the multi-agent autonomous decision making model with the existence of uncertainty was studied. The problem of state reconfiguration is formulated as a loosely coupled Markov decision process (MDP). Then the multi-agent autonomous decision making model based on MDP is established and the optimal policy is obtained by using genetic algorithm. Finally the performance of the optimal policy is examined in the numerical calculation and the autonomous state reconfiguration is realized by following the received policy. This method can increase the autonomous decision making level and enhance the mission effectiveness of shipboard chilled water system.

Key words: multi-agent, shipboard chilled water system, autonomous decision making, state reconfiguration, Markov decision process (MDP), genetic algorithm

中图分类号: