强化学习在基于多主体模型决策支持系统中的应用--以湖泊水环境决策支持系统为例  

Reinforcement learning for DSS based on multi-agent model:A case of lake water environment DSS

作　　者：倪建军[1] 刘明华 任黎[2] 张传标[1] 

机构地区：[1]河海大学计算机与信息学院(常州),常州213022 [2]河海大学水文水资源学院,南京210098

出　　处：《系统工程理论与实践》2012年第8期1777-1783,共7页Systems Engineering-Theory & Practice

基　　金：河海大学常州校区创新基金(XZX/09B002-02);河海大学自然科学基金(2009423111)

摘　　要：利用研究复杂系统和多主体(multi-agent)建模的相关知识与方法,将湖泊水环境中的各种实体,如政府、排污企业以及各种水生生物等抽象为具有一定智能的主体,建立湖泊水环境智能决策支持系统.并将强化学习方法应用到智能决策支持系统中,实现湖泊水污染的智能预测与预警.最后,以太湖流域为应用背景,进行了初步的仿真实验,实验结果验证了该方法的有效性.The lake water environmental problem has been more and more serious. It is a very important subject to find a more effective way of water pollution control. In this paper, the lake water environ- ment decision support system （DSS） is set up, using the knowledge and methods of complex system and multi-agent modeling. The various entities in the lake water environment （such as government, polluting enterprise and a lot of aquatic organisms） are abstracted as the agents, which have some certain intelli- gence. A method based on reinforcement learning is proposed to achieve the intelligent prediction and warning of the lake water pollution. At last, a preliminary simulation experiment is conducted on the application of Taihu Lake basin. The experiment results show that the proposed method is effective.

关 键 词：强化学习 决策支持系统 多主体建模 水污染治理 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]