Reinforcement Learning Based Multi-Agent Cooperation for Water Price Forecasting Decision Support System

Water price forecasting is a very important and difficult decision-making issue of any water resource decision support systems. Because there are many factors that would affect the price of water resource, there aren't any accurate models, which can be used in the water price forecasting. In this paper, an intelligent decision support system based on multi-agent is set up and applied to water price forecasting. An improved reinforcement learning approach is proposed to realize the multi-agent cooperation in the intelligent decision support system. In the proposed approach, a dimension reduction algorithm is used, which make the proposed approach have some good performances, such as the higher learning speed, the better convergence performance, and the better practicability. The experimental results show that the proposed approach can deal with the water price forecasting correctly and quickly.