基于DQN的二次供水系统运行优化研究

OPERATION OPTIMIZATION OF SECONDARY WATER SUPPLY SYSTEM BASED ON DQN

摘要: 二次供水系统是饮用水到达用户的最后关键环节，针对二次供水运行中水龄较长影响水质的问题，提出一种基于深度学习Q 学习算法（Deep Q-Learning Network，DQN）的运行优化模型。该模型将水压、水龄、能耗优化目标综合计算成对应的奖励，基于水力模拟的运行工况为输入，进水池、水泵的运行指令为输出。以某二次供水系统为例，利用EPANET软件构建水力模型，基于DQN分别对组件运行进行优化。结果显示，优化后均在保证供水压力的前提下达到降低水龄的目标。

Abstract: Secondary water supply system (SWSS) is the key process for water supply to reach the users' tap water. Aimed at the problem that water age in SWSS affects water quality, the optimized operation model based on deep Q-learning network (DQN) is proposed. The optimization objectives of water pressure, water age and energy consumption were calculated to the corresponding rewards. In the model, the inputs were the hydraulic state of SWSS, and the outputs were the operation instructions of the pool inlet and the pumps. Taking a SWSS in the residential community as an example, the hydraulic model was established by using the software EPANET. Based on DQN, the operation instructions of the pool and the pumps were optimized. The results show that on the premise of ensuring water pressure, the water age is reduced.