Pump Scheduling Optimization in Water Distribution Networks Incorporated with Deep Neural Networks
作者单位：Department of Automation Shanghai Jiaotong University
会议名称：《第32届中国过程控制会议(CPCC2021)》
会议日期：2021年
学科分类：12[管理学] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 08[工学] 081104[工学-模式识别与智能系统] 0815[工学-水利工程] 0835[工学-软件工程] 0811[工学-控制科学与工程] 0812[工学-计算机科学与技术（可授工学、理学学位）]
关 键 词：Water Distribution Network Pump Scheduling Deep Neural Network Particle Swarm Optimization
摘 要：Pump scheduling optimization is an essential guarantee for the safety and stability of water supply in water distribution networks(WDNs).However,the existing pump scheduling optimization methods are typically based on artificial experience with no systematic guidance and the empirical policy is relatively simple and not adaptive which makes it hard to implement since the real-world WDN is becoming *** this article,we attempt to address this issue by exploring data-driven approaches for pump scheduling optimization while considering arbitrary topology and no need for a hydraulic ***,the deep neural network is used to obtain the steady-state model of the ***,we use an adaptive policy to construct the loss function and use particle swarm optimization to implement online *** perform simulation studies using a realworld network in Shanghai to evaluate the effectiveness of the steady-state model and the control policy.