Study Of Real-time Scheduling Of Integrated Valves And Pumps In Water Distribution System

In recent years,the massive waste has been caused by serious energy consumption in water distribution system.In order to solve this phenomenon,many countries have processed a large number of technology experiments of saving energy.It is essential to save energy in China.Therefore,a kind of pumps and valves integrated scheduling is presented in this paper,which can save energy and cost of waterworks departments.A case of city T is employed to prove the precision of this method.Firstly,deep belief network combined with symlets time series model(SW-DBN)is presented for predicting the hour water usages of water plant in city T using MATLAB program.The results show that the relative errors meet the requirement where is less than 5%at time 1-8,10%at time 9-10 and 20%at time 11-12.Compared with DBN,this model has a better precision.Secondly,Convolution Neural Network model(CNN)is used to predicted delivery pressures.In this model,SCADA pressures monitors in pipes and water usages in water plant are regarded as independent variables.The historical pump delivery pressures are regarded as dependent variables.A pressures predicted model is built by training of CNN and the new hours pressures are predicted.The relative errors of this model meet the requirement where is less than 5%compared predicted with real values.Thirdly,parallel pumps parameters are proofread until the sum of pumps flows equaling to the real flows using MATLAB program.The fitness of curve is finally fitted.The results show that.the max of ?b?k are 6.30 and 2.54×10-5,meeting the requirement.Fourthly,introducing leakages and electricity cost as objective function,a kind of real-time pumps scheduling model is presented using MATLAB combined with Epanet and Csharp programming.And it can be solved by cuckoo search(CS).Next,using fuzzy c mean(FCM)cluster model,water distribution system is divided into different zones by the opening degrees of valves matrix.Finally,introducing opening degrees of throttle check valve(TCV)as factor and leakage costs as objective function,a kind of real-time TCVs scheduling model is presented based on Cuckoo search,which is used for each time scheduling that can balance pressures of network and reduce leakage volumes.According to the case study in water distribution system of city T,the total of leakage cost is 45938.59 CNY after optimization which cut cost by 5.04%than only pumps scheduling,2.06%than only valves scheduling and 5.67%than before optimization.