Optimal Scheduling Of Water Supply Pump Station Based On Calibration And Renewal Of Pumps' Characteristic Curves

Water-supply pump station,as an important part of urban water supply and drainage system,ensures the users and factories' daily needs of water in the city.However,the traditional scheduling method,which heavily relies on experience of the engineers,stratifies the needs of the corresponding service area while consuming large amount of energy and it suits the current resource-intensive society no more.Applying intelligent algorithm onto scheduling becomes more and more popular nowadays.It can not only meet the needs of water flows and pressure on the pipe network,but supplies the most energy-efficient combination of pumps to the engineers in the water-supply pump station.In order to improve the feasibility of the new scheduling method,here the scheduling method based on verification and updating of the pump characteristic curves is proposed.By re-drawing the characteristic curve of the pump,the optimized scheduling scheme can be independent of characteristic curves' shift caused by aging or other reasons of a pump so that the new scheme can be more suitable to the real working conditions.First,BP artificial neural network,RBF artificial neural network and improved real-coded genetic algorithm are used to fit the characteristic curve based on real data.To do so,the parameters in each method are first set with the help of the initial pump characteristic curves.Then the data of the pumps is simulated used to polyfit the characteristic curves of the pumps and through comparing the advantages and disadvantages of each method,they will play their roles respectively here.However,in view of the fact that the real data of the water pumps can only cover the high efficiency area of their whole working area,the confidence evaluation system for the polyfitted curves based on real is put forward here and the necessary conditions under which,these methods can be applied is also given.In the process of comparing the re-plotted water pump characteristic curves with the initial curves,the idea of the pump fault diagnosis based on the new pump characteristic curves is put forward.Three kinds of common pump faults are listed as examples in this paper.Then,how the optimal scheduling is realized is introduced by listing the specific steps of scheduling the pump units based on generic algorithm.And then a model of a real pump station in Shanghai is built.In order to get the scheduling scheme which meets all the constraints and needs in the real working conditions,the standard generic algorithm is developed here in this paper,through introducing adaptive genetic operators and adding constraints with penalty weights based on simulated annealing algorithm into the model.After applying both the standard generic algorithm(SGA)and the improved adaptive generic algorithm(IAGA)onto the model,it is found,that the scheduling scheme based on IAGA can converge to the optimal solution more quickly,and the result can be more in line with the constraints in the model.Finally,according to the actual needs of the real conditions,a series of energy-saving measures are further proposed: First,according to the real working condition,the impact of switching pumps brought to the pump station as well as the pipe network must be taken into consideration,so in order to avoid a frequent pump-switch condition,several solutions are given here based on the already mentioned model and scheduling method.Secondly,to investigate the effect of clean water tank on energy consumption of a water pump station,the relationship between the level of a clean water tank and the energy consumption of a water pump station is quantified and it is found out that,to save more energy,the water should be kept on high level as long as possible and according to the daily water flow rate of a certain clear water pool in Shanghai,the influencing factors and the reference water level is deduced at last.