Leakage Control Of First-level Division And Pumps Scheduling On Multi-source Water Supply Network Of City B

Water supply network system is one of the important infrastructures to ensure the normal operation of the town,which is related to people’s life,enterprise production and social development.It is one of the important carriers closely connected with water purification plant and users,and plays an important role in the storage,transportation and scheduling of drinking water.The leakage of pipe network is an important factor affecting the service level of water supply,especially with the expansion of the city,the increase of the number of water sources,the continuous extension of the pipe network,and the aging of the old urban pipe network.If the water division lacks management and scheduling planning,it will result in the uncoordinated distribution of water quantity among multiple water sources and the unbalanced distribution of pipe network pressure,which makes the leakage problem more serious.The loss of a large number of water resources in the process of transmission and distribution not only causes waste and is not conducive to the sustainable use of water resources,but also increases the cost of water production and sale,which restricts the rapid development of economy and society.As a common water supply system,how to improve the operation management level by dividing the water supply areas,coordinating the operation of the second pump station of each water plant,and managing the leakage has become an urgent problem.In view of the above problems,this paper analyzes the leakage control of multisource water network.Firstly,the dynamic hydraulic model of the pipe network is improved based on the short-term water demand prediction of the pipe network to provide guidance for the optimal operation;and the leakage phenomenon is accurately simulated by studying the pressure driven leakage theory.Then,the theory of first level partition and the theory of multi-objective optimal operation of pump group are studied,and the effect of static and dynamic adjustment on leakage control is discussed,which is applied in B city pipe network.The operation state of water supply network is not only related to static attributes(pipeline attributes and topology),but also affected by dynamic data,especially node water demand.This study analyzes the actual water consumption and water leakage of users,and adjusts the micro model from these two aspects to accurately simulate node water consumption and water leakage.In the aspect of actual water use simulation,DBSCAN-XGBoost combined forecasting method is proposed,which uses the robust spatial clustering algorithm to identify and remove the abnormal water use data of outlying distribution,with good recognition effect and fast removal speed.This study use XGBoost algorithm to predict the hourly water demand,and the decision coefficient can be kept above 0.9,which fully verifies the accuracy of the combined forecasting method.According to the prediction results,the node water demand is modified to realize the fine regulation of the actual water consumption of the micro model.In the aspect of water loss,in view of the lack of leakage simulation in the classical hydraulic model,this study uses water balance analysis to calculate the water loss,analyzes the pressure driven leakage theory,calculates the leakage coefficient of user nodes,and checks the pressure and flow data.According to the above method,the actual water consumption and leakage water consumption of users can be modified,which can accurately simulate the hydraulic state of the pipe network,and provide the basis for the further study of the first level zoning and optimal scheduling.The static topology adjustment is based on the first-level zoning of the pipe network.The existing research of water supply network zoning focuses on the bottomup zoning,and the first level zoning method is insufficient.Therefore,this paper applies the sensitivity analysis method of material concentration to the first-level zoning of water supply network,adjusts the material concentration of different water sources one by one,and selects the water source with the most sensitive node concentration.Compared with the minimum power dissipation path zoning method,sensitivity analysis can better reflect the actual hydraulic state effect.The actual water consumption in each first-level zoning is similar to the design water consumption,and a certain water supply capacity is reserved,so the performance of the first-level zoning is better.The dynamic adjustment is realized by the multi-objective optimization of the working condition of the pump set.A multi-objective optimization model with water leakage,power consumption and average water age as objective functions is established.The constraint conditions are transformed into penalty terms in different forms according to the soft and hard constraints.Two multi-objective evolutionary algorithms are used to solve the problem,namely,the second generation and the third generation of non-dominated sorting genetic algorithm(NSGA-II and NSGA-III).The multiobjective evolutionary evaluation index is used to evaluate the convergence,diversity and universality of solution set,and the technology for order preference by similarity to an ideal solution(TOPSIS)is used to analyze the advantages and disadvantages of solution set.The results show that although NSGA-III algorithm can reduce the diversity and universality of solution space partition,its convergence is stronger than NSGA-II,and its optimization ability is stronger.Compared with the optimization before and after the partition,it is found that the fitness of the solution set after the partition is greater than that of the solution set without the partition,which proves that the material concentration sensitive partition can effectively improve the operation of the pipe network.Taking city B as the application case of the above methods,the paper provides solutions to the problems of insufficient operation planning,uncoordinated multi water sources,high leakage rate and so on.In the aspect of short-term water demand prediction,XGBoost and ARIMA models are compared,path search and sensitivity analysis methods are compared in the first level zoning,multi-objective optimization model of pipe network before and after the zoning is adopted in multi-objective optimization,NSGA-II and NSGA-III solution models are adopted respectively,and finally the solution set of optimization algorithm is evaluated.The results show that the optimal operation scheme of the pump group can be obtained by NSGA-III solution of the pipe network after partition,which can not only reduce the power consumption and average water age of the pump station,but also save the power consumption of the pump station.The key point is to obtain a good control effect of water leakage.The leakage rate is reduced from the original 31.58% to 29.35%,reducing 2.23% and 8100.75 cubic meters.It has the practice of guiding the multi-source pipe network to control the leakage Value.