Research On Optimal Schedule Of Water Distribution Network Based On Smart Monitoring System

With the ease-implementation and high accuracy of pressure monitoring sensors, pressure data can be gathered more systematically instead of gathering information from only worst points. However, the utilization of these data is very low. Operational scheduling of the network is still on manual experience. To meet the requirement of real-time and intelligent scheduling, this paper studied the optimization technology of water distribution networkso that a new optimization method based on real-time pressure data can be established.Through analyzing basic hydraulic theory, this paper proposed the concept of digital network model which calculate the network pressure by interpolation.The robustness of the digital network model was tested by experiments under different conditions. By doing this way, the problem of real-time simulation was solved.Aoptimal sensor placement is the basis of digital model. The objective function of optimization problem is the minimization of simulated errors. PSO algorithm is used to solve the problem. Considering the topology of a real network, we proposed a theory-experience method to place the sensors in the network. By doing this way, the accuracy of the digital network model is guaranteed, with the calculated error less than 3m.According to the periodic feature of pressure data, a SVM-based prediction model is established to predict the pressure of monitoring node. The parameters of SVM model are also discussed in the paper. Simulation experiment is done after theoretic analysis.The results proves that the SVM model can forecast the tendency of pressure at the monitoring nodes accurately. In terms of relative error,85% of total predicted results are less than 5%.Also, this paper proposed a pipe roughness coefficient calibration method by clustering the pipes with similar features of material, ages, diameter and length. Thus a micro-model can be constructed. Based on the predicted state and hydraulic theory, nodal demand can be inversed. The inversed demand coincides well with the actual values with the relative error of total demand 5.1%.According to the pressure and demand data, an direct optimization problem is proposed and solved by PSO algorithm. To test the feasibility, the proposed method is applied to a real network. The pressure decreased a lot after optimization withthe change of the pressure only 1.13 m.The results show that the water pressure has decreased after optimization which is helpful to loss control.