The Research Of An Artificial Neural Networks’ Application In Optimal Scheduling Of Water Supply

The goal of Optimal Scheduling of Water Supply is to satisfy the water supply pressure, flow and water quality requirements in order to reduce the direct and indirect costs of production as much as possible. Meantime, safety, stability of production and transport processes are also improved. Pipe network modeling is an effective way to forecast the dynamic conditions of the pipe network for water supply. It’s helpful for the scientific and modern management of the pipe network as well as the scheduling optimization of water supply systems. It’s also helpful to realize water supply on demand and reduce losses and leakage. The microscopic model is the traditional method for modeling of water supply networks. However, there are some shortcomings in the practical application, so this model cannot be fully effective used in the optimal scheduling of water supply.At first, this paper built the prediction model of the city water supply system by BP ANN. Using the previous24hour water supply data as input, the model could computed the next quantity in the next hour, which can be used as the parameter of the optimal model. Utilizing the macro-modeling method, the model of water supply networks was built. The flow of waterworks and pump stations were used as input and the output were the pressure of measure point as well as waterworks and pump stations. By BP ANN, we can get the model of water supply networks. With the history data, it can be proved that the result by our models about were accurate.Given the two models above, the optimal function was built. The goal of the optimization was to minimize the cost of water supply, such as power consumption. At the same time, the total flow and pressure should be satisfied. Heuristic algorithm was used to solve this problem. The problem was first transformed to a unconstrained problem and then solved by Genetic Algorithms (GA).In the end, utilizing the history data, the test result shows that our method could reduce the cost while satisfying the pressure and flow of water supply. Therefore, the proposed optimal model can adjust the scheme of water supply, and provide theoretical guidance for the scheduling. This model can be combined with SCADA in future in order to make a more efficient automatic scheduling system of water supply.