Study On Prediction Of Water Consumption And Inverse Analysis Of Node Demand

As we know, water distribution modeling is becoming more and more important for construction of digital city and modernized management of water supply system nowadays. Even though the theory has been developed maturely, there are still many obstacles on practical use. Water supply network is a sophisticated and dynamic system. On practical application of advance of freshet analysis, experience is playing a very important role. The uncertainty of node demand is an existing difficulty in modeling. The method is described in the paper to resolve this problem. To the unsolved problem of calibrating model, inverse analysis of node demand for water supply network is discussed detailedly in the paper. There are a few aspects discussed in this paper.1. Study on the prediction of water consumption: The paper summarizes the original method of water prediction. A new Kalman filter model with two parameter vectors is proposed by improving the original model. The model is applied to the prediction of water consumption in Hangzhou. The result shows that, the prediction based on improved model has the advantage of high precision. The seasonal model proposed in this paper is a feasible and effective model of forecasting water consumption.2. To go around the problem where the hourly water consumption contains seasonality, seasonal parameter is proposed. In this paper the time-vary ARMA model is used to simulate the hourly water consumption. The result shows that, the prediction based on time-vary ARMA model has the advantage of high precision.3. An integrated research frame of node demand of water supply network is provided. The inverse analysis of node demand includes existence of solution, uniqueness of solution parameter identifiability, algorithm of solver, stability and evaluation of solution.4. Utilizing generalized inverse matrix theory can make for the searchorientation of optimize, which will improve efficiency greatly. During the simulation, when Wiggins method is integrated into optimize algorithm, noise can be suppressed effectively.5. The model is applied to the water supply network in Hangzhou. The result shows that, the prediction based on generalized inverse matrix theory has the advantage of high precision, and it can meet the requirement of real project. At the same time, a new method about how to evaluate the digital model is proposed in this paper.