Municipal water consumption modelling by artificial neural networks

Municipal water demand forecasting is the first and one of the most important steps of planning water recourses. Determination of the types and dimensions of the required infrastructure that are used for finding, collecting and transmitting the water resources to consumers would depend on the forecasted water demand. If we take into consideration the costs of investments required and the importance of safe water supply for the people, how reliable the forecasts should be could be understood more clearly. In our country, water demand forecast is made by using a simple formula based on exponential population forecast according to the codes of Iller Bank. However, it is necessary to use more comprehensive data and methods to obtain reliable results because of the importance of this subject and the abundance of related parameters. To this end, artificial intelligence methods could offer alternative ways for forecasting water demands, which could lead the most effective use of water resources. The aim of this study is to forecast of municipal water demand by using Artificial Neural Networks (ANN). Monthly water usage of zmir have been modeled by ANN, taking into account a series of econometric (population, water bill, number of household, gross domestic product) and climatic (temperature, precipitation and moisture) parameters. Using the model constructed, water demand forecasts for next 20 years have been made.