Prognosis of urban water consumption using hybrid fuzzy algorithms

For the optimal operation of waterworks it is necessary to predict the expected water consumption of the following days as accurately as possible. However, there are no conventional methods to predict the water demand. In this paper a prediction model based on hybrid fuzzy algorithms is introduced. The software automatically creates a fuzzy rule system out of a training database using the so-called VISIT (Variable Input Spread Inference Training) algorithm. A fuzzy neural network (FNN) system is created. Rules are trained with back propagation (BP) and least squares estimate (LSE) methods. The parameters of the algorithm are optimized with a simple genetic algorithm. As a result, one gets a rule system that delivers higher accuracy than a common statistically based model. Calculations and results are presented in this paper.