Study on Fractal Prediction Model of Urban Hourly Water Consumption

A fractal prediction model for urban hourly water consumption was put forward in this paper, which was based on fractal theory. Taking the hourly water consumption historical data of an urban in north China as an example, firstly, the rescaled range analysis method was used to compute the Hurst exponent H. The result shows that the Hurst exponent H was 0.9158, which proved that the urban hourly water consumption time series presents fractal characteristics. Secondly, based on fractal collage theory, the fractal interpolation algorithm was used to find the iterated function system whose attractor is close to the historical hourly water consumption data. Then, the fractal prediction model for urban hourly water consumption was established according to this iterated function system. Through application it found that the average prediction error was 1.8%, which achieved the precision needed and could be the decision support for urban water supply optimization operation.