A neural network forecasting model of Beijing´s water supplied and consumed based on set pare analysis

By using neural networks, Beijing´s water supplied and consumed is forecasted, and connection number and total partial connection number of the set pair analysis (SPA) are obtained. Principal factors and development trend are sequenced based on absolute relative error between sample value x_i and predominant value y_i so as to set up a mathematics model of forecasting Beijing´s water supplied and consumed. The instance analysis shows that it´s a scientific and practicable method of system analysis with a higher forecasting accuracy.