Study On The Intelligence Methods Of The Regional Water Demand Prediction

The regional water demand prediction supplies some important basic data for water security management, which is the basis of the water supply planning. The regional water demand prediction system is a complex system, which is affected not only by the gross of regional water resources, but also by the regional Socio-economic development, the per capita living standard, the water supply establishment, the water supply price, the water transfer, and so on. For a long time, the regional water supply has been one of the most important factors that restrict the economic development, and with the development of economy, the enlargement of regional scale, the contradiction between water supply and demand in area will be more outstanding. Therefore, the regional water demand prediction can provide scientific basis for the water resources reasonable allocation guiding macroscopically the regional water supply planning and the water resources management.In this paper, the grey system model and BP neural network are established based on the introduction of each method of the regional water demand prediction, besides, the two models are improved. The existing parameters and the added parameter in the GM (1,1) are optimized by accelerating genetic algorithm; the training data in BP neural network are pretreated by the principal components analysis, for the urpose of improving the forecasting precision. The result of the examples show that the improved methods are reasonable. Because the single forecast method has certain limitations, the combining forecast model for the water demand prediction in area used set pair analysis was presented in this paper, the forecast precision of each single forecast model was estimated by qualitative and quantitative analysis based on identical, discrepant and contrary degrees using the connection degree formula of set pair analysis, and the weight of the every single forecast model was determined.