The Calculation And Predication Of Large-scale Domain Irrigation Water Demand

With the development of the economy and the society, the conflict between the provision and the demand for the water resource will become more and more serious,which cause a series of social problems and has affected the healthy development of the economic and society, and has become an important factor of restricting the economic and society development. Accurately predicting large-scale domain irrigation water use plays an important role in programming water resource scientifically and resolving the conflict between agriculture and industry and other water demanding department. The calculation and prediction of large-scale domain irrigation water demand is mainly studied in the dissertation.Theories and scientific research on the irrigation water requirement and the forecasting methods in china and other countries are discussed and reviewed comprehensively in this study. The advantages and drawbacks of common methods and their application conditions were elementary compared and analyzed in the dissertation.Influence factors of large-scale domain irrigation water use were filtered, and some important influence factors were determined, such as precipitation, evaporation and atmospheric temperature. Multivariate linear regression model with different periods of time was established. The different periods of time, factors and series of data on the predicting precision of irrigation water use were compared.Regional planting structure of farm crops, which is the precondition of accurately predicting large-scale domain irrigation water demand is difficult to acquire. By generalizing and predigesting the irrigation schedule, a conceptual model was established. Suitable condition of the annual amount and sharing-time prediction model, and different series of data on the precision of predicting the irrigation water use were analyzed.Artificial neural network (BP network) model to predict the irrigation water demand was established, self learning ability of artificial neural net was simulated through slide prediction, and a comparative study of different input layer neuron was carried on, the best number of hidden node of different months was optimized, and the influence of prediction precision on irrigation water use that adopted assimilation number was analyzed.