Application Of Neural Network Method On Water Requirement Forecasting In The Irrgation Area

With the rapid development of economy, water supplyment has become one of the most sensitive and complex key problems in the current society. Healthy water resource systems provide survival basis for human reproduction and development. However, spoliatory developing and consuming patterns in natural resources, as well as the extensive mode of social development and economic growth, bring gradually increasing pressure to the water resource systems. In China, water shortage in many areas has become a serious obstacle to the development of economy and society, directly affecting China's sustainable economic and social development, especially in Tarim River Basin in Xinjiang. In this paper, the theory and analysis method of water requirement forecasting in the irrigation area associated with Tie Gan Li irrigation area is presented, the present reaearch aims to the sustainable development and effective utilization of the water resources and the decreasement of the deference between the supply and demand of water. The paper mainly covers the following aspects:1. Expounded the background and the significance of the water requirement forecasting in the irrigation area, previous research of the artificial neural networks, main research contents of the papper, ideas, objects as well as the technical approach of the project. Such efforts provide theortical support and guidance for the comprehensive planning of local water resource.2. Detailedly explained the research tool of the water requirement forecasting in the irrigation area, names BP neural network. The problems which should be attentioned in the application were also discussed. By means of implementing BP neural network in Matlab, some problems in practice can be solved. Matlab can be utilized as a powerful tool in discussing the examples in the next chaper.3. According to the following five factors, such as the population in the irrigation area, area of the tilth, GDP per person, area of the orchard, area of the plantation, the local water requirement was studied. The main influence factors was used as the input of artificial neural network(ANN), the water requirement of the irrigation was used as the output of artificial neural network(ANN), and then the ANN model about water requirement forecast was set up. A case study indicates that the average relative difference between the actual value and the function value of the water requirement was 1.19%(The maximum difference was 5.43%, and the minimum difference was 0.02%), demonstrating a good correlation of the model, the relative difference between the forecasted value and the actual value of the water requirement in 2006 was 1.66%. It seems that the model can be adopted in predicting the water requirement in the irrigation area. Based on SPSS software, the regression equation for the relationship between the year and influencing factors could be developed, the time-series could be forecasted and could be used as the input of ANN, finally, we could get the water requirement forecast on the base of ANN, so as to provide a guide for water requirement forecasting in the irrigation area.4. The main factors which affecting the water requirement forecasting in the irrigation area can be classified into three categories, namely, population factor, industry factor, and ecological factor. Each factor was discussed and the management method for water requirement forecasting in the irrigation area was suggested. Considering the shortages of water resources in the irrigation area, effective measures have to be adopted to strengthen the management of water resources, improve the efficiency of water utilization and reduce the water consumption per unit output, so as to achieve the sustainable development and utilization of water resources.