Research On The Forecast Of Regional Soil-Water (Salt) Environment Based On Artificial Neural Networks Theory

To meet the need of water environment evaluation and management decision for planning and design in the water saving reconstruction project of large-scale irrigation district, this dissertation based on BP model and RBF model of artificial neural networks theory, which is important branch of artificial intelligence technology, to build the forecast models of regional soil-water (salt) environment. The reasonability and feasibility of ANN models which are applied in the predication of regional water environment are discussed and analyzed. Then, the ANN technology is systematically applied in the forecast and evaluation of regional soil-water (salt) environment based on the two different scales experimental zones (Jiefangzha irrigation area and Shahaoqu experimental area) of Hetao Irrigation district of Inner Mongolia. The main conclusions are as follows:Through the different structure and algorithm application of BP model in the predication of regional groundwater hydrology, the hidden layers number, learning rates, neuron number of hidden layer and training errors of BP model and Accelerated BP algorithm which influence the convergence effects and test results of model are compared each other. Some application technology related parameters of BP structure design are put forward. This will provide concrete reference for parameters values which mainly depend on the experiences. According to the selected BP model and related water table depth records and other information, the annual and monthly mean water table depth trends in the future planning year (2005, 2010) are forecasted on condition that water saving reconstruction projects are accomplished in the larger scale experimental zone (Jiefangzha) of Hetao Irrigation district. Then, the results of BP are compared with that of RBF model and regional water equilibrium method. This tests the reliability of forecast results. From the prediction results of several methods we have known that: the annual mean water level in the 2010 will decrease about 0.6m comparing with present average level. The decrease range will relatively significant. Furthermore, the fall range in the crop growing period will larger thannon-growing's. This will worthy of noticing for the influence of regional water environment in the irrigation district.Based on five years regional water-salt monitoring data and related hydrological and weather information in the smaller scale experiment zone (Shahaoqu), the regional water-soil (salt) environment regime including grounder water table depth, water quality, superficial layer (0-40cm) and middle layer (40-70cm) soil moisture and saline concentration are simulated, tested and predicted using BP model. At the same time, the RBF model is be used to calibrate the results of BP. The calibrated results are proved to be reasonable. We draw the conclusions as follows from prediction results: the total salinity of groundwater will slightly increase compared with present mean value, when the irrigation water amount is reduced 10% of present normal water quantity in the 2005. But, when the total water quantity is reduced 30% of present average amount in the 2010, the total salinity of groundwater will have significant increase in the different period (pre-summer irrigation, post-summer irrigation, pre-autumn irrigation and pre-freezing up of soil four periods). The water quality will be likely to deteriorate; in addition, the prediction models of soil moisture and saline concentration are established. The soil water moisture in the total tin-saturated layers will decrease with the irrigated water amount reduction in the 2005, 2010. But, the saline concentration will produce different results in the different layers. The saline concentration in the 0-40cm layer will produce desalination trend in the 2010. However, the saline concentration in the 40-70cm will be slightly accumulated. This indicates that soil salinity will transport to middle and deep layer from superficial layer. As a whole, the water-saving reconstruction project will be helpful f...