Research Of Hydrological Simulation Based On Artificial Neural Networks

Along with the fast development of modern technology, the computer and communications are applied to the hydrological science widely used .At the same time ,the people improve themselves in the cause and effect of every hydrological phenomena factors by study deeply. The research of physical-based distributed and conceptual models are improved fast, so they can applied to every hydrological fields. Such as the rule of base hydrologic, the control of flood and drought diseases, water resource evaluation and water development utility , the protection of water environment and ecosystem, climate change, man's activity. Artificial Neural Networks-ANN is kind of artificial intelligence tool which is widely applied in the fields of time series analysis, pattern identification etc. The possibility of using ANN models to simulate the rainfall-runoff relationship of urban areas will be explored first.The Back Propagation model is the most representative in Artificial Neural Networks. The Back Propagation model appeared in the field of hydrological science lateness, but it's advantage and potential have opened up. In this application, BP Model will be set up to combine to produce and practice. The thesis presentation the technology of applied and the rub problems. The mostly content as follows:1 , It studies the mostly characteristic of BP model in hydrological forecast and analyse the presence problems;2, It studies some arithmetic of BP model and the problems of discipline parameter;3, It simulates the rainfall-runoff relationship of studied basin by BP model;4, It simulates the rainfall-runoff relationship of studied basin by XINANJIANG model;5, It simulates the rainfall-runoff relationship of studied basin by Sacramento model;6, The analysises of rationality and comparison are ultimately applied for the performance. Then we discuss the feasibility the precision the excellent and theshortcoming of BP model in hydrological forecast.