The Application Of Artificial Neural Network In The Flood Forecasting

Hydrological system is not only an open, complex giant system,but also subject to the influence of climate and human activities, showing characteristics of complex behavior, so it is a nonlinear complex system. Due to the complexity of hydrologic phenomenon, it is difficult to use content method of hydrologic phenomenon (especially in the mid-to-long-term hydrological forecasting) to describe at present.There are many traditional method analysis to describe the mid-to-long-term hydrological forecasting that they are hydrology statistical, time series analysis method, etc. But the traditional forecasting methods mainly exist in the following problems, such as through the methods of mathematical statistics parameters describing hydrological process is too coarse; Each method has its own lack of limit, so it cannot achieve favorable forecast purpose when It is used alone;all the traditional theory and methods of need explicit function, but these functions often be obtained, or only approximate expressions to describe, so cause bigger errors. The mid-to-long-term hydrological forecasting existed prediction accuracy of lower due to the above causes, so it cannot guide effectively the production practice, etc.To solve the above problems in hydrological forecasting, this article read a lot of hydrological forecasting of domestic and foreign literature, from artificial neural network (Ann) in the study of hydrological forecast and combined with the actual situation of fengfeng mine area to do the flood forecasting and formulate rational planning of the city.Combined with the actual situation of fengfeng mine area, the aticle analysis comprehensivly of the natural conditions in a region, meteorological hydrology, river system, and the social economy, etc., and analyzes emphatically the fengfeng flood situation and existing problems.The focus of this research is that the artificial neural network is applied to flood forecasting and forecast the features for the value of Cv, Cs/Cv of hydrological sequences eigenvalue. The hydrological sequence are divided into the training sample, validation sample and prediction samples and use the training and validation samples to determine the model samples. Using the determined network model to use the predict the sample to predicted Cv, Cs/Cv, combining the P-â…¢curve, for flood-related calculations. According to the calculation results, we formulate relevant flood control planning scheme, flood control planning measures combined the actual situation of urban fengfeng mine area. To ensure that promote engineering flood control function fully foundation, established all sorts of non-engineering measures especially in the flood cases.The conclusion shows that using artificial neural network to the flood forecast enriched the contents of hydrological forecast and the research results has practical value.