Run Law Analysis And Prediction Of The Upper Reaches Of The Yellow River

The water is all biological reproduction of the most basic survival conditions, Human life is irreplaceable natural resources. The water resources is the most active of ecological environment, the most extensive influence factor, but also the human society sustainable development restriction factors, in people’s living and national economy has an important position. And runoff in the water resources system play a leading role, climate and human activities by the influence of factors, the variation in the complex, and the show nonlinear and randomicity, such as by the co mplex characteristics. Therefore, how to reasonable description of the runoff evolution, and the future an accurate prediction of runoff reasonable, so far are the domestic and foreign research problems and hot.Now commonly used runoff description and forecasting methods are generally hypothesis hydrological system is linear, but the hydrologic system is actually a nonlinear system, runoff is the hydrologic system in the typical nonstationary and nonlinear properties of the sequence. According to the characteristics of runoff, this paper introduces a kind of can handle the smooth, nonlinear signal method, namely Hilbert-huang transformation method. This method contained empirical mode decomposition method (empirical mode decomposition, EMD) and Hilbert transform two parts. EMD method’s main purpose is to get the order IMF component and the Hilbert transformation, get Hilbert spectrum. And then to the changing law of runoff is analyzed. This paper, based on runoff material of the Guide, Xiaochuan and Lanzhou hydrological stations, researched the change rule of the analysis and prediction.Although Hilbert-Huang transformation method can process the non stationary nonlinear problem, but there are also many problems need to be solved, including boundary effect is a very difficult issue. In the process of decomposition for EMD boundary problem, this paper puts forward the interception of data directly at both ends, mirror image method and extreme value point symmetry extension method. Compared to no end point that process and the results show that after the endpoint processing, the latter’s effect is the former a lot better. Compared to three methods, extreme value point symmetry extension method is a mirror image method in foundation, increased endpoint or judgment is extreme value point for symmetric point processing, the effect is better, so choose the method to deal with this paper of all the data endpoint effect. Calculation results show that the method to enhance the treatment effect, can restrain to spread in error.Using EMD method runoff series can be decomposited limited natural modal functions (Intrinsic Mode Function, the IMF), to which the analysis, can master the runoff and change rules. On the analysis of the characteristics of the IMF component based on the artificial neural network for modeling prediction. Neural network approach can be used for nonlinear function function to describe the runoff series. This paper discusses the BP and RBF neural network model and structure, learn the rules, and construct the prediction model of the runoff. USES the establishment two neural network model for prediction of the specific runoff series, the result is good. And two network carries on the comparison, all have their respective advantages and disadvantages. Results show that BP and RBF neural network model has good prediction ability.In this paper,using Huang Hilbert transformation and artificial neural network method, based on annual and monthly runoff variety rule analysis and forecast of the upper reaches of the Yellow River,Guide, Xiaochuan and lanzhou three stations, provides scientific basis for the regional water resources development and utilization of the planning and design.