Research On Models Of Mid-and-Long Term Runoff Forecasting And Integrated Application

Given the long-term runoff forecasts have predicted a long period features, correct and timely long-term runoff forecasting, scheduling beneficial for reservoir regulation and significant. Due to the long-term variation of hydrological factors are very complex, and cognitive abilities and limitations of objective scientific and technological level, although currently there are many long-term hydrological forecasting models, but its accuracy can not meet the demand forecast production sector. Numerous studies found that various existing models can not be applied to all river basin, so by analyzing the process of watershed runoff characteristics and runoff characteristics, and in accordance with applicable conditions of the model to select the model that makes sense.Based on the theoretical study and practical overview of domestic and international long-term hydrological forecasting model, a brief summary of the advantages and disadvantages of each model and the applicable conditions and select several commonly used model, through case studies carried out on the applicability of the model discussed, but also provides a valuable reference for instance choose a suitable long-term runoff forecasting models. The main findings are as follows:(1) Based on the current domestic and international long-term hydrological forecasting model, a brief summary of the advantages and disadvantages of commonly used models and application conditions.(2) By conventional methods of time series analysis of ARIMA model Huanren Reservoir and Biliuhe different periods runoff, Songhuaba reservoir dry season runoff, Tanghe reservoir dry season runoff forecast. First, the time series stationary test and treatment, and then, using SPSS to identify the model, parameter estimation, model checking, conducted ARIMA model prediction.. The results show that the model for the relatively stable hydrological processes less volatile effect is more ideal, such as dam reservoir dry season runoff and Songhua Tanghe reservoir dry season runoff, it has some promotional sense.(3) The use of fuzzy identification method to describe human experience and knowledge of everyone for water, Shenwo annual and annual runoff in reservoirs to predict bridge. First, determine the evolution of the hydrological cycle sequence, on the basis of forecasts looking to be close to the point where the cycle period; and then, a fuzzy recognition model, based on the hydrological cycle and the element values close to changing trends for long-term predictions. The results show that the model for periodic obvious effect is better hydrological processes, such as cyclical Dahuofang Reservoir annual runoff and annual runoff Shenwo has important reference value.(4) The use of BP neural network model Biliuhe flood runoff and the annual runoff everyone room to predict. First, to ensure that the model is operable and could be applied to 74 circulation characteristics of basic data, choose relatively easy to obtain and is a major factor as the model hydro-meteorological factors; then, using a neural network model JAVA long-term forecast,. The results show that, due to the Biliuhe Basin maritime climate, climate-sensitive area, it is a factor more closely selected physical contact, more reliable model, the forecast accuracy of the results even better.(5) The application of several common long runoff forecasting model applicability of this article, is an ideal long-term forecast accuracy of prediction of the actual project provides a basis, but also for its similar characteristics have long basin hydrological processes Runoff Forecast provides some reference value.