Study On The Application Of Runoff Mid-long T-erm Combination Forecasting Model

Mid-long term runoff forecast play a decisive role in ensuring the safety of people's lives and property,the safety of industrial and agricultural production,the development of water conservancy facilities and the use of water resources.Mid-long term runoff forcast have relatively high requirements for accuracy and forecasting in actual production and life.According to the changes of the hydrological factors themselves,it has the characteristics of long term changes,and the law of change is quite complex.At present,due to the limitations of objective cognition and the level of science and technology development,there are many Mid-long term runoff forecast models.The prediction accuracy an the stability of the forecast still cannot meet the actual application requirements of the production department.A large number of studies and practical application requirements of the production department.A large number of studies and practical applications show that any current Mid-long term runoff forecast model is not applicable to any other established watershed,and the characteristics of the watershed and characteristics of runoff and model applicability conditions are analyzed.Long-term forecast of runoff will have more engineering significance.This paper takes the Zhangjiashan monthly runoff as the research object,based on the sequence data from 1932 to 2001,combines the features of the basin and characteristies of runoff process,based on the single model forecast,and analyzes the GM(1,1)model respectively.The advantages and disadvantages of the BP neural network model,the wavelet analysis model,and the ARIMA model,and the applicable conditions,based on which the weight gcometric average,the weight arithmetic average,the induced ordered information set settlement sub-theory were introduced,and the combined forecasting model was established from different scales.Designed and developed a Mid-long term runoff forecast system based on a comprehensive integrated platform,presenting a typical application of the system.The main research content is as follows:(1)Inductive analysis of the Mid-long term runoff individual forecast model.The GM(1,1)model?the BP neural network model?the wavelet analysis model and the ARIMA model were established respectively from the Mid-long term forecast model of runoff.The basic theory and model parameters of each model were optimized for the Mid-long term runoff combination forecast model.The establishment provides the basis for the foundation.(2)Establish a Mid-long term runoff forecast model.Based on the characteristics of the single forecast model for Mid-long term runoff,a combined forecast process was established to analyze the effectiveness of the combined forecast method and its feasibility.A combination forecast model with changes in the weights was constructed as a key research object.A combined forecast model system based on runoff data was established.The relative accuracy of R-Bias and root mean squared error(R-RMSE)were used to test the model.The results show that the forecast accuracy and stability of the Mid-long term runoff combined model are better than that of the Mid-long term runoff single model.(3)Realize example applications of runoff Mid-long term forecast.Taking the runoff sequence data of Zhangjiashan from 1932 to 1999 as the data base of the application object,the forecasting results of four kinds of medium and long term runoff single forecast models and Mid-long term runoff combined forecast models were compared and analyzed.The results show that the forecast effect of the mid-long runoff combined forecasting model is better than that of the single forecasting model as a whole,while the forecasting result of the variable weighted optimal non-negative variable weighting factor combination forecast model is better than that of the weighted invariant combined forecast model result.(4)Designed and developed a Mid-long term runoff forecast system based on a comprehensive integrated platform.Starting from analysis of runoff forecast requirement.According to the system,the overall architecture?functions and database of the system are designed.Mid-long term runoff forecast system based on a comprehensive integrated platform is developed.The system can quickly perform man-machine interaction,modify parameters,and improve runoff forecast accuracy and real-time.