Study On The Runoff Evolution Law And Forecast Model Of The Upper Yellow River

Water resource has always been one of the necessary conditions for human survival,and river is one of the main sources of water resources in China.This paper selects the hydrological data of two typical hydrological stations(Jimai station and Tangnaihai station)in the upper reaches of the Yellow River from 1961 to 2012,a total of 52 years and 624 months.On the basis of previous studies,this paper analyzes the basic mathematical statistical characteristics and evolution law of the Yellow River runoff,and constructs a prediction model to predict and analyze the monthly runoff series and annual runoff series in the source area of the Yellow River,so as to provide reference for the relevant departments to carry out water conservancy work in the future This paper provides a reference for the forecast work.The main research contents and achievements of this paper are as follows(1)Through the analysis of the basic statistical characteristics,the distribution characteristics and the annual variation rules of the runoff data in the source area of the Yellow River,the research shows that the runoff in the source area of the Yellow River mainly concentrates in the flood season,and generally,the annual distribution of the runoff of the Yellow River is uneven and the annual change is large.(2)In Runoff Trend analysis,the moving average method,m-k order test method and Spearman rank correlation test are used.The results show that the annual runoff of Jimai station does not increase significantly,while the annual runoff time series of Tangnaihai station shows a significant decrease trend.Combined with m-k mutation test method,sliding t test and ordered clustering method,the year 2006 and 1986 are determined as the runoff mutation years of Jimai station and Tangnaihai station respectively.The results show that Jimai and Tangnaihai stations in the upper reaches of the Yellow River have multi time scale characteristics.The small time scale changes are embedded in the large time scale cycle,and both have the first major cycle of 13 year time scale.(3)BP model of artificial neural network and GA-BP model optimized by genetic algorithm are established to predict the monthly runoff of two hydrological stations in the source area of the Yellow River.Among them,the BP model has a large error,while the improved GA-BP model has a faster speed and higher accuracy in runoff prediction.The calculated prediction accuracy is 79.17% and 75% respectively,and the accuracy reaches grade B(70%-85%),which meets the requirements,and can be used in hydrological forecast.(4)GM(1,1)model and R / s-gm(1,1)model optimized by R / S analysis method are used to forecast annual runoff.The results show that: for GM(1,1)model,the prediction accuracy of Tang Naihai and Jimai station is 69.48% and 66.49%,respectively,which do not meet the model accuracy requirements.The relative error of the combined R / s grey prediction is 18.10% and 11.68% respectively.Compared with the single grey prediction,the prediction accuracy of the combined R / s Grey Prediction in Tangnaihai and Jimai stations is improved by 12.42% and 21.83%,which reaches the accuracy required by the grey prediction,indicating that the model can be used to predict the annual runoff in the source area of the Yellow River.