Study On The Mid And Long Term Hydrological Forecast Of The Longyangxia Reservoir

The Longyangxia reservoir is the first one of the Yellow river upper stream reservoirs in cascade. It plays an important role in aspects of flood-control, drought prevention and electric power generation etc. Because the mid and long term hydrological forecast of the reservoir has not been completed, the reservoir management is always in a passive situation when arranging flood operation, electric power generation etc.Aiming at the actual existing problems and according to the hydrological and geographic characteristics of the reservoir, the basic factors influencing the runoff are determined in this paper from the point of view of the physical cause and statistics analysis through the analysis on the mechanism of the runoff generation and afflux. Using chaotic dynamic system analysis, the maximum possible forecast time dimension is determined. Using multivariate linear regression analysis and artificial neural network forecast method, the Longyangxia reservoir inflow is forecasted .The purpose is to establish a mid and long term forecast model that can be used in practice and has higher dependability and to offer worthy consult basis for the reservoir flood prevention and electric power generation. The main research results are as follows:(1) Broadly referring to related domestic and international materials, the present condition of the domestic and international research of the mid and long term hydrology forecast and forecast means are generalized. Based on this, the research method, that is genetic analysis and chaotic dynamic system analysis method, is determined according to the characteristic of the river basin and the status of the hydrology data.(2) The factors, which influence the Longyangxia reservoir inflow, are analyzed and researched systemically by means of physical cause analysis and statistics analysis. The results show that the rainfall is the main factor influencing the Longyangxia reservoir inflow in high flow period and while the recession of the river basin and river network in low flow period. Inside the forecast time interval the influence of the rainfall distribution on the runoff forecast is small.(3) The chaotic dynamic analysis method is introduced into the hydrology forecast. Through the analysis on the chaotic dynamic characteristic of the time series of the Longyangxia reservoir inflow, the maximum possible forecast time dimension of the inflow time series and the minimum imbedding dimension of the series are determined. These offer a scientific basis for the determinations of the forecast period of the mid and long term forecast model and the number of forecasters..(4) Making use of the main influence factors determined, models respectively based on multivariate linear regression analysis method and artificial neural network forecast method are established inside the range of the maximum possible forecast time dimension and the minimum imbedding dimension. The checking of forecast and the contract analysis on the forecast results of different models are executed. The results show that the effect of the runoff forecast is good and can entirely reach the need of practical uses. The improvement of the forecast scheme is expected.Researches above indicate that the genetic analysis and chaotic characteristic analysis on the Longyangxia inflow are rational and consistent with the practical situation of the runoff. The mid and long term forecast model based on the time series analysis has a good forecast effect. It can be used in the practice forecast. The results attained can offer worthy consult basis not only for the operation of the Longyangxia reservoir and the power station, but also for the joint dispatch of the Yellow river upper stream reservoirs in cascade.