Research In Medium And Long-Term Forecasting Schemes Of Monthly Runoff Volume In Hongjiadu Hydropower Station

Along with the national economy fast development, the scarce energy has become the restriction economy development "the bottleneck". It appears especially importantly that water resources are used fully and the function of the hydropower station is maximum limit displayed. The medium-and-long-term runoff foresting is the foundation of long-term electricity generation plan. Increasing precision of medium-and-long-term foresting is the question which the production unit is urgent needs to solve, also is the important content of hydrologic forecast research.This dissertation unifies the medium-and-long-term forecasting task of the Hongjiadu hydropower station in the upstream of WuJiang, and thoroughly studies part of the medium-and-long-term hydrologic forecasting mathematical statistic methods. We take the monthly runoff volume of Hongjiadu hydropower station as the forecast object, and take increasing forecasting precision as the goal, use the non-linear methods of wavelet decomposition and artificial nerve network, with the traditional methods of time series analyzing, period mean superimposed method, stepwise regression, ridge regression, principal component regression and so on, establish the forecasting models of the monthly runoff volume of Hongjiadu hydropower station. And the combination forecasting model is built on the basic of comparing the result of all the models. Finally, the superior model is chosen out as the last forecasting result of the monthly runoff monthly runoff of Hongjiadu hydropower station.It is indicated that, the fitting and forecasting result of stepwise regression and principal component regression achieved the anticipated precision request, and may be used in the monthly runoff volume of Hongjiadu hydropower station, which provide the reliable basis for optimizing dispatch, maximum limit displaying its economic efficiency of hydropower station.