Research On The Mid-long Term Runoff Forecast And Optimized Operation Of Hydroelectric Plant In Power Market

With the development of electrical innovation, the generation competition of individual power enterprises is an inevitable trend, which holds far-reaching effect on the management of hydroelectric plants. In order to fully utilize water resources and promote economic benefits of hydroelectric plants, it is of great theoretical and practical importance to study the optimized operation of hydroelectric plant in power market.In this paper, subjects mainly focused are as follows:To meet with the requirement of forward business and the establishment of mid-long term generation planning of hydroelectric plant, the yearly runoff and the monthly runoff are studied in this paper, providing many kinds models that suited to min-long term runoff forecast, including the time series analysis, the nearest neighbor bootstrap regressive model, the grey topological model, recession curve model, threshold auto regression, mean generating function, and ANN model etc. The forecast result proves that these models are useful.To improve the forecast accuracy, several methods are adopted to analyze the trend portion, the periodic portion and the jumping portion; meanwhile, the effect caused by people's activity is also taken into account.In this paper, mean generating function is applied to the forecast of monthly runoff in a creative way. The forecast result proved that the mean generating function is a feasible method in monthly runoff forecast.From the point of forward business, a mid-long optimized operation model aiming at the maximal benefit of hydroelectric plant is provided. In this model, it accounts not only the runoff but also the different price in different season. When this model is used to cascaded reservoirs on Longxiriver, result of simulating operation is good.Considering amendatory forecast runoff value will affect the operation strategy, a step-by-step decision method is proposed, which is useful for manager's making an optimized decision.The mid-long runoff forecast system of Longxiriver is developed with the C++ language.