Research On Ultra-Short-Term Wind Power Forecasting Based On Combined Model

With the continuous development of wind power technology,the scale of wind power and grid-connected wind farms continues to expand,and the proportion of power demand is also increasing.This trend makes the impact of wind power on the grid more and more obvious,in order to meet the demand for power supply.To ensure the stable operation of the power grid and the reliability of the power supply system,and to rea lize the effective control of the wind turbine,the prediction of the wind power becomes an indispensable link.Wind power is fluctuating and intermittent,but decomposition can effectively reduce the fluctuation of wind power series.Firstly,the principle of empirical mode decomposition,wavelet decomposition and variational mode decomposition are introduced and spectrum analysis is carried out.It can be seen from the analysis results that although the empirical mode decomposition method has a strong adaptive ability,the mode mixing phenomenon has a greater impact on the accuracy of the decomposition,and the prediction effect of the high-frequency component IMF1 is poor.The method of variational modal decomposition can not only avoid the phenomenon of mode aliasing,but also avoid the choice of wavelet basis function,so the experimental steps are reduced by using the method of variational mode decomposition,which can avoid the phenomenon of mode aliasing and avoid the choice of wavelet basis funct ion first,otherwise it will have a great influence on the experimental results.Secondly,the time series model is applied to the single-step prediction of ultra-short-term wind power.In the first step,the basic principle and prediction steps of the ARIMA model are introduced,and in the second step,the ARIMA model is applied to the single-step prediction of wind power,and the prediction accuracy is low.Then,a single-step wind power prediction method based on VMD-ARIMA-GARCH combined model is proposed.In the first step,the wind power series is decomposed by variational mode decomposition to reduce the non-stationary characteristics of the wind power series.In the second step,the ARIMA model is established for each component,and then the residual sequence test is carried out for each component,and the ARIMA-GARCH model is established for the component with heteroscedasticity.In the third step,the final prediction of wind power is obtained by superimposing the prediction results of each component.Finally,in order to solve the problem that too many decomposition layers lead to long prediction time and large amount of calculation,the fuzzy entropy theory is introduced.Because the empirical mode decomposition method has strong self-adaptive ability,the decomposition method is improved at the same time.The improved method of CEEMD is cited.In the first step,the wind power series is decomposed by complementary set empirical mode decomposition,and in the second step,the complexity of the decomposition result of CEEMD is evaluated,and the adjacent components with similar complexity are superimposed to form a new combined component.So as to reduce the prediction time and calculation;the third step of the new combination of components to establish the CEEMD-FE-ARIMA-GARCH combination model,so as to carry out ultra-short-term wind power multi-step prediction.At the same time,the interval forecasting is introduced,which is of great significance to the rational distribution and real-time dispatching of wind farms,and can effectively improve the economic benefits of wind farms.