Research On Ultrashort-Term Wind Power Combination Forecasting Considering Meteorological Information And Frequency Characteristics

With the rapid development of the global economy,the continuous consumption of traditional fossil energy has not only triggered a worldwide energy crisis,but also caused a serious deterioration of the ecological environment.Because of the technological maturity and achievability,wind energy has received widespread attention as a replacement for clean energy.However,wind power has the features of fluctuating,intermittent and unstable.As the proportion of wind power installed capacity continues to increase,the impact of wind energy on the power system and the subsequent wind abandonment are becoming more and more obvious.Therefore,in order to suppress the influence of strong randomness of wind power,ensure the safe operation of the power grid,and improve the wind energy consumption rate,accurate prediction of wind power is particularly important.The ultrashort-term prediction of wind power is of great significance in increasing wind power consumption,reducing wind turbine regulation pressure and ensuring safe operation of the power grid.This paper takes the ultrashort-term wind power prediction as the research object and proposes a novel ultrashort-term wind power prediction model that takes account of meteorological information and seasonal trends,improves the data quality of the original wind power sequence,considers the coupling relationship between meteorological characteristics and wind power.reduces the complexity of a large amount of meteorological information,establishes a prediction model that memorizes long-term series,and extract the seasonal trend of wind power to correct the prediction results.Based on the above research,the hybrid model proposed further improves the prediction accuracy of ultra-short-term wind power.The specific work is as follows:The distribution and characteristics of various meteorological features were statistic,such as wind power,wind speed,and wind direction.The meteorological feature information that affects and reflect wind power were analyzed,and the integrity of the original wind power data was checked.The reasons and filtering methods of abnormal data are analyzed.The strong coupling effect of meteorological information and wind power was fully considered.The meteorological features were added to the input feature vector of the combined prediction model,and the principal component analysis algorithm was proposed to perform dimensionality reduction processing,in which a large number of meteorological features were converted into a few linearly independent comprehensive features.Under the premise,the modeling complexity and prediction time were significantly reduced.Aiming at the instability of wind power sequences,an improved empirical mode decomposition algorithm is proposed to decompose the original wind power series into several sub-feature components to weaken the volatility of wind power.Considering the different frequency characteristics of the components,different prediction models are used for independent prediction,and then the prediction values of all sub-feature components are reconstructed to obtain the final prediction result.Aiming at the problem that the low-frequency component has a large time span in each period and the problem of gradient disappearance and explosion in the modeling process,a long short term memory network was proposed to flexibly adapt to the timing properties.Aiming at the problems of complex changes of high-frequency component and slow model convergence,an extreme learning machine model with excellent generalization performance and fast convergence speed was proposed.In order to verify the effectiveness of the proposed model,this paper compared common wind power forecasting models and conducted experimental verification on three datasets collected from a wind farm in Laizhou,Shandong.Experiments showed that the combined model can effectively improve the prediction accuracy of wind power,and has certain practical application prospects.