Wind Speed Forecasting And Probabilistic Assessment Of The Output For Wind Farm

With the energy supply pressure increased,the development of renewable energy will be the trend of the times.Because of its mature technology and the reducing of the cost,wind power has become the most promising renewable energy sources.Wind speed forecasting(wind power prediction)plays an important role in the wind energy,accurate predictions on wind speed is not only beneficial to the scheduling department in a timely manner to adjust the scheduling plan,thereby reducing spinning reserve capacity and the operation cost of power system,but also provide some reference or advice for the grid planning.This paper first introduces the study status of wind speed forecasting and probabilistic assessment of the output for wind farm,and summarizes the commonly used methods for wind speed forecasting.What's more,reliability assessment for wind farm is also introduced.Based on cubic exponential smoothing method,a new self-adaptive and dynamic forecasting method is presented,also the multi-step prediction formula.The results of simulation and error analysis show that the new self-adaptive cubic exponential smoothing method is more accurate and efficient than traditional method.In this paper,a variable weight combination forecasting method based on golden section algorithm is proposed.The new combined forecasting method is on the basis of self-adaptive exponential smoothing method and gray model.The combined forecasting method updates raw data in a timely manner,and selects smoothing coefficient and weight coefficient by timely use of golden section algorithm.Simulation results show that the adaptive combination forecasting method based on golden section algorithm can effectively improve the accuracy of one-step wind speed forecasting,but in two-step wind speed forecasting,its forecasting result is not pleased,especially worse than a single exponential smoothing method.Taking wind farm output as the research object,the stochastic nature of wind speed,stochastic production,fault state and derating state on output are considered to set up theprobability model of multi-state output by means of the Monte-Carlo method.Then the evaluation process is given.Combined with actual example,algorithm of probabilistic assessment of multi-state output for wind farm is realized by MTALAB.Example analysis shows that the model can effectively assess the output for wind farm,which of course has better applicability than the traditional three state models.The study of this paper was based on the real data of wind farms,so the conclusions have practical significance and application value.The research of this paper has not only enriched the methods of wind speed forecasting,state evaluation for the output of wind power,but also accumulated some theoretical basis for safe and efficient use of large-scale wind power.