Research On Wind Power Forecasting And Dispatching Based On Time Frequency Analysis

In today’s background of carbon neutrality,the use of renewable energy such as wind and solar energy to replace traditional fossil energy for electricity generation will help achieve the carbon neutrality target earlier.Among them,wind energy has been widely concerned and applied because of its advantages such as wide distribution,rich accumulation and less affected by bad weather.However,wind energy has the inherent characteristics of high volatility,intermittency and randomness,also leading to the fluctuation,intermittency and randomness of wind power generation.When large-scale wind power is connected to the grid,it will bring great challenges to the operation reliability and dispatching economy of the power system.Therefore,it is of great significance to study wind power forecasting,wind power risk assessment and wind power dispatching.This thesis studies wind power forecasting,wind power risk assessment and wind power dispatching from the perspective of time-frequency analysis.Firstly,a short-term interval forecasting method of wind power based on timefrequency analysis is proposed.The noise component and the trend component of wind power are obtained by wavelet transform and singular spectrum analysis.The probability distribution fitting of the noise component is performed to get the confidence interval of the noise,and the trend component is forecasted deterministically by using the short and long memory network.Then,interval forecasting results of wind power are obtained by superposition of the above two.This method improves the forecasting accuracy of the trend component and also takes into account the uncertainty of noise.Secondly,a risk assessment method of wind power based on time-frequency analysis is proposed.The S-transform of wind power is carried out to extract timefrequency domain characteristic indexes which reflect the volatility,intermittency and randomness of wind power.The characteristic indexes of wind power are evaluated comprehensively,and the evaluation value reflecting the risk of wind power is obtained.This method can comprehensively evaluate the characteristics of wind power in the time-frequency domain,and provide a reference for risk level assessment of wind farms and wind power grid-connected dispatching.Finally,a day-ahead dispatching method of wind power based on sequential wind abandonment penalty is proposed.The risk level of each wind farm is obtained by the characteristic evaluation method of wind power based on S transform,and it is transformed into the abandonment order constraint of wind farms.Combining the abandonment order constraint of wind farms and the segmented wind abandonment penalty cost coefficient,a wind abandonment penalty model is established.On this basis,a day-ahead dispatching model of wind power,considering the operating costs of thermal power units and wind abandonment penalty costs,is constructed.This method can ensure the stable operation of the power system and reasonably arrange the abandoned wind power of different wind farms at the same time,so providing a reference for the orderly connection of wind farms to the power grid.