Research On Hierarchical Coordinated Control Of Reactive Power And Voltage In Wind Farm Cluster Based On Predictive Reactive Power Margin

Wind energy has received more and more attention as a renewable energy source.A large amount of capital is invested in the wind power industry.Wind power technology has ushered in vigorous development.However,as the scale of wind power continues to expand,the problem of grid voltage fluctuations caused by the randomness and volatility of wind energy is becoming increasingly prominent.The technology of improving the quality of the grid-connected voltage while reducing the active power loss of the line has attracted much attention.It is also of great concern to give play to the ability of wind turbine reactive power regulation to improve the stability of wind turbine output voltage while reducing the risk of reactive power overrun.In view of the above problems,the wind farm cluster composed of doubly-fed wind turbines is taken as the research object.Ultra-short-term wind power prediction,reactive power margin prediction of doubly-fed wind turbines and reactive power hierarchical coordinated control strategy of wind farm cluster will be studied.This research has certain theoretical significance and application value.First,the reactive power characteristics of doubly-fed wind turbines are analyzed.The reactive limit relationship of doubly-fed wind turbines is derived.The reactive voltage characteristics of the wind farm cluster grid connection point are analyzed.The importance of reactive power compensation for maintaining the voltage stability of the cluster grid connection point is analyzed.In order to reduce the impact of wind power volatility,the causes of its fluctuations are analyzed.Finally,the role of wind power forecasting in reducing the impact of power fluctuations is clear.Secondly,in order to reduce the prediction error and accurately solve the reactive power margin prediction information of wind turbines,a combined prediction model based on empirical wavelet transform and least square support vector machine is proposed.During data preprocessing,empirical wavelet transform is used to decompose the original data sequence.The non-stationarity of the original data series is reduced.Before the prediction,historical data is used to train and test the least squares support vector machine learning algorithm.The generalization ability of the prediction model is improved.After data reconstruction,prediction error correction is used to further improve prediction accuracy.Through the simulation verification based on the measured wind farm data,the validity of the prediction model is determined.Finally,a multi-level joint optimization wind farm cluster control strategy is proposed.By decomposing optimization goals layer by layer,the operability of the strategy is improved.Among them,at the cluster level,improving the safety and economy of the reactive power regulation process is taken as the optimization goal.At the wind farm level,the minimum fluctuation of the outlet voltage of the wind turbine and the maximum predicted reactive power margin are taken as the target of reactive power distribution.At the wind turbines level,it is a prerequisite to give priority to the reactive power regulation capability of wind turbines.Based on the above method,the reactive power tasks of all layers are coordinated and managed to realize the closed-loop control of the reactive voltage of the wind farm cluster.In the modeling and simulation of Matlab/Simulink,the effectiveness and feasibility of the proposed control strategy are verified.