Compressed Air Energy Storage Configuration Based On Non-parametric Kernel Density Estimation

In recent years,wind power generation(WPG)is driven by natural resources and has been developed enormously with its clean and renewable characteristics.However,due to the high randomness and volatility of WPG,the high proportion of WPG connected to the grid brings severe challenges to the balance of power supply and demand,power system security and power quality.In order to deal with the volatility of WPG,the fluctuation of wind power generation can be smoothed by large-scale energy storage system.Compressed air energy storage(CAES)technology with its large capacity,low cost,high efficiency and other characteristics,becomes the best choice to smooth the fluctuation of WPG and improve the utilization rate of WPG.However,the capacity of the CAES will directly affect the effect of wind power stabilization and the economy of the whole system.Based on the aforementioned reasons,the compressor / expansion rated power and the capacity of the air storage carven of the CAES is studied in this thesis.In this thesis,the principle of wavelet transform and neural network are introduced,and the neural network structure is analyzed first.Secondly,considering the prediction error and energy deviation of WPG,the probability distribution of WPG prediction error is obtained by using nonparametric estimation method and empirical distribution model,then the scheduling period of energy storage system is obtained by the uncertainty of the probability distribution of WPG.On this basis,the nonparametric kernel density estimation method is used to model the prediction error interval of WPG.Finally,the thermodynamic model of the CAES is established.The CAES is regarded as an energy storage system to smooth the fluctuation of WPG.the probability density distribution function and the cumulative distribution function of wind power generation prediction error are established by empirical distribution model.On this basis,the nonparametric kernel density estimation method is used to establish the wind power prediction error.The CAES compression/expansion rated power and the capacity of the air storage carven are determined by the error analysis of the wind power generation prediction error.Finally,the feasibility and the effectiveness of the proposed method are verified by the simulation.