Short-Term Wind Power Prediction Based On FFA-VSELMAN

With the continuous consumption of fossil fuels such as coal,oil and natural gas,renewable energy issues have become the focus of world attention.Wind energy is a renewable and clean energy source with strong randomness and intermittentness.As the fluctuation of wind power output will reduce the reliability of the power system,it is important to establish a high-precision wind power forecast model for wind farms.This paper focuses on the following aspects of wind power forecasting:(1)Aiming at the data preprocessing problem,that is,how to select the meteorological features with strong correlation with wind power as the input of the prediction model,the Copula theory is introduced for correlation analysis.First,the overall analysis is performed on the correlation between wind power and local meteorological characteristics,and the optimal Copula function is selected using the rank correlation coefficient and the Euclidean distance.Second,the BEMD decomposition algorithm is used to decompose the wind power and meteorological characteristics at a microscopic angle and further analyze the two.Correlation between the two,and finally select the meteorological features with strong correlation as the input of the prediction model.(2)Considering the training problem of Elman neural network,a variable learning rate Elman neural network(VSELMAN)is proposed.This method allows the network to always be trained at the maximum rate.(3)First consider the parameter optimization problem of VSELMAN neural network,and use the firefly algorithm(FA)to optimize;then for the problem of fixed step size of the firefly algorithm,introduce fuzzy logic to construct a fuzzy firefly algorithm(FFA),and compare the firefly with experiments The algorithm and fuzzy firefly algorithm show that the calculation speed of fuzzy firefly algorithm is higher than that of firefly algorithm.Finally,a FFA-VSLEMAN wind power prediction model is established.(4)Based on the measured data of a wind farm in the north(15 2MW wind turbines),the established FFA-VSELMAN neural network model is used for wind power prediction,and the prediction effect of the model is evaluated by averaging absolute percentage error and root mean square error.The experimental results show that FFA-VSELMAN neural network has better convergence speed and model prediction accuracy than FA-VSELMAN neural network,VSELMAN neural network,ELMAN neural network and BP neural network.The model has practical feasibility,and the proposed prediction method is reasonable and effective.