Short-term Wind Power Prediction Research Based On Extreme Learning Machine

The output of wind power is fluctuant and intermittent.The large-scale wind power accessing to power grid will bring adverse effects on the power quality of the power grid,which poses a serious challenge to the safe and stable operation of the power grid.The accurate prediction of wind power is one of the effective measures to solve the large-scale wind power accessing to power grid and increase the wind power proportion.At present,the domestic research on wind power prediction method is still in the theoretical research stage.The development and application of the actual maturity prediction system is relatively less and lack of practical experience.Therefore,it is of great theoretical and practical significance to s tudy the wind power prediction method.In this study,the short-term prediction method of wind farm power is selected.This study proposes a novel wind power prediction method based on empirical mode decomposition and extreme learning for wind power with complex nonlinear and nonstationary characteristics.This method firstly decomposes the original wind power time series by using empirical mode decomposition method,then establishes the appropriate extreme learning machine prediction model according to the characteristics of each component,and finally superimposes the prediction values of the components to obtain the final prediction value.The results show that EMD reduces the difficulty of modeling and forecasting,and improves the precision of wind power forecasting.Furthermore,extreme learning machine has more advantages than traditional BP neural network in learning speed and normalization performance.To improve the model instability caused by the random selection of hidden layer parameters in the extreme learning algorithm,this study presents a pre diction method based on empirical mode decomposition and kernel extreme learning machine.The simulation results show that the kernel extreme learning algorithm introduces the kernel function mapping instead of the hidden layer mapping of the extreme learn ing algorithm,and the prediction model has a great improvement in the stability and prediction accuracy.To further improve the prediction accuracy of the model,this study presents a method based on empirical mode decomposition and multi-core extreme learning machine power prediction combined with multi-core learning algorithm.The multicore function gathers the characteristics of many kinds of basic kernel functions,and can extract the feature information between data samples better,and has stronger learning ability.The simulation results show that the proposed method is effective in predicting wind power,and the model has better prediction accuracy and stronger generalization performance,which verifies the effectiveness of the proposed method in wind power prediction.