Short-term Wind Speed Prediction Based On IVMD-PSO-LSTM Model

As one of the most promising renewable energy sources,wind energy has received wide attention from many countries.However,due to the large volatility and uncertainty of natural wind,wind power generation can also fluctuate,which seriously affects the reliability of wind power systems and brings challenges to large-scale grid connection of wind power.Wind speed prediction is of great significance to ensure the safe and stable operation of wind power generation systems.In this paper,a combined short-term wind speed prediction model based on FE(Fuzzy Entropy),VMD(Variational Modal Decomposition),ARMA(Autoregressive Moving Average Model),PSO(Particle Swarm Optimization algorithm)and LSTM(Long-short Time Memory network)is proposed.The details are as follows.First,the problem that the number of decomposition layers in VMD is difficult to determine is addressed.The FE algorithm is used to calculate the entropy change of the residual sequence.When the information in the original signal is fully extracted,if the number of decomposition layers continues to increase,then the FE value of the residual component will not continue to decrease,and the optimal number of decomposition layers of the VMD algorithm can be judged according to this criterion.Then,the optimized VMD algorithm is used to decompose the original wind speed data to extract different feature information in the sequence and reduce the non-smoothness of the sequence.Then,the parameter optimization problem in the LSTM network is addressed.The input dimension of the neural network is determined adaptively using an ARMA model to improve the prediction performance of the neural network;according to the principle of minimizing the mean square error,a PSO algorithm is used to optimize the weights and thresholds in the LSTM network to avoid the problem of falling into local optimum and improve the stability and robustness of the network.Finally,the wind speed dataset of Quebec wind farm in Canada is modeled and predicted by analyzing two different time periods,and MAE,MSE,RMASE,MAPE and R are used as prediction model error evaluation indexes.Compared with the single BP model,ELM model and LSTM model,the prediction performance of the combined model proposed in this paper is greatly increased;and the comparison with the combined EMD-PSO-LSTM model proves the effectiveness of the optimization algorithm used in this paper,and further verifies the higher prediction accuracy of the model proposed in this paper,which is feasible and superior in short-term wind speed prediction.