Short-term Wind Speed Prediction Based on IVMD-PSO-LSTM Model
The original wind speed sequence is a kind of nonlinear and unsteady wind speed sequence,and the modeling and prediction of wind speed sequence directly are not very accurate.In this paper,a model based on IVMD(Improved Variational Mode Decomposition)-PSO-LSTM(Particle Swarm Optimization-Long Short-Term Memory)network is proposed to predict the short-term wind speed sequence.The IVMD algorithm can adaptively determine the number of decomposition layers,so that the original wind speed sequence can be transformed into several different frequency,stationary sub-sequences,and has good com-pleteness.In this paper,an appropriate decomposition layer is selected for VMD algorithm by calculating the fuzzy entropy value of each sub-sequence under different decomposition layers.Then,the original wind speed sequence is computed and decomposed by VMD algorithm to obtain a series of stationary sub-sequences.Then,the optimal parameters are found by PSO algorithm optimiza-tion of LSTM model.The optimized combination model is established to predict the subsequence,and the final prediction result is obtained by summation of the prediction results of the subsequence.The simulation results show that the proposed hybrid model of IVMD-PSO-LSTM is more accurate than the single model of BP,ARMA and LSTM,and meets the existing wind speed prediction standards.
国家科技期刊平台
NSTL
万方数据
