Short-term Wind Speed Prediction Based On Hybrid Optimization Model

Wind energy is a promising renewable energy driven by climate.It helps to overcome global warming and environmental pollution caused by fossil fuel combustion.It has very high social and economic benefits.The use of renewable energy is an inevitable choice for the sustainable development of human society.Wind energy was introduced as alternative energy in the 1970 s.Since the 1990 s,the growth rate has been faster than any other energy.However,with the rapid development of wind power generation,many problems have arisen.One of the main problems related to wind power generation is that the instability and persistent fluctuation of wind speed greatly affect the regulation of power system and the dynamic control of wind turbine.Therefore,accurate wind speed prediction is very important for the stable operation of wind energy conversion system.This paper presents a short-term wind speed prediction model based on data preprocessing technology and neural network technology,namely WD-VMD-MEAElman hybrid model.In this paper,we study how to improve the accuracy of wind speed prediction by data preprocessing technology.We use variational mode decomposition(VMD)technology which is better than empirical mode decomposition(EEMD)to decompose the wind speed data set.To solve the uncertainty of the optimal mode number K and the optimal penalty parameter alpha required for VMD decomposition,we introduce cuckoo search algorithm(CS)to determine K and alpha adaptively,and apply one.A new algorithm called thought evolutionary algorithm(MEA)is proposed to optimize Elman neural network.The proposed model firstly uses wavelet decomposition to denoise the original wind speed data,then uses cuckoo search algorithm to self-adaptively determine the optimal parameters required for variational mode decomposition(VMD),and uses the optimal parameters to decompose the denoised data into several subsequences of intrinsic mode(IMF).Finally,Elman neural network optimized by thought evolutionary algorithm is used to decompose the denoised data into several subsequences of intrinsic mode(IMF).Subsequence prediction.The model combines the advantages of wavelet denoising,VMD decomposition,optimization algorithm and neural network,and effectively improves the accuracy of wind speed prediction.The model is validated by real wind speed data of Jiuquan and Zhangye in Hexi Corridor and compared with WDVMD-GA-Elman,WD-VMD-PSO-Elman and WD-EEMD-MEA-Elman models.The results show that the hybrid model can achieve more accurate prediction than other comparative models.