Establishment And Assessment Of A Hybrid Wind Speed Foresting Model

At present,wind power has become China's third largest power source after coal power and hydropower.As a new type of clean energy,wind power generation has the advantages of mature power generation technology and rapid cost reduction.Therefore,wind power has been continuously improving its position in the power industry.The proportion of power generation increases rapidly year by year with the expansion of installed capacity.According to relevant data of the world wind energy association,by the end of 2017,the total installed wind power capacity of the world has reached 539GW,of which 52.6GW were added in 2017.It can be seen that the wind power industry has been quite scale and continues to maintain the trend of rapid growth.Although China now has the world's largest installed wind power capacity,the problem is that wind power generation accounts for a relatively low proportion of total power generation.By contrast,in Europe,Denmark,Portugal and Spain.In the face of large-scale development of wind power industry and the need for wind power generation,high-precision wind power prediction is of great practical significance.Providing high accuracy wind energy forecasting allows us to improve the economic benefits of wind power,which reduces the generation costs and improves the security of the wind power system.In this paper,a novel hybrid forecasting model called E-SA-BP,which combines ensemble empirical mode decomposition(EEMD),a simulated annealing algorithm(SA)and a back-propagation neural network(BPNN),is developed to perform wind speed forecasting.As a widely used artificial intelligence prediction method,BP neural network has the advantages of reliable basis,rigorous derivation process,high precision and good universality.Therefore,BP neural network is selected as the basis of mixed prediction model in this paper.However,considering that BP neural network is prone to fall into local minima,the selection of network structure cannot be unified,and the determination of the number of nodes in hidden layers lacks theoretical basis,the following improvement methods are proposed in this paper:1.EEMD is used to process the original wind speed data and remove the noise interference in the original wind speed data.2.Select the weights and thresholds of the BP neural network optimized by SA to improve the network structure.Since the learning and memory of BP neural network are unstable,SA is introduced into the wind speed prediction model to select the weights and thresholds of the BP neural network to improve the prediction accuracy.SA has strong local search ability,can find the global optimal solution of combinatorial optimization problem,can avoid the search to fall into the local optimal solution,and the time consumed in the optimization is also more satisfactory.Therefore,in this paper,S A is selected as the weight and threshold of BP neural network search to generate the minimum prediction error,and the network structure of neural network is improved.The addition of SA also improves the accuracy of prediction to some extent.Therefore,based on BP neural network and combining EEMD and SA,a new hybrid prediction model called E-SA-BP is proposed in this paper.Last,the data of three wind speed observation sites in Jiaodong Peninsula of China are chosen to test the performance of the forecasting models.The results show an effective improvement in the forecasting accuracy of E-SA-BP when it is compared with the MA(1),ES(1),ES(2),ARM A,ARIMA,BP,SA-BP and E-BP models,and E-SA-BP model is suitable for wind speed forecasting.Although the hybrid prediction model proposed in this paper has achieved some breakthroughs in the accuracy of wind speed prediction and model stability,there are still some deficiencies in this study.1.The hybrid prediction model proposed in this paper focuses on the application of statistical modeling method and artificial intelligence method in wind speed prediction.Other wind speed prediction methods,such as spatial correlation method and physical simulation method,are not involved,and the selection of methods is limited.In the further research,multiple types of wind speed prediction methods can be considered comprehensively,and they can be combined and compared with each other.2.The hybrid wind speed prediction model proposed in this paper only makes single-step prediction,and whether the multi-step prediction will have a better prediction effect and can further improve the prediction accuracy,all of which need to be considered for further research...