Short-term Prediction Of Wind Power Based On LSSVM

With the rapid development of society and economy,non-renewable energy,such as traditional fossil fuels,is becoming more and more scarce.Looking for clean energy has become an important means of sustainable development.However,the randomness and volatility of wind energy make the development and utilization of wind energy more difficult.Accurate wind power prediction can mitigate the adverse effects of wind power on the whole power grid.The least square support vector machine(LS-SVM)is widely used in wind power prediction,and the research method in this paper is based on this method.On the basis of this,the method is improved and applied to wind power prediction.In different stages of prediction,several algorithms,such as EEMDand mixed cuckoo algorithm,are used for combined prediction.The main contents of this paper are as follows:(1)The least square support vector machine(LSSVM)is used as the basic tool for wind power prediction to realize the short-term prediction of wind farm output power.(2)The parameter optimization problem of least square support vector machine(LSSVM)modeling is studied,and a hybrid cuckoo algorithm is proposed to improve the parameter optimization ability in SVM instead of the traditional method,which is verified by numerical optimization.Compared with the traditional mesh parameter optimization method,the proposed hybrid algorithm is more efficient and accurate,and can improve the accuracy and accuracy of short-term wind power prediction.(3)The output power of wind farm is affected by uncertain factors.Considering that the noise contained in the wind power data will have a certain influence on the inherent trend,the EEMD decomposition method is used in this paper to deal with the noise reduction.Then the improved least square support vector machine is used to predict the de-noised wind power sequence in real time.The final prediction results are obtained by phase space reconstruction.The established model is a combination prediction model based on EEMD and improved LSSVM.The model established in this paper is verified and compared with other prediction models.The feasibility and good of the combined model are obtained.