Wind Power Prediction Based On Empirical Mode Decomposition And Neural Network

Wind energy is not only a renewable energy source,but also has the advantages of low cost,convenient development and utilization,and green pollution-free.As society's demand for energy grows and people's awareness of environmental protection grows,wind energy begins to shift from complementary energy to major energy.With the development of China's science and technology level,China's wind farms have made great progress and progress both in terms of scale of development and technology.Wind power has great potential for development in China.However,due to the instability and randomness of wind energy,the accuracy of the output power of the wind farm is difficult to guarantee,which has an impact on the operational safety and stability of the power system.Improving the prediction accuracy of wind power has important practical significance for improving the operational stability of wind farms and ensuring the safety of power systems and the quality of electrical energy.The wind power time series is susceptible to meteorological factors such as wind speed and wind direction,and has high instability and strong nonlinearity.It is difficult to use linear prediction method to ensure its prediction accuracy.In view of the above problems,this thesis studies the short-term wind power prediction method for a wind farm in Liaoning.The research contents are as follows:(1)The short-term wind power of the wind farm is predicted by a back propagation(BP)neural network.The BP neural network model is established by the actual running data of the wind farm,and the prediction model is simulated.The research results show that the BP neural network model can describe the nonlinear characteristics of the wind farm power prediction model,but the accuracy needs to be further improved.(2)In order to further improve the accuracy of the model,the signal decomposition method is used to preprocess the data to obtain a relatively stable subsequence to reduce the influence of the randomness of wind power on the system.In this thesis,the empirical modal decomposition(EEMD)is used to decompose the wind power,wind speed and sine value cosine sequence of wind angle to obtain a series of periodic,regular and relatively stable eigenmode(IMF)components.Remaining component.(3)The initial value of the weight and threshold of BP neural network is optimized and improved.The BP neural network is optimized by genetic algorithm,and the optimized weight and threshold initial value are optimized by genetic algorithm.(4)Combining empirical modal decomposition with genetic algorithm optimization neural network(EEMD-GA-BPNN)method to establish a short-term wind power prediction model,using multiple neural networks to predict each IMF component and residual component.The simulation experiments show that compared with the simple BP neural network model,the average absolute error,average relative error and mean square error are reduced,which proves the effectiveness and feasibility of the proposed method.