The Fusion Model For Wind Power Prediction Based On Model Optimization

Fossil fuels as non-renewable energy are the primary energy consumed by the human, and the burning of fossil fuels will bring the environmental pollution. Wind energy as a kind of clean, renewable and infinite energy has become main energy source for changing the structure of energy consumption and reducing the environmental pollution. Generate electricity is the main form of use of wind energy. As the wind energy is intermittent, randomness and instability, so the wind power is also intermittent and instability. Wind power connected to the grid in large-scale will affect the power grid scheduling, the power quality and reliability of the power system operation seriously. Wind power prediction can adjust the schedule of the grid, improve security and stability of power grid operation and provide the foundation for the efficient use of wind power.There are several kinds of single forecasting models. But relatively large errors may appear, because the single forecasting models utilize the data information incompletely, which reduces the prediction accuracy. To further improve the prediction accuracy, the fusion model for wind power prediction is established in the thesis supported by National Natural Science Foundation of China (51277127).The main research content in this thesis is summarized as follows: (1) The background and significance of the wind power prediction are elaborated. The wind power short-time forecasting models are summarized. And insight into the wind power generation technology.(2) The fusion modeling method chooses some classical predictive models firstly to establish the single model database, including time series model, regression analysis prediction model(RAPM), back propagation neural network(BP), Elman neural network, radial basis function neural network(RBF), generalized regression neural network(GRNN) and support vector machine (SVM).(3) On the basis of the single models, the combination model is established based on arithmetical average method, the reciprocal variance method, the simple weighted average method, the binomial coefficient method and the entropy evaluation method. The simulation results show that the prediction accuracy of the combination model is better than that of every single model.(4) Not all of the single forecasting models can effectively improve the prediction accuracy of fusion model, which is found in the simulation, so the single forecasting models should be optimizated. The fusion modeling method initially selects the higher precision models with grey correlation analysis, and then judges the redundancy of the higher precision models with information matrix of forecasting errors, eliminates the redundant models to simplify the fusion model establishment. The simulation results show that the fusion model simplifies the computation of the model and saves the prediction time. The prediction accuracy of the fusion model is slightly better than that of corresponding combination model. And the fusion model solves the problem that the Shapley value calculates the negative degrees of fusion. The fusion model provides the foundation for the efficient use of wind power.