Research On Wind Power Forecasting Based On Improved Neural Network

The sustainable development of national economy requires the country to further deepen the reform of energy structure and expand the scale of renewable energy.As an important clean energy,wind energy has developed rapidly in recent years.However,the inherent intermittence,randomness and volatility of wind power bring challenges to the safe dispatching and stable operation of large-scale wind power grid.Wind power prediction technology can predict the output power of wind turbine in the future,provide effective basis for the control personnel to ensure the smooth operation of the grid,and help to reduce the cost of wind power generation,and create conditions for improving the scale of wind power grid connection.Therefore,the prediction technology of wind power output is the key to the safe dispatching and stable operation of the national grid.It is of great significance to improve the accuracy of wind power prediction.In order to improve the accuracy of wind power output prediction and the training efficiency of the existing neural network model,this paper first analyzes the main factors affecting the wind power output and the main characteristics of wind power generation,and then takes the historical data of a single wind turbine in Yunnan Province as an example to carry out simulation tests,and compares the accuracy of BP,ELM and RBF neural networks for wind power prediction Learning efficiency.Through genetic algorithm and principal component analysis,the traditional BP neural network wind power prediction model is optimized.The initial weight and threshold of BP neural network are improved by genetic algorithm,and then the data dimension is reduced by principal component analysis.On the premise of ensuring the accuracy,the machine learning training efficiency is improved compared with the traditional BP neural network.Finally,the simulation results show that the optimized BP neural network has better accuracy,model stability and training efficiency than the traditional BP neural network.