WRF Model-based Short-term Wind Speed Forecasting And Correction

Wind energy which is a clean, renewable and pollution-free new energy, has not only received wide attention of the countries all over the world but also developed rapidly. The growth of wind power capacity has increased rapidly in recent years, especially in our country, which has now become the country with the largest installed capacity of wind power in the world. However, wind power has the characteristics of volatility, intermittent and randomness. There is a severe challenge to the security of the power system stability and the efficient operation of social economy when large scale wind power is combined to the grid. In order to improve the utilization of wind power as well as reduce the negative influence of wind power generation on power grid operation, it is necessary to predict wind power accurately in advance, and the wind speed is the most important input factor of wind power prediction. Therefore, accurate wind speed forecasting is the premise and foundation of wind power prediction.In order to predict the short-term wind speed of wind farm effectively, this paper makes some in-depth researches on short-term wind speed forecasting and its correction methods. The main works are as follows:Firstly, the WRF model was used to predict the short-term wind speed of wind farm in this paper, the data of model is from NCEP/GFS global weather forecast data, and the prediction time is 0-72 hours. The WRF model was used to forecast wind speed, wind direction and other meteorological elements of wind farm area for a whole year, and the forecasting results were interpolated to the wind tower hub height (70m) by bilinear interpolation method. Then the WRF model forecast data are compared with the actual data from wind speed and direction of wind tower. It shows that, WRF model forecast wind speed is greater than the actual wind speed of wind towers, but the trend of forecast wind speed is more consistent with the actual wind speed; the forecast and the actual wind direction also has a good consistency. So, the wind speed output by WRF model used in short-term wind power prediction is effective and feasible, but the accuracy must be further improved.Secondly, in order to improve the accuracy of WRF model forecast results, the Extreme Learning Machine method (ELM) was introduced to make the correction for forecast wind speed. Due to the fitting performance of traditional ELM algorithm might be affected by network structure in a certain extent, the Differential Evolution algorithm (DE) was used to optimize the ELM weight of network input layer and the bias of hidden layer, and the Conjugate Gradient algorithm (CG) was used to solve the output weight value, which made the ELM algorithm optimized further. Based on the improvements above the problem of training large sample of data was solved and the generalization ability of the model was also improved. The experimental results showed that the correction errors of the optimized ELM algorithm had significantly reduced compared with the traditional ELM algorithm, and the effectiveness of the optimized ELM algorithm got verified. The error of revised wind speed is smaller than the error of WRF model forecast wind speed, and the revised wind speed is closer to the actual wind speed.Finally, owing to the wind direction, temperature, pressure and other meteorological elements also have significant influence on the accuracy of wind speed forecasting, a second short-term wind speed correction model was established based on Principal Component Analysis (PCA) and Radial Basis Function (RBF) neural network. PCA method was used to process the revised wind speed of optimized ELM algorithm and other meteorological elements data, eliminating the correlation between the data, and then the data which processed by PCA was input into RBF network. The experimental results showed that the second correction model based on PCA and RBF neural network could further reduce the error of forecast wind speed, ameliorated the trend differences of forecast wind speed and improved the accuracy of short-term wind speed forecasting.