Research On Wind Speed Prediction Based On State Space Model

Wind power is one of the new energy which has gained more attention. But in the process of wind power generation and its interconnection, also faces many challenging problems. And Wind power prediction is one of the key issues. This paper, based on the measured data of wind farms, by means of statistical analysis to obtain the data characteristics of wind power, sets up wind speed and wind power prediction model, implements the short-term prediction of wind power.The wind data selection method based on the copula function and grey correlation degree. First, preprocessing the missing data of target wind farm. And then calculate the rank correlation coefficient of the target wind power and the other. On the basis of rank correlation coefficient selection, choose the strongest relation wind farm with the target one. Further, calculate the grey correlation degree of target wind farm and the selected. According to the grey correlation, draw the final input data. The data selected with this method not only use the spatial correlation, but also temporal correlation.Speed prediction model based on state space model. With time-series model and support vector machine model the linear state-space model and nonlinear state space model especially. During model the nonlinear state-space model, UKF is used, and the scale parameter of UKF is optimal. Stimulation shows that the prediction precision is improved with the UKF. This model is real-time, and use less data.