The Spatial Correlation and Lead Time for Wind Speed Prediction in East and Southeast Coastal Region of China

One of the fundamental tasks for security and economical operation of smart grid is to accommodate every kinds of renewable green energy. The future wind speed prediction by spatial correlation is a new approach to improve the reliability and accuracy of wind power forecasting. For the future local wind speed prediction, it is theoretically deduced by the causes of wind in dynamic meteorology and characteristics of binary linear regression in mathematical statistics: although the pressure gradient force is the primary force of wind, the existence of measurement errors makes that the upstream wind speed is the primary influencing factor where the spatial distance is short; both the pressure gradient force and the upstream wind speed are used where the spatial distance becomes farther. As a case study, to improve the reliability and accuracy of wind power forecasting during monsoons in east and southeast coastal region of China, the correlation numerical analyses and regression predictions show that about 600 km is the distance to change influencing factors. The partial least squares regression, support vector regression, long short-term memory, and deep belief network are employed to the regression predictions. A regional spatial correlation prediction network is suggested to be establish in this region. And for the wind speed predictions of monsoons within lead time10 hours, one hour lead time increasing makes about 0.15 m/s increasing of the root mean square error.