Study On Short-term Wind Speed Prediction Of Wind Farm At Shangdu And Huade County Of Inner Mongolia

Wind power is unstable and uncontrollable. Large-scale development of wind power brings pressure and challenge to the power grid dispatching. Chinese wind farms are mainly located in the region of large capacity power grid construction is relatively weak. In the policy background of the rising proportion of wind power in the power grid, we must carry on the accurate forecast of short-term wind speed and power to ensure the safety and stable operation of power system, and the quality of electric energy, this is propitious to the development and application of wind energy business in China and the sustainable development of national economy.The paper is based on the typical terrain undulating hills of Inner Mongolia as the study area and using1953-2011a total of53years of conventional and non-conventional data to explore the characteristics and laws of large and mesoscale weather background and local complex boundary layer meteorological processes. It using time series statistical forecast method, numerical weather prediction of BJ-RUC and model forecast based on least squares support vector machine focuses on the applicability of Shangdu Huade short-term wind speed of wind farm prediction. Conclusions of the study are as follows:The predominant wind direction is west wind and northwest wind in Wind Area, and the maximum wind speed occurred in spring while the minimum wind speed occurred in summer. The daily variation regularity of surface wind speed is significant, wind speed during the day is stronger than wind speed during the night and the maximum wind speed occurred in the afternoon. The average wind velocity showed declining trend in53years. The wind velocity decreased by0.36m s-1/10a. There were four sudden changes during the period. The vertical distribution characteristics of ground layer wind filed showed that the diurnal variation of the wind speed curve phase characteristics changes with height is small. The daily amplitude has maximum in spring and has minimum in summer. The daily variation in10m height was larger than in70m height.Time series statistical forecast method in prediction of1-4hours super short term wind speed prediction error is smaller. But error increases quickly with the increase of forecast time. The results for forecast verification show that the method for forecasting aging for4hours super short term wind speed prediction can be applicable. Beijing rapid update cycle of assimilation forecast system (BJ-RUC) can forecast a longer time and forecast wind speed change trend well. Error is relatively stable within the prediction aging. The method is better suitable for short-term wind speed forecast comparing with time series statistical forecast method. The least squares support vector machine (LS-SVM) was applied to revise BJ-RUC numerical forecasting production. The error is decreased obviously after correction than the original numerical forecast products. The Correction method based on LS-SVM obtains the better forecast effect.