Study On Forecasting The Wind Speed And Wind Power Based On The Measured Data Of A Wind Farm

Wind power is one of the renewable energy power generation technologies which are growing fastest and the most mature. In recent years, China's installed capacity of wind power increases rapidly, and its proportion in power grid continues to increase. Wind power accessed to the power grid has a greater impact on the power grid, leading to peaking power surge problems. In order to reduce this negative impact, forecasting the wind speed and wind power will help power grid scheduling department know the wind power to be injected to the power grid, to facilitate a reasonable grid scheduling and ensure good quality of supply.This paper is based on the measured data of Tao Nan wind farm in western Jilin Province, after the discussion of daily variation characteristics of wind speed and wind power, fluctuations characteristics and the relationship between them, wind speed characteristics and distribution of hourly wind speed, based on historical data method, the short term prediction model of wind speed and wind power is researched.Firstly, ARMA time series technique is researched in detail. Through the model identification, parameter estimation and model testing, ARMA (6,1) model is proposed, which is applicable to the wind farm. ARMA model is feasible to forecast the short-term wind speed, and prediction accuracy ahead of multi-step doesn't decline much than ahead of one-step, which indicates that dynamic prediction can improve the prediction step within a certain range of allowable error.Then the chaotic time series theory is introduced into the wind speed prediction. Using C-C method to calculate the parameter of reconstructed phase space; Time series'Lyapunov index greater than zero is the standard of measuring the original system has a chaotic nature, Lyapunov indexes of wind speed, wind power calculated with a small amount of data are both greater than zero, which prove they are of chaotic characteristics. On this basis, wind speed is forecasted by using a first-order weighted local prediction method, relative to the ARMA method, the error indicators of chaotic time series prediction results improve significantly and prediction results are better.Finally the chaotic time series prediction method is used to directly forecast wind power, and it is compared with the power curve transformation method. It is found that the effect of direct prediction is better, and the error can be reduced two percentage points. In the case of meteorological and other factors unknown, using historical data simply, it is proposed that chaotic time series method is applied directly to forecast the wind power in a wind farm.