基于季风特性的风电空间相关性预测的基础研究-控制科学与工程专业论文.docxVIP

ABSTRACT In recent years, wind power industry develops quickly in the world. But wind power generation mainly depends on the wind speed, so the wind speed prediction accuracy contributes to the stable operation of power system, the best efficiency of wind energy, which has very strong social value and economic value. The wind speed data obtained from wind field has a certain noise, its existence affects wind speed upper limit accurately. How to separate the noise from wind speed data accurately, and study the correlation between wind speeds and how to forecast wind speed reasonably, are the focus of this paper. The contents of this paper are as follows: A low order nonlinear transform—Square Root Transform (SRT) is studied to the effect of wavelet threshold denoising. In this experiment, the standard signal plus outliers and jumping point is tested, it proves when outlier value L ??[-0.7340, 0.4314], after SRT, becomes L??[-0.7376, 0.48035], the interval length of L increases 4.509%, the stability of wavelet threshold denoising effect is enhanced, and give reasonable theoretical interpretation. Based on Statistical Principle - Difference together with the wavelet denoising theory, we respectively separate white noise from wind speed in the monsoon regions of China during four periods, making a lot of experiments and analysis, The research belongs to prediction accuracy upper limit and forecast research of mechanism + identification prediction strategy. Taking advantage of Pearson Product-Moment Correlation Coefficient （PMCC），we judge the degree of wind speed spatial correlation in monsoon region, and the Fisher z transformation is used to estimate the confidence interval of PPMCC, providing the precondition for the partial least squares to predict wind speed. We provide a new theory—Wind Power Automatic Prediction System based on the Internet of Things, improving the accuracy of wind power forecasting further. KEY WORDS：Wind Speed Forecasting, White Nois