Abstract
Knowledge of wind speed variation is important for response of small wind turbines. In order to investigate the characteristics of speed variation of natural wind in the time domain of several seconds, a method of detecting specific wind speed patterns by using both correlation coefficient and root mean square (RMS) has been proposed. Threshold of correlation coefficient was determined as 0.98 and that of RMS as 0.35. Wind speed data in the year 2007 was analyzed using the method; and parts that resemble up-ramp patterns or down-ramp patterns were detected. The results showed that natural wind tends to increase in relatively short time but decrease in longer time and that probability of up-ramp patterns with short rise time (about 1 s) becomes higher when wind speed level is high.
