Prediction of Mean Winds at an Indian Coastal Station Using a Data–Driven Technique Applied on Weibull Parameters

Abstract

Statistical analysis of the wind speed has been carried out in the surface boundary layer at an Indian coastal station for the years 2006 and 2007 using a best fit two-parameter Weibull distribution. The seasonal variation of Weibull scale and shape parameters has been studied. Further, the predictability of the mean wind speed has been investigated using genetic algorithm in conjunction with Weibull parameters. The prediction of winds has been carried out for half an hour, one hour and one and a half hour in advance. The performance of the algorithm as reflected in the correlation coefficient and root mean square error has been found to be quite satisfactory.