Evaluation of genetic algorithm alternatives for wind speed modelling using grey relational analysis

In this study, parameter estimates of the Weibull distribution, which is widely used for modelling wind speed distributions and is preferred in reliability analyses, are obtained by Genetic Algorithm (GA) optimisation based on the Maximum Likelihood method. A fitness function is proposed in the GA method to estimate the shape parameter of the Weibull distribution. Different GA parameter values such as population size, crossover rate and mutation rate were used to optimise the proposed fitness function. The proposed alternatives for the GA method are ranked by using the Grey Relational Analysis (GRA) and Entropy weighted GRA methods based on Kolmogorov-Smirnov (KS), R2, Root Mean Square Error (RMSE), Akaike Information Criterion, Bayesian Information Criterion and Power Density Error criteria over three different wind speed data.