Estimation of wind direction distribution with genetic algorithms

Directional and stream data are common in many research fields. Wind speed and direction are the most important variables for effective wind energy utilization. It is also well known, that wind significantly influences the current-carrying capacity of overhead power transmission lines. This shows the importance of knowing the annual wind direction distribution for specific locations, e.g. where wind farms or power transmission lines are situated. In this paper, a new method of wind direction distribution determination is presented. The statistical model is composed of a finite mixture of circular von Mises distributions. Parameters of the model are estimated using the heuristic search method of genetic algorithms. The quality of computed distribution is evaluated by Pearson´s chi-squared test. The entire proposed procedure is tested using a case study. The results show that the model composed of a finite mixture of von Mises distribution corresponds to the input data with high significance level.