Wind power calculation based on sums of diffusion method and a novel power curve approximation model

Wind energy estimations using wind speed models and power curve approximations are mandatory for the optimal regulation of wind electricity generation. A weak stationary process can be probabilistically modeled as a sum of several diffusion processes generated by a class of stochastic differential equations. In this study, a residual dataset obtained by eliminating the non-stationary component of a wind speed dataset recorded at Kokkilai in Northeastern Sri Lanka, has been considered as a weak stationary process. The correlogram of residual dataset is bounded by an exponentially decreasing theoretical autocorrelation function that includes several time scales. Hence the residual process is probabilistically approximated by three sums of diffusion models based on the infinite divisibility and the ρ-mixing property of the underlying distribution. By combining each approximated residual process and the non-stationary component of the dataset, three wind speed distributions were constructed. Then each distribution model was incorporated to calculate the average energy generated using the power curve data of the recommended wind turbine for the site using a novel power curve approximation model. Consequently, a comparative study was performed on energy calculation based on each model and the most appropriate model was selected and the feasibility of the recommended wind turbine was discussed.