A new stochastic approach to weather condition for wind energy applications

摘要

Weather conditions are related to wind turbine power output and maintenance therefore, it now becomes interesting to use stochastic approaches to study regional weather conditions. In this paper, this approach was taken, based on real sampled data for a statistically representative time period. Results showed an adequate curve fit of the different probability density functions like Exponential, Gaussian, Log-Normal, GaussCum, Gamma, WeibullCum and Weibull distributions. Further, a common probability density function for all these weather variables was understood as a Weibull peak function with ten minutes of real sampled data, over a ten-year period. Once the Weibull model for each year and variable was defined, the relationship was analysed using the Pearson's correlation factor that was employed as a validation tool of weather forecast. Results showed that the partial vapour pressure of moist air resulted in a better approach with GSI than moist air dry bulb temperature, and particularly thermal comfort indices, specifically humidex, showed a good approach. Aiming to define a common model for all the weather variables, the determination factor between all the weather model constants was calculated for each year showing a relation between pressure and temperature and wind velocity. Finally, a practical case study of stochastic approach of wind power output was developed and the probability density functions of the main control parameters of the wind power station were defined.