Copula Function To Explore The Relationship Of Wind Speed In Inner Mongolia Autonomous Region

Wind energy,as a renewable energy,has been widely developed and used all over the world,effectively reducing the serious environmental pollution caused by global industries.Wind speed is an important factor affecting wind energy.It is particularly important to discuss the application of wind energy and the selection of wind farm locations for wind speed and wind speed changes in neighboring areas.When discussing the relationship of wind speeds in neighboring areas,it is generally assumed that the wind speeds at different locations are independent or ignored the correlation intentionally or unintentionally in most cases.In fact,wind speeds in neighboring areas are interdependent because they are affected by similar geography,weather,climate,terrain,and other factors,which makes it difficult to describe the wind speed relationship between the two places.Fortunately,Sklar proposed the Copula function to model and measure the multivariate dependent structure of the marginal distribution.In this paper,the Copula function is used to study the wind speed and forecast wind speed in the neighbouring areas of Inner Mongolia(Hohhot,Baotou,Jining),and to provide parameters for the selection of wind farm locations and wind energy applications.This paper first describes the definition and properties of wind speed distribution,parameter estimation method,Copula function,Copula regression function and quantile curve,P(X > Y)Calculation method etc.Secondly,the 2013-2017 daily average wind speed data of three weather stations in Hohhot,Baotou and Jining were analyzed and the Copula model was established two by two.The Copula function and the P(X > Y)theory were combined from five years(2013-2017)Compare the wind speed with the angles of the four seasons.Choose three types of two-parameter distributions: Weibull distribution,Log-normal distribution and Gamma distribution to describe the Copula marginal distribution,and use maximum likelihood estimation and moment estimation to solve the parameters.Root mean square error(RMSE)and KS values to find the best fitting distribution.The results show that whether it is 2013-2017 overall data or seasonal data,the optimal marginal distribution of Copula function is Lognormal distribution.Establish Copula function,Select the Archimedes Copula function,use maximum likelihood estimation and Kendall’s τ method to estimate the Copula function parameters,and find the optimal Copula function by comparing the root mean square error(RMSE)and KS value.Next,the Copula function is combined with the P(X > Y)theory to compare the wind speed in Hohhot,Baotou and Jining,and the results are compared with the theoretical distribution(assuming(X,Y)is independent)and The empirical distribution of P(X > Y)value comparison,the final conclusion shows that compared with the other two cases,Copula theory can better fit the relationship between the two places,and whether it is2013-2017 overall Both data and seasonal data are in Hohhot with the highest wind speed,followed by Baotou and Jining.Finally,the Copula regression function and the Copula quantile curve are used to make point predictions and interval predictions for the monthly average wind speeds in Hohhot,Baotou and Jining in 2018.The monthly average wind speed data from 1981-2017 in Hohhot,Baotou and Jining are used as basic samples,2018 The annual average data is used as a prediction sample,and the monthly average data of 1981-2017 is used to establish the neighboring month Copula regression function.The function and the Copula quantile are used to make point predictions and interval predictions for the 2018 monthly averages of the three places.The results show that after removing the data from Hohhot in January 2018 and Baotou in July,the remaining monthly average real values are all within the prediction interval,which shows that the Copula prediction method used in this paper is effective and feasible.In addition,for point prediction In general,the prediction effect of Jining point is better than that of Hohhot and Baotou;for interval prediction,the prediction effect of Baotou is better than that of Hohhot and Jining.