Uncertainty Analysis Of Wind Speed Based On Mixture Distribution

When designing and constructing wind farms,local wind resources need to be evaluated.Uncertainty analysis is an important and indispensable task in the assessment of wind resources.There are many factors that affect the uncertainty of wind speed,including measurement errors,errors in the estimation of wind speed probability distribution,and changes in wind resources.Among them,the accurate estimation of the probability distribution of wind speed can help to more accurately assess the local wind energy potential level,contribute to the construction and planning of wind farms,and at the same time contribute to the operation efficiency of wind farms and reduce costs.This study analyzes the uncertainty of wind speed from two aspects of probability distribution of wind speed.First,the heavy-tail characteristics of wind speed data are discussed,and a method for judging the heavy-tail data is given.The generalized extreme value distribution(GEV)and the generalized Pareto distribution(GPD)commonly used in engineering are used as components of the mixed distribution to fit the wind speed data.The Weibull distribution and eleven mixed distributions are used to fit the wind speed data in the mountainous area of Yunnan,China.The maximum likelihood method is used to estimate the parameters of the selected distribution,and the optimal wind speed probability distribution is selected using the comprehensive evaluation criteria consisting of 5 test statistics,the distribution density curve of the histogram and the QQ chart.According to the results of the comprehensive evaluation criteria,the normal-normal-generalized extreme value distribution(NNGEV)has the best effect in fitting wind speed data in the mountainous areas of Yunnan,China.Secondly,the characteristics of the parameters in the wind speed probability distribution with time are studied,and the time-varying parameter model is used to describe the low-order autocorrelation of the time-varying parameter sequence.Mainly consider the situation that the weight parameter in the normal mixed distribution changes with time.The Gibbs sampling,MH algorithm and Kalman filter are used to estimate the parameters of the time-varying parameter model.The method of data simulation is used to analyze the time-varying parameter model.This model is effective and feasible.This study believes that the accurate distribution of the wind speed data with probability distribution is very important for the analysis of the wind energy potential of wind farms.The research from the aspects of heavy tail characteristics of wind speed data and time variability of wind speed probability distribution parameters has obtained good results.Results of the analysis to analyze the uncertainty of wind speed.